Don't look now, it's in the e-mail!!

Thanks for responding to my unabashed plea to get free advise and
tie up your time while I enjoy reruns of Green Acres and play gin
rummy on my PC!

I'm sending my brain child as a separate package, but first must
make clear:

     1. this is a work in progress!!
     2. some non-ascii letters got substituted with pirate signs
or completely left out, like my nice BETA symbol.
     3. there seem to be a few extra lines thrown in courtesy of
the e-mail postmaster or something.  Take them out and send them
back to said master.
     4. I do not have data for 1983 (the only year I'm currently
missing between 1970 and 1986).  Thus, some Tables have * where
they should have values.  Also, some ** mean the value is missing
for the same reason.
     5. Chapter Six refers to a diagram which can't be sent via
ASCII.  The text might make things clear enough.  Besides, by
Chapter Six--if anybody even gets there!--anything will look
right!!
     
     Although I bill this as a dissertation on public policy
using chaos theory AND evolutionary models, you will see that I
deliberately blur the distinction and eventually use one half to
inform the other half.  Damn cleaver of me, if I must say so!

     Stan
                  Chaos Theory and Evolutionary Models
                 for Economic Development Public Policy



                               Chapter One
                                    
                            Research Proposal
                                    
                                    
     CHAPTER ABSTRACT: Moving toward the goal of enhancing
     research on subnational policy-adjustment behavior,
     this chapter introduces three models drawn from
     complexity and chaos theory--deterministic,
     equilibrium, and dissipative.  The different
     assumptions of each model are discussed with reference
     to past research on states' public policy adjustment. 
     The goal is to introduce the possibility of using
     orientations drawn from complexity and chaos theory to
     further clarify and explain subnational policy-
     adjustment behavior in economic development.
          The overall target of the present inquiry aims at
     showing that a different analytical approach to public
     policy analyses can produce results which have eluded
     past research.  Specifically, within the overall
     framework of chaos theory and complexity, two
     approaches are relevant: (1) assessments from physics
     of log-linear "noise" and (2) models from biology
     concerning evolution within a system.  These two
     orientations generate models which can be applied to
     the record of policy adjustment by states.  By testing
     the fit of each model to subnational behavior, results
     can help highlight the dynamics which influence policy-
     adjustment rates in states.  


FAILURE OF ANALYSES

     No clear explanation exists concerning the adoption, use,
and spread of state-level public policies (see Chapter Two). 
This shortfall is not due to inattention in the professional
literature.  In fact, several aspects of state policy-adjustment
behavior have received considerable attention and generated a
lively debate.  However, more inquiry has produced less agreement
and more inquiry.  After over thirty years of serious attempts to
explain possible relations between state-level characteristics
and the public policy record, a clear explanatory model remains
elusive.
     So the question remains: What factors explain state policy-
adjustment rates?  One familiar means of answering this question
has been to relate state-specific socio-economic and/or political
characteristics to policy preferences.  This approach has a
logical appeal to it.  Differences between states, for instance
in apportionment or gubernatorial authority, would suggest
different policy goals and histories.  Nevertheless, statistical
analyses have failed to verify hypotheses drawn on such thought
experiments.  Noting this failure, what else remains?
     To answer this question, a new approach is needed to study
subnational behavior.  This approach starts with Jantsch (1980)
who describes three systems--deterministic, equilibrium, and
dissipative.  A deterministic system is based on a Newtonian
vision of cause-effect as found mostly in physics.  Systems
exhibiting deterministic characteristics are predictable, linear,
and controllable.  Small stimuli cause small outcomes, and large
stimuli, large outcomes.  Events are ahistoric; past experiences
do not change the outcome of stimuli.  
     Equilibrium systems, in contrast, recognize that
perturbations can move the system away from stability at least
temporarily.  However, equilibrium systems have coping mechanisms
to restore stability after a shock.  Also, in an equilibrium
system, "primary emphasis is placed on the relations among
separate components of the systems under analysis.  In this way,
a system is defined as a set of elements which interact with each
other and the environment" (Leifer, 1989; see Von Bertalanffy,
1975). 
     Last, dissipative systems are most dissimilar to
deterministic systems while equilibrium systems are intermediate.
Dissipative systems are recognized in complexity and chaos theory
as important examples of systems involving complicated yet
deterministic interaction between agents.  Dissipative systems
are open to environmental influences and undergo real change and
restructuring based on inherent stabilites.  Unlike an
equilibrium system, when a dissipative system is perturbed,
changes within the system will create a new equilibrium which is
different from previous points in time.  (Equilibrium systems, by
contrast, will show momentary disequilibrium before settling back
into its previous equilibrium.)  Toffler (1984), writing in the
introduction to Prigogine's and Stengers' ORDER OUT OF CHAOS,
offers the following overview of how a dissipative system on the
edge of chaos undergoes change.  He writes:

     In Prigoginian terms, all systems contain subsystems,
     which are continually "fluctuating".  At times, a
     single fluctuation or a combination of them may become
     so powerful, as a result of positive feedback, that it
     shatters the preexisting organization.  At this
     revolutionary moment--the authors call it a "singular
     moment" or a "bifurcation point"--it is inherently
     impossible to determine in advance which direction
     change will take: whether the system will disintegrate
     into "chaos" or leap to a new, more differentiated,
     higher level of "order" or organization, which they
     call a "dissipative structure". ....
          One of the key controversies surrounding this
     concept has to do with Prigogine's insistence that
     order and organization can actually arise
     "spontaneously" out of disorder and chaos through a
     process of "self-organization". (p. xv)

     To show the differences between the deterministic,
equilibrium, and dissipative analyses, consider the research
approaches which study the history of state-level public policy. 
A survey of the literature shows two tracks: deterministic and
equilibrium.  The deterministic track monitors state-specific
characteristics in terms of socio-economic, structural, or
enclave-group variables.  The equilibrium approach recognizes
diffusion and intra-state (regional) influences.
     A more detailed discussion of these two approaches as
explanatory models for subnational policy-adjustment behavior is
reserved for Chapter Two.  For the moment, LET IT BE NOTED THAT
AN EXTENSIVE SEARCH OF THE LITERATURE REVEALED NO EVIDENCE THAT
AN ANALYTICAL APPROACH OF THE STUDY OF SUBNATIONAL POLICY
HISTORIES HAS UTILIZED A COMPLEXITY OR CHAOS THEORY APPROACH IN
GENERAL NOR THE DISSIPATIVE MODEL IN SPECIFIC.
     Perhaps one reason for the absence of the dissipative model
is that either the deterministic model or equilibrium model is
successful in capturing the dynamics of the system.  In fact,
neither model very convincingly explains state policy-adjustment
performances.  In considering deterministic analyses, internal
state characteristics which may influence policy adjustment
records include a variety of factors such as the governor's
electoral margin of victory, inter-party competitiveness, party
turnover, length of a governor's mandate, and the point in the
election cycle (see, as a sample, Lowi, 1963; Schlesinger, 1966;
Hofferbert, 1964, 1966; and Walker, 1969).  However, empirical
work has failed to find convincing support for the supposed
cause-effect relationship of the deterministic model.
     A special case of the deterministic model is Olson's (1982)
attention to the outcome of special-interest groups.  Here again
no consensus has been reached on the "institutional sclerosis"
thesis despite its frequent evaluation in the recent literature
(see and note bibliographies in Gray and Lowery, 1988; Brace,
1991; Hendrick and Garand, 1991; Garand, 1992; and Brierly and
Feiock, 1993).  Further, Grady (1987) reported finding little
evidence to support linkage of policy adjustment to high activity
levels by interest group and lobbyists (see Harrison and Kanter,
1978; and Jacob, 1986).  Overall, deterministic analyses have not
built a convincing case for explaining the history of subnational
policy adjustment.
     Turning to the equilibrium model, diffusion of innovation
stresses the role of external stimuli on state behavior and
recognizes that policy adjustment acts to re-establish
equilibrium between the state and its environment (see Walker,
1969, 1973; Gray, 1973a, 1973b; and Plaut and Pluta, 1983). 
Empirical evidence is more supportive of the equilibrium model
than the deterministic model.  For example, in addition to
testing two deterministic models, Grady (1987) analyzed state
behavior using an arms-race (diffusion) model (thus following
Peretz [1986]) to show influences on policy adjustment.  Grady
found that "...the data in general support the arms race hypothe-
sis" and therefore "it appears that the arms race model comes
closest [of the three models tested] to explaining the dynamics
of the growth and spread of business incentives across the
states" (Grady, 1987:92).  Grady's support of the arms-race model
is, however, less than enthusiastic.  He wrote that the arms race
model was only the best of the three models, the other two being
deterministic models for which he failed to find any support. 
More on Grady's findings later....
     No doubt both internal state characteristics and diffusion
do help explain some aspects of subnational policy histories. 
For instance, if the level of a state's political competitiveness
is sufficiently high, a minority-party governor will encounter
resistance in passing legislation through a hostile legislative
branch.  Further, contiguous states cannot fail to be aware of
what each other are doing.  However, without denying the
contribution these approaches make, THE FOCUS OF THIS RESEARCH IS
TO EMPHASIZE THAT BOTH THE DETERMINISTIC AND EQUILIBRIUM MODELS
FAIL TO CAPTURE MANY FUNDAMENTAL DYNAMICS OF SUBNATIONAL ECONOMIC
DEVELOPMENT EFFORTS.  IT FOLLOWS THAT THESE MODELS CANNOT
ADEQUATELY ANALYZE THE PHENOMENON.
     Noting the lack of success of deterministic and equilibrium
studies, however, is a necessary but not sufficient condition for
advocating research using the dissipative model.  The next step
is to suggest reasons why the dissipative system is a legitimate
contender (without the assumption of its victory).  Waldrop
(1993) and Leifer (1989) show why dissipative systems might be
important to the study of this phenomenon.  Waldrop's discussion
of complexity suggests a behavioral dynamic which might apply to
state behavior.  Viewing states as "agents", Waldrop (1993)
explained:

     These agents might be molecules or neurons or species
     or consumers or even corporations.  But whatever their
     nature, the agents were constantly organizing and
     reorganizing themselves into larger structures through
     the clash of mutual accommodation and mutual rivalry. 
     Thus, molecules would form cells, neurons would form
     brains, species would form ecosystems, consumers and
     corporations would form economies, and so on.  At each
     level, new emergent structures would form and engage in
     new emergent behaviors.  Complexity, in other words,
     was really a science of emergence. (p. 88)

The reference to molecules or neurons will be relevant later (see
Chapters Three and Four) as will the reference to species (see
Chapters Three and Five).  For the present, note that the system
of states might also display characteristics of complexity--
"mutual accommodation and mutual rivalry" along with "emergent
structures"--just as do molecules, species, and corporations.
     The dissipative model is also suggested by learning from the
failure of the two rival models--deterministic and equilibrium. 
Leifer (1989) suggested that the deterministic and equilibrium
models are ill-equipped to internalize the dynamics of the
phenomenon under study.  Leifer (1989), by drawing on Tichy and
Ulrich's (1984) work, points out that even an equilibrium system
"does not describe the profound transformational, discontinuous
change brought about by fundamental changes to the basic
political and cultural systems of the organization...as a result
of uncontrollable environmental turbulence".  Rather, those
changes within a dynamic environment such as state politics and
global influences on states are best captured in a dissipative
system.
     As noted, a thorough search of the literature reveals no
analysis using a dissipative system to explain subnational policy
adjustment.  This lacuna may be due to the relatively recent
recognition of dissipative system models in social science
phenomena.  WHATEVER THE REASON FOR THE ABSENCE IN THE LITERATURE
A DISSIPATIVE-SYSTEM APPROACH TO UNDERSTANDING POLICY BEHAVIOR,
THE PRESENT RESEARCH WILL EVALUATE SUBNATIONAL ECONOMIC
DEVELOPMENT POLICY-ADJUSTMENT RATES USING MODELS WHICH EMPHASIZE
THE DYNAMICS OF A DISSIPATIVE SYSTEM WITHIN THE LARGER CONTEXT OF
COMPLEXITY AND CHAOS THEORY. 

BUT WHY "NOISE"
AND EVOLUTION??

     Systems come in different types and display different
dynamics.  As for different types, systems can be random or they
can be organized along an infinite variety of deterministic
rules.  Physics helps determine whether behavior is random or
deterministic.  To make this distinction it refers to "noise". 
"Noise" is defined as "small changes" or "fluctuations" in a
variable.  Of particular importance to the present discussion is
"flicker noise".  Consider the implications of flicker noise in
light of the following observations.  Bak, Tang, and Wiesenfeld
(1988:34) wrote

     The same interdependence of species also makes the
     ecosystem very susceptible to small changes or "noise". 
     However, the system cannot be too sensitive since then
     it could not have evolved into its present state in the
     first place.  Owing to this balance we may say that
     such a system is "critical".  We shall see that this
     qualitative concept of criticality can be put on a firm
     quantitative basis.
          Such critical system are abundant in nature.  We
     shall see that the dynamics of a critical state has a
     specific temporal fingerprint, namely "flicker noise",
     in which the power spectrum S(f) scales as 1/f at low
     frequencies.  Flicker noise is characterized by
     correlations extended over a wide range of time scales,
     a clear indication of some sort of cooperative effect.
     ....  We shall argue that flicker noise is in fact not
     noise but reflects the intrinsic dynamics of self-
     organized critical systems. 

     As a first step to researching the "intrinsic dynamics of
self-organized critical systems" and thereby determining (even
measuring) state influences on each other, distinguish noise with
three colors: "white noise", "pink noise", and "brown noise". 
Notably, white noise and brown noise (a.k.a. Brownian noise and
Weiner process) describe random movement.  As Schroeder
(1990:121) describes it

     the POWER SPECTRA..., often known as NOISES, seem
     addicted to simple, homogeneous power laws in the form
     f-  as functions of frequency f.  Prominent among these
     is WHITE NOISE, with a spectral exponent of  =0.  Thus,
     the power spectrum of white noise is independent of
     frequency.  But white noise, that is, a noise with a
     constant flat power spectrum, is a convenient fiction--
     a little white lie. ....  If we integrate a white noise
     over time, we get a "brown" noise, such as the
     projection of a Brownian motion onto one spatial
     dimension.  Brown noise has a power spectrum that is
     proportional to f-2 over an extended frequency range.
     (emphasis in original) 

     Pink noise, in contrast, which has a power spectrum f-  such
that 0>-BETA>-2, has been described as self-organized criticality
which shows deterministic behavior.  Thus, a clear line is drawn
between pink noise and white/brown noise: the former is
deterministic in origin and the latter are random.  It follows
that IF DATA ON SUBNATIONAL POLICY ADJUSTMENTS REVEAL PINK NOISE
(TO THE EXCLUSION OF WHITE AND BROWN NOISE), THE SYSTEM UNDER
STUDY FUNCTIONS IN A NON-RANDOM MANNER.  Further, it is incumbent
on this research to provide a preliminary description of the non-
random characteristics of the system.
     To pursue this research objective, two models--white and
brown noise combined and considered separately from pink noise--
are borrowed from physics for their direct application to
subnational policy-adoption rates.  A more detailed analysis of
each type of noise is reserved for later presentation (Chapters
Three and Four); in the meantime it should be noted a limited
number of technical articles have already appeared linking
physics noise to a variety of social science phenomena.  Musha
and Higuchi (1977) related noise to traffic behavior.  Sornette
and Sornette (1989:202) foresaw that SOC 

     would also be relevant for other problems such as the
     average seasonal temperature, annual amount of
     rainfall, rate of traffic flow, etc.  In these systems,
     there is also a competition between an average incoming
     stationary flux with inhomogeneous boundary conditions
     which is released by sudden and sometimes catastrophic
     events.  With the appropriate translation, the
     preceding discussion of the size-frequency relationship
     and on the return time of extreme events should apply.

     Determining which type of physics noise best explains
actions by U. S. states is the first step in this research
(results are presented in Chapter Four and discussed in Chapter
Six).  The second step is to determine what competitive dynamics
characterize the states.  To achieve this second objective, four
models of evolutionary biology are used.  At first blush
evolutionary biology may seem an odd selection, in particular
coming after physics.  However, the study of physics noise has
been linked to evolution of species in the above citation from
Bak, Tang, and Wiesenfeld, as well as separately by Kauffman
(1993) and Gingerich (1993).  Kauffman's (1993) work has
suggested that biological evolution may be attributable to
interaction between species in the ecosystem.  Gingerich (1993)
applied logarithmetic analyses to the study of directional change
or stasis in species.  Furthermore, recall that evolution takes
place in an (eco)system and involves agents (species) seeking
their individual goals where resources may be finite.  Don't
states (species) also operate (at least sometimes) in the federal
system and seek their individual goals where supply may be
finite?
     Evolutionary models have been applied to additional social
science phenomena, most notably by Axelrod in his "An
Evolutionary Approach to Norms" (1986), THE EVOLUTION OF
COOPERATION (1984, 1991) and other treatments of the iterated
Prisoners' Dilemma situation (1980a, 1980b, 1987, 1988 [with
Dion], 1981 [with Hamilton]).  Additional work on evolution as it
applies to norms is found in Hayek (1967) and Peyton Young (1991,
1992).
     Other systems where competition is believed to be present
have been described by reference evolutionary models.  Notable
are McCain's (1992) A FRAMEWORK FOR COGNITIVE ECONOMICS and Meier
and Haury's "A Cognitive-Evolutionary Theory of Economic Policy"
(1990) for bring biological models in economics.  Further,
Alchian's 1950 article "Uncertainty, Evolution and Economic
Theory" and its 1953 sequel "Comment: Biological Anologies in the
Theory of the Firm" sparked discussion of evolutionary forces in
firm behavior (Enke, 1951, 1953; Penrose, 1952, 1953, 1959;
Winter, 1964, 1971, 1975, 1987; Nelson, 1974, 1977; Nelson and
Winter, 1973, 1982; Nelson, Winter and Schuette, 1976; Schuette,
1980; Schuette and Winter, 1975, Iwai, 1981a, 1981b; and Winter
and Rothblum, 1982).  Thus, as further explained by Chattoe
(1994):

     Although Alchian does not provide a formal model, he
     explicitly attempts to specify a complete analogy
     between biological and economic evolution in the
     context of the behaviour of firms.  The 'genes' of each
     firm are the strategies it uses in order to reach
     decisions or conclusions, for example deciding to
     implement marginal cost pricing or 'cost-plus' pricing. 
     These genes are 'exchanged' when firm imitate the
     observed strategies of others and 'mutated' by trial-
     and-error innovation.  Selection pressure is exerted by
     the need to make positive profits in order to remain in
     business.  

     Chattoe (1994) also discusses many other applications of
evolutionary models in social science, including the stock
market, innovation and technological progress, use of money as a
medium for exchange, the existence of self-fulfilling prophecies,
speculative bubbles, and technological lock-ins.  And, of course,
the numerous reference to "social darwinism"  fall into this
category.

OBJECTIVES AND
RESEARCH STRUCTURE

     This chapter opened by referencing a question which deserves
an answer: What factors explain state policy-adjustment rates? 
The answer to this question has been the goal (grail?) of
distinguished researchers and numerous publications; however, the
goal has not been reached.
     But failure is not defeat.  A clear opportunity for new
inquiry exists for learning both how subnational behavior relates
to economic development policy adjustment, and how it relates to
other features of states in a federal structure.  This
opportunity has emerged due to the recent attention to complexity
and chaos theory and their applications to the social sciences.  
     The chapters which follow draw on complexity and chaos
theory by utilizing economic development as a case study of
subnational policy-adjustment behavior.  It is expected that this
analysis will open a third source of inquiry into state behavior,
adding the dissipative model to the deterministic and equilibrium
models.  
     To achieve this goal, six chapters are presented.  In
writing a dissertation entitled "Log-Linear 'Noise' and Models of
Evolutionary Biology in Public Policy Research: An Application to
Subnational Policy-Adjustment Behavior in Economic Development",
Chapter One (Research Proposal) identifies the lacuna in the
study of state behavior and justifies the approach used in the
present research.  
     Chapter Two (Literature Review) and Chapter Three (Models,
Hypotheses, and Data) present the orientations and means of
analysis for carrying out the objectives of the inquiry.  Chapter
Two identifies the bodies of literature which are relevant to the
discussion.  Chapter Three provides an overview of the six
models, how each can be tested, and the source of data.
     The next two Chapters--Chapter Four (The Physics of "Noise")
and Chapter Five (Biological Models of an Evolutionary System)--
put the data to the test.  Chapter Four identifies whether U. S.
states behave as described in a self-organized system or
according to Brownian movement.  Chapter Five tests four
biological models of competition: Red Queen Hypothesis,
Punctuated Equilibrium, Stationary Model, and Turnover-Pulse
Hypothesis.
     Chapter Six (Conclusions) presents the conclusions of this
research.  Work is summarized, implications identified, and
potential for future work recognized.
     The next chapter reviews the literature of state policy-
adoption behavior.  The discussion is divided according to the
models used: deterministic or equilibrium.  No inquiry has yet
used a dissipative analysis.  Also, a brief introduction of
complexity and chaos theory, as well as their application to the
social sciences, is provided.





                               Chapter Two
                                    
                            Literature Review


     CHAPTER ABSTRACT: This chapter presents an overview
     (select bibliography and discussion) of past research
     into state policy-adjustment behavior as well as
     complexity and chaos theory.  The joint emphasis is
     placed on understanding research opportunities and
     restraints which result from the theoretical
     orientations used.
          A unique division is observed in the presentation
     of past work: analyses which stress the deterministic
     model are presented separately from analyses using
     equilibrium models (to the extent possible given some
     treatments deal with both).  Thus, the discussion of
     past research on state policy-adjustment behavior
     presents two discussions on deterministic models
     (Olson's thesis and all others) and one on equilibrium
     models.
          The chapter closes with a look at the theoretical
     orientations of complexity and chaos theory and their
     role in social science inquiry.


POINT OF DEPARTURE

     Writing a dissertation on the application of a dissipative
system to a public policy issue has had three interesting
aspects.  First, to explore this field requires consulting
journals such as EVOLUTIONARY THEORY, JOURNAL OF GEOPHYSICAL
RESEARCH, and PHYSICAL REVIEW LETTERS as well as the more
familiar JOURNAL OF POLITICS and AMERICAN POLITICAL SCIENCE
REVIEW.  Journal articles carry the pulse of the inquiry;
however, even journals do not appear quickly enough.  Perhaps
indicative of the times, some of the information needed for
preparing this dissertation has been gathered via e-mail and
Internet through direct contact with researchers and lists such
as NLIN-SYS (non-linear systems) and EMSS (evolutionary models in
the social sciences).  If ever there is a case to be made for
spontaneous order out of chaos, Internet lists will have to be
cited.  Will all dissertations in the future ride the
superhighway?
     The second aspect of work on complexity and chaos theory is
also evident from scanning the bibliography: it can be noted that
50% of the sources cited (excluding the literature review) were
not even in print at the time I took my qualifying exams in 1988.
Indeed, one essential model--systematic testing of self-organized
criticality in sandpiles--appeared in the literature only in
1990.  
     Third, research in the public-policy application of
complexity theory has returned me to my intellectual roots.  As
an undergraduate I majored in Human Evolution.  In that pursuit I
encountered many of the concepts in evolutionary biology which
I've rediscovered and used in the present research (although I
must admit the physics has been a new experience).  It has been
fun (and more than a slight case of dj vu) to consider once
again the nuances of phenotypes, speciation, and natural
selection.  Also, the literature of evolutionary biology has
recalled to mind the intellectual accomplishment of a personal
hero--Charles Robert Darwin.  Again seeing his influence, strong
after so many years, informing contemporary, cutting-edge thought
has reaffirmed my own conviction concerning his personal and
intellectual triumphs. 

SUBNATIONAL POLICY-ADJUSTMENT BEHAVIOR

     Dawson and Robinson (1963) clarified the rationale and
approach to research on the history of public policy adoption. 
They stated

     We begin with the assumption that public policy is the
     major dependent variable which political science seeks
     to explain.  The task of political science, then, is to
     find and explain the independent and intervening
     variables which account for policy differences [between
     states]. (p. 266)

This admirable goal--that of explaining part of the policy
process--has not been realized, even given the thirty-year
interval.  More sophisticated empirical tests have refined the
work.  Also, the phenomenon has been further sorted into policy
topologies and outcome expectations.  However, a consensus has
not emerged.  
     Following observations made in Chapter One, the present
discussion of subnational behavior separates this discussion
according to the deterministic or equilibrium approach used for
analysis.  Further, to facilitate presentation, the deterministic
model is divided into (first) an examination of Olson's (1982)
thesis as stated in THE RISE AND DECLINE OF NATIONS and (second)
other deterministic explanations.

     SUBNATIONAL BEHAVIOR - DETERMINISTIC MODEL

     PRIMARY SOURCES Dawson and Robinson, 1963; Schubert and
     Press, 1964; Hofferbert, 1964, 1966; Sharkansky, 1967;
     Fry and Winters, 1970; Jennings, 1979; Dye, 1980, 1984;
     Olson, 1982; Mueller (ed.), 1983; Grady, 1987; Gray and
     Lowery, 1988, 1989; Ambrosius, 1985, 1989; Brace and
     Cohen, 1989; Brace, 1985, 1988, 1991; Garand, 1992; and
     Brierly and Feiock, 1993.


     OLSON'S (1982) THESIS
 
     Studying the "mysterious decline or collapse of great
empires or civilizations" (p. 1), Mancur Olson (1982) identified
nine factors which account for this outcome.  Part of Olson's
thesis rests on the behavior of enclave groups and special
interests in a larger system.  As such, Olson's Points Two,
Three, and Four are relevant to state economic development. 
Olson noted (p. 74)

     2. Stable societies with unchanged boundaries tend to
     accumulate more collusion and organizations for
     collective action over time.

     3. Members of "small" groups have disproportionate
     organizational power for collective action, and this
     disproportion diminishes but does not disappear over
     time in stable societies.

     4. On balance, special-interest organizations and
     collusion reduce efficiency and aggregate income in the
     societies in which they operate and make political life
     more divisive.

     Dubbed "institutional sclerosis" by Dye (1980), the thesis
stated that older organizations, as found in stable democracies,
develop special-interest subgroups which act to protect and
promote themselves at the expense of the larger group.  The more
active the older and better-positioned subgroups, the slower the
growth rate for the group as a whole.  As this may apply to
states seeking economic development, a special-interest group may
succeed in securing legislation which benefits itself but which
results in a loss of greater benefits for the state as a whole.  
     Early work supported Olson's thesis, even before THE RISE
AND DECLINE OF NATIONS appeared in print.  At a December, 1978,
conference in College Park, Maryland, support for Olson's thesis
was marshalled through a variety of theoretical and empirical
lines of evidence using a wide range of countries (although all
Western) for support.  In all, thirteen articles appear in Dennis
C. Mueller's (1983) anthology of research articles from the
conference.
     Other empirical work on Olson's thesis has been offered. 
According to Gray and Lowery (1988), Olson's thesis was
empirically examined on separate occasions by Choi (1979), Brace
(1985), and Brace and Dudley (1985).  However, these studies,
given as conference papers, have not appeared in print.  Also not
in print are analyses by Ambrosius (1985) and Brace (1988). 
Nevertheless, when taken together, these numerous sources seemed
to provide a strong testing of institutional sclerosis.
     Dye (1980), working from a preliminary draft of Olson's
longer work, considered the age of a state in his analysis of
policy adjustment.  (State age began with admission to the Union
or, for the Southern states, at the end of the Civil War.)  Dye
found that "age...turned out to be a very influential determinant
of economic growth in every problem in which it was entered" (p.
1100).  He concluded that "'older' states cannot maintain the
economic growth rates of 'younger' states, independent of any
other characteristic of younger or older states.  Age itself
appears to be a barrier to economic growth" (pp. 1100-1101).
     Gray and Lowery (1988) reviewed the above studies and were
not convinced, however.  They concluded that "when better
measures are employed [than in earlier studies] and the key
linkages are specified, the supportive evidence for Olson in the
case of the U. S. states largely evaporates" (pp. 128-129).
     However, what goes around, comes around: Brace and Cohen
(1989) were not convinced with Gray and Lowery's analysis.  Brace
and Cohen point to two weaknesses in Gray and Lowery's research. 
First, they incorrectly quantified the power of groups.  Second,
they mishandled numerous specification problems.  According to
the argument, Gray and Lowery failed to adequately handle
exogenous variables.  As such, Brace and Cohen wrote "we want to
suggest that more of an effect should be made to IDENTIFY MAJOR
EXOGENOUS VARIABLES AND TO CONTROL FOR THEM in the empirical
analysis of the impact of endogenous factors on the economic
performance of states" (p. 1301; emphasis added).
     Accompanying Brace and Cohen's critique was a reply by Gray
and Lowery (1989).  They acknowledged some aspects of the
challenge while rejecting others.  Overall they point to a
general failure by past studies to support Olson's thesis and
cited the inability to adequately test it empirically given the
availability of data and the problems with the thesis itself.
     Work on a different approach to Olson's thesis by Grady
(1987) has already been mentioned (Chapter One).  Testing for the
impact of special-interest groups on state policy adjustment
records, Grady is unable to confirm Olson's thesis that states
with more active groups would adopt greater numbers of policies.
     Empirical work by Ambrosius (1989) took a larger sweep at
Olson's thesis.  Ambrosius linked Lindbloom's attention to
political influences to Olson's logic.  Thus, "if the focus
shifts, from Olson's discussion of coalition activity in creating
ECONOMIC obstructions, to the POLITICAL influence of occupational
interests, whether through group activity or through Lindbloom's
state of 'privilege', then a sequence of hypothesized effects can
be examined" (p. 55; emphasis in original).  Using this approach
to study states for a fourteen-year period, Ambrosius concluded
that "the impact of the occupational interest strength variable
was consistently positive for all categories of economic
development policies" (p. 64).  The data thus confirmed that high
occupational interests at the state level correlate to higher
levels of adjustment of economic development policies.
     Recent work by Garand (1992) sought to "explore the degree
to which the relationship between state age and economic growth
is time-dependent" (p. 471).  He used data from 1946 to 1984. 
His first step was to determine whether the states demonstrated
stability in levels of their economic growth over time, thus
fulfilling the assumption of cross-sectional test on Olson's
thesis.  Garand reports that the data "suggest substantial
instability in economic growth rates for the American states over
time" (p. 474).
     Next, Garand looked at the influence of age on economic
performance.  He was clear in his hypothesis and findings.  He
wrote:

     the hypothesis underlying this model, one would
     expect older states (i.e., those presumably dominated
     by organized interests) to exhibit lower relative
     levels of economic growth than younger states (i.e.,
     those with less-developed organized interests).
          For the single-year models, one finds wide
     variability in both the direction and magnitude of the
     state age coefficients.  Of the 39 coefficients, 25 are
     in the expected negative direction (19 of which are
     significant at the .10 level), while 14 are positive
     (with two significant). (pp. 475-476)

     Interestingly, the period which best supports Olson's thesis
falls between 1968 and 1982, the years used by Olson (1982), Dye
(1980), and Choi (1979) when they found support for the thesis. 
Thus, Garand continued, "when one moves beyond the period from
1968 to 1982, one finds much less support for the expected
relationship between state age and economic growth" (p. 478).
     Garand concluded against Olson's thesis while nuancing his
statement.  He summarized: 

     State economic growth is not highly stable in the post-
     World War II era, with some states exhibiting high
     economic growth during some years, and others enjoying
     high economic growth during other years.  Most
     importantly [sic.], however, is the finding that the
     relationship between state age and economic growth is
     unstable and highly time-dependent, with the inferences
     drawn from cross-sectional tests of Olson's thesis
     shown to be a function of the years utilized in one's
     analysis.  All of this calls into serious question the
     degree to which cross-sectional methods are adequate to
     test Olson's thesis, as well as the degree to which one
     can accept findings of support for Olson's thesis in
     the first place. (p. 480)

     The latest research in the literature again found support
for Olson's thesis.  Brierly and Feiock (1993) operationalized
economic growth in states as the change in total personal income
for each state over three six-year periods (1970-1975, 1975-1980,
and 1980-1985).  They summarized their findings as follows:

     State economic growth appears to be driven by capital
     and organizational changes.  As expected, change in
     capital was positively related to income growth.  The
     negative organizational relationship is consistent with
     the Olson thesis and previous results for the 1970-1985
     time period (Garand, 1992).  The period dummy variable
     estimates further illustrate the significant slowdown
     in income growth from 1975 to 1985.  Controlling for
     the influence of capital, these results provide modest
     support for the Olson contention that the slowdown in
     growth during this period is attributable to
     organization. (p. 664)

     In summary, despite attention in the literature to Olson's
thesis, the debate has not culminated in a strong case in favor
or in opposition to it.  Indeed, some of the further work which
must be done on the thesis is providing a clearer statement about
the thesis itself and how its variables can be measured, tested,
and conclusions drawn (see Gray and Lowery, 1989). 
 
     OTHER DETERMINISTIC ANALYSES

     Dawson and Robinson (1963) reminded scholars that "political
science is concerned with how various formal and informal
institutions, economic, social, philosophical and geographical
conditions influence the adjustment and implementation of policy"
(p. 265).  They pointed to a list of other research on the topic,
including David Easton's THE POLITICAL SYSTEM (1953) and Harold
Lasswell's (1951) "The Policy Orientation".  Additional stand-out
works include V. O. Key's SOUTHERN POLITICS IN STATE AND NATION
(1951) and Thomas Dye's POLITICS, ECONOMICS, AND THE PUBLIC
(1966).
     In their own research, Dawson and Robinson set out "to
investigate the relation between political process and the
policies adopted by political systems" (p. 265).  Specifically,
they posited that "the greater the degree of inter-party
competition within a political system[,] the more extensive or
'liberal' the social welfare policies a political system will
adopt" (p. 281).  They concluded that their "findings seem to fit
the relations hypothesized in the theoretical scheme" (p. 287). 
However, they added that the system variables are held constant.
     Not all deterministic analyses were so successful.  As
evidenced in what follows, factors which THEORETICALLY influence
policy adjustment are many.  Verifying the influence using data,
however, has been no easy task.  For instance, using Schubert and
Press' (1964) data, can malapportionment levels be correlated to
policy adjustment preferences?  In fact, the Spearman's Rank
Correlation value ( , rho) is 0.074, or almost entirely
unrelated.  Using economic development records, results were
generated for Hofferbert's (1964) inter-party competitiveness
scale with a rho =0.055 and  rho =0.038 for his welfare
orientation (Hofferbert, 1966).
     Some factors, in contrast, have been shown to be correlated
to state behavior.  For Hofferbert (1966), however, only
environmental factors are relevant.  According to him,

     The thesis advanced here is that differences in policy,
     at least in certain substantive areas, are more readily
     explained in terms of differences in the socio-economic
     environments of the states than by an examination of
     structural variables. (p. 73) 

Thus he found results in keeping with his expectations.  Notably,
he wrote:

     Thus far, the limited evidence presented indicates
     there is no significant relationship between
     apportionment and public policy, apportionment and
     partisan competition, apportionment and divided
     control, divided control and public policy, and between
     the party in power and public policy. (p. 81)

Far from becoming the last word on the topic, Hofferbert's
conclusions did not keep Jennings from different findings
concerning inter-party competition.  Jennings (1979) wrote:

     The weight of the evidence produced by this analysis
     provides strong support for the thesis that the nature
     of partisan alignments and partisan control of
     government under conditions of class-based politics
     influence welfare policy outcomes.  Class-based and
     non-class-based political competition produce quite
     different sets of welfare policy outcomes in the eight
     states included in this analysis.  Control of
     government appears to be important when parties or
     factions divide the electorate along lines of economic
     cleavage. (p. 427)

Even more notable is Fry and Winters' (1970) rebuttal to
Hofferbert's emphasis on socio-economic variables.  They
concluded:

     We have been able to develop a statistical model that
     accounts for more than half the variance in
     redistribution among the 48 states [excluding Alaska
     and Hawaii] included in the analysis and more than two-
     thirds of the variance in non-Southern states.  Second,
     and more significantly, we have found that the
     political variables employed in the model are
     considerably more powerful than the socio-economic
     variables in explaining interstate variations in
     redistributive patterns.  Thus, our data not only
     support the assertion that politics makes a difference,
     they suggest that politics plays a dominant role in the
     allocation of the burdens and benefits of public
     policies. (p. 522)

     Better results linking state characteristics to its behavior
have been shown by analyzing single factors rather than large
categories.  Thus, Dye (1980) sought to explain effects which
state taxing and spending policies have on rates of economic
development.  He performed a time-lagged analysis to show
possible influences on state economic growth.  He found "little
association between tax policy and economic growth" (p. 1098). 
In contrast, "spending policies of states in the late 1960s are
indeed associated with economic growth rates in the 1970s" (p.
1098).  In particular, highway spending emerged as the strongest
correlate of economic growth of all the variables he studied.
     Dye tested fifteen other variables under the rubrics of
taxing (six), spending (four), redistribution (two), and size and
development (three).  His findings were generally consistent:

     Aside from highway spending, there were no policy
     variables which independently affected growth rates. 
     Right to work laws did not help or hinder economic
     growth.  None of our taxing measures--total tax burden,
     sales tax burden, income tax burden, corporate tax
     rate, or "worker" or "executive" income tax liability--
     were INDEPENDENTLY associated with economic growth. 
     Spending for education, welfare, and health were not
     INDEPENDENTLY related to the economic growth.  Our
     hypothesis that educational spending in the 1960s might
     contribute to economic growth in the 1970s did not
     stand up under controls for age [i.e., age of the
     state] and highway spending.  Indeed, the only
     discernable effect of educational spending on economic
     growth, controlling for other variables, is a negative
     one. (p. 1101; emphasis in original)

     Findings by Grady (1987) apply here as well.  He tested a
correlation between state policy adjustment behavior and
unemployment levels.  His analysis rejected the association,
showing that policy adjustment rate were not related to
unemployment.
     Brace (1991) started with the premise that "the influence of
states on their economies is not constant but changing and that
this influence is conditioned by many forces beyond state
control" (p. 298).  The discussion of this second premise is
reserved for treatment later (under equilibrium models).  As for
Brace's analysis, he divided his data into three panels: 1968-
1973, 1974-1979, and 1980-1985.  He concludes that "for the
majority of the period studied, states had no statistically
discernible impact on their economies" (p. 312).  However,
increasing economic activism is reflected in more recent years.
     Overall, the definitive link between state characteristics
and policy preferences remains elusive.  Two sources of error in
these analysis might be present.  First, it could be that the
phenomena under study--welfare orientation, malapportionment,
competitiveness--are too complex or not easily quantifiable for
the uses made of them.  Sharkansky (1967), in his analysis of
state government expenditures, properly noted that "it is
apparent that changes in expenditures respond to a large number
of stimuli that vary in their composition from state to state"
(p. 184).  Not surprisingly, Sharkansky was unable to find a
strong relationship between expenditure changes and any of his
independent variables.  The same problem could be at hand with
the studies cited above.  
     A second explanation for failed results might be the
methodology used to test the hypotheses.  In fact, of the issues
raised in this literature, methodological disputes rank high
(Cnudde and McCrone, 1969; Brace and Cohen, 1989; Gray and
Lowery, 1989; and others).  Indeed, Cnudde and McCrone made the
case clear with they wrote:

     Our analysis leads us to several major conclusions. 
     First, our examination of the methodology utilized in
     previous research into the effects of political
     variables on state expenditure policies convinces us
     that the spurious model now predominant in state
     politics literature rests on shaky empirical
     foundations.  Reliance on correlational analysis can
     lead to unwarranted causal inferences. (p. 865)   

     Significantly, there is some awareness in the deterministic
assessments that the system of states influences individual state
behavior and preferences (see citations above from Brace and
Cohen, 1989, and Brace, 1991).  While there is no assumption that
this influence is similarly perceived or acted upon across the
fifty states, the role of the system of subnational agents must
be acknowledged if state behavior is to become clearer.
     In addition to deterministic models of subnational behavior,
inquiry has proceeded using equilibrium models.  The next section
of this discussion treats this second focus of inquiry.

     SUBNATIONAL BEHAVIOR - EQUILIBRIUM MODEL

     PRIMARY SOURCES: Walker, 1969, 1973; Gray, 1973a,
     1973b; Plaut and Pluta, 1983; Grady, 1987; Jones, 1990;
     Brace, 1991; and Hendrick and Garand, 1991.

     EQUILIBRIUM ANALYSES

     The equilibrium approach to understanding subnational
behavior recognizes the systemic (regional and/or national)
influences on member agents.  Given an equilibrium system,
movement by one agent will result in new influences exerted on
other subnational agents.  The degree to which one state's
behavior influences other states, and the resulting behavior by
other states, is expected to vary over time and distance.  Coping
mechanisms within the system will result in subnational behavior
which restores the systemic steady state.
     How subnational agents interact--either through learning,
diffusion, cooperation, competition, or combat--is a necessary
aspect of this research.  To begin the review, recall Walker's
(1969) discussion of the diffusion of innovations among U. S.
states.  He was responding to deterministic studies (cited above)
and attempted to show the role of diffusion in "one of the most
fundamental policy decisions of all: whether to initiate a
program in the first place" (p. 880).
     At the end of his analysis he produced composite innovation
scores for the states.  These scores were based on how quickly a
state adopted an innovation relative to other states' adjustment
(or non-adjustment) behavior.  The score ranged from New York
(the high with 0.656) to Mississippi (the low with 0.298).  With
the maximum at 1.000 and minimum at 0.000, the range was 0.358.
     Walker's score was based on the adjustment record of 88
policies.  He selected "six to eight different pieces of
legislation in each of these areas: welfare, health, education,
conservation, planning, administrative organization, highways,
civil rights, corrections and police, labor, taxes, and
professional regulation" (p. 882).  Thus, the score combined
"beauticians licensing" with "fair housing--urban renewal areas"
and "probation law".
     Walker's paper prompted a challenged (Gray, 1973a), defense
(Walker, 1973), and truce (Gray, 1973b).  As this dialogue
relates here, Gray discussed Roger's (1962) expectation of an S-
shape curve.  Thus, innovators and laggards account for the
extremes in a normal distribution of policy adjustment. 
Elaborating, Gray pointed out:

     Actually, the normality or non-normality of the adopter
     distributions is independent of the theoretical
     assumption that ideas spread because adopters somehow
     influence nonadopters.  OTHER CURVES, PARTICULARLY THE
     LOGISTIC CURVE OF POPULATION GROWTH, have been widely
     used to fit the same kind of data (innovations) with
     the same goodness of fit. (p. 1176; emphasis added)

Finding non-linear logistic curves is important since (in keeping
with complexity and chaos theory, discussed below) it suggests a
different, non-linear system dynamic from one which produces
normal distribution curves.  
     Further, Gray challenged the use of such a broad list of
policies to arrive at an index of innovativeness for the states. 
She preferred that the policies be segregated into policy domains
before analysis (an opinion shared in the present research). 
Only such an approach to the data would reveal differences
between policy adjustment based on characteristics of the policy
domain itself.
     Walker (1973), in his response to Gray, repeated his point
of departure.  He stated:

     My research specifically was designed to show that the
     American states are NOT a completely intermixed
     population, but rather are organized into a complex
     system of pioneers, regional leaders and laggards.  I
     identified some important developments that were
     creating pressures for national uniformity, such as the
     growing cosmopolitan communities of public officials
     and policy experts, and made some rather oblique
     efforts to measure these trends, but my conclusion was
     that regional differences were remarkably persistent
     and unlikely soon to disappear. (p. 1187; emphasis in
     original)

Walker's reference to "a complex system" is unaware of the use of
that term in complexity and chaos theory.  Nor is he aware (cited
below) of the same significance of his remarks to "an immensely
complicated social system" and "random occurrences and chance
factors".  Nevertheless, Walker does show, in the vocabulary of
complexity and chaos theory, an emerging awareness of the
presence of complexity in this field of inquiry.  As such, Walker
wrote:

     We are studying AN IMMENSELY COMPLICATED SOCIAL SYSTEM
     IN WHICH RANDOM OCCURRENCES AND CHANCE FACTORS are
     prominent ingredients, but regardless of these
     difficulties I am not prepared as yet to give up my
     emphasis on organization factors as keys to understand
     [diffusion of innovation]..." (p. 1190; emphasis
     added).

     Independent of Walker, Plaut and Pluta (1983) developed a
disequilibrium-adjustment model to study how state
characteristics might influence growth.  Their model is based on
the assumption that economic growth opportunities will go to
states having the greatest number economic development policies. 
According to the model, economic growth accumulates in the states
with more policies until other states adjust for the disparity by
adopting policies to restore the equilibrium.  Their model does
not, however, explain why states would discontinue economic
development polices.
     For their own findings, they showed that tax effort,
expenditures, and the "business climate" did not correlate to
subnational growth.  Further, they showed that employment growth
was slowed by tax and "business climate" variables.
     This model was used, with some modification, by Jones
(1990).  He "added CONTROLS FOR REGIONAL DIFFERENCES AMONG
STATES" (p. 222).  He concluded that 

     This analysis of state expenditures and economic growth
     during the period 1964-1984 using a disequilibrium-
     adjustment model indicates that overall expenditures do
     not cause economic stagnation.  It may be the case that
     total public sector expenditures can interfere with
     economic vitality, but not within the range of existing
     variation among states. (p. 230)

     Grady (1987) also discussed diffusion of policies using the
arm's race model of Peretz (1986).  In comparing the arms race
diffusion model to two others he tested (see above), Grady found
that the arms race model best explained the phenomenon under
study.  However, his findings must be put in the context of his
failure to find any verification at all in favor of enclave
interest groups (Olson's thesis) and unemployment rates
(deterministic model).  When put in perspective, Grady's findings
showed BETTER, but perhaps not ADEQUATE, results.
     Brace (1991), already discussed above, also treated states
as agents in a federal system.  Brace noted that states became
the primary agents for action as "the national economy has been
placed under greater pressure globally, and as the limits of
federal policy have been reached domestically" (p. 312).  To test
his hypothesis, Brace used multiple variables, including 1)
institutional capacity of governors and legislatures; 2) economic
development policies; 3) state fiscal polices; 4) general
influences of regional/national effects and period effects; and
5) specific influences of energy, unionization, and defense
expenditures.
     The notable aspect of Brace's list of variables was that
they included variables drawn from the state, regional and
national levels.  He explained: 

     Most certainly national and regional forces exert
     influences on state economies that state level
     political actors cannot control.  Regional indicator
     variables are employed to control for regional effects. 
     The effect of the national economy is controlled by the
     inclusion of time point indicator variables that
     measure variation common to all states (i.e., national)
     in a given year. (p. 301)

Thus he recognized that states exist as separate decision making
entities but are within a system.  
     Last, Hendrick and Garand (1991) were very concerned with
separating state, regional, and national influences on states. 
They pointed out early in their analysis that "it is clear that
state economies are (at least to some extent) interdependent, and
that there exist national and regional economic factors that have
relatively strong, uniform impacts on economic growth across
states" (p. 1094).  
     To test the degree of control a state has over its economic
development, Hendrick and Garand used a variance components model
similar to one used in assessing electoral nationalization
(Stokes, 1965, 1967; Claggett, Flanigan, and Zingale, 1984; and
Vertz, Frendreis, and Gibson, 1987).  Personal income data were
used to assess growth.  
     Tests showed the relative influence of state, region, and
nation.  They wrote:

     the state component of variance in state economic
     growth dominates the national and regional components
     during the postwar era.  For each indicator of economic
     growth, the state component represents at least 61% of
     the variance, with the national component representing
     between 25% and 33%, and the regional component taking
     up the remaining 5%-6%.  HENCE, BETWEEN 31% AND 38% OF
     VARIANCE IN STATE ECONOMIC GROWTH IS EXOGENOUSLY
     DETERMINED, THAT IS, ATTRIBUTABLE TO NATIONAL AND
     REGIONAL EFFECTS GENERALLY OUTSIDE THE CONTROL OF THE
     STATE ECONOMIC AND POLITICAL SYSTEM.  THE REMAINING 62%
     TO 69% IS THE COMPONENT OF THE VARIANCE THAT IS UNIQUE
     TO THE STATES AND PRESUMABLY WITHIN ITS CONTROL. (pp.
     1101-1104; emphasis added).

     Hendrick and Garand pointed to three reasons for the role of
the national influences: the active postwar role of the federal
government; intergovernmental grants and policies of
redistribution; and international competition with uniform
impacts across regions.
     Overall, Hendrick and Garand presented a fitting summation
of the work on deterministic and equilibrium approaches to state
behavior.  They noted

     The question of exogeneity in the determination of
     state economic growth is not a trivial one.  Numerous
     scholars have attempted to assess the linkage between
     the political and economic systems within the American
     states, primarily in their attempts to test Olson's
     (1982) organized interest theory of economic growth. 
     With only a few exceptions (Brace, 1989), these
     scholars have depicted this political-economic
     interaction as endogenous to the states, with virtually
     no impact on exogenous factors on state economic
     growth.  Our results suggest that such an assumption is
     unrealistic, and that models of state economic change
     relying on state-level characteristics are
     underspecified.  Clearly, based on our results,
     ATTEMPTS TO DEVELOP MODELS OF STATE ECONOMIC GROWTH
     MUST ALLOW FOR THE POSSIBILITY THAT THE DEPENDENT
     VARIABLE IS A FUNCTION OF EXOGENOUS FACTORS THAT SEEM
     TO EXERT SUBSTANTIAL INFLUENCE AND AFFECT THE VARIABLE
     DIRECTLY AND INTERACTIVELY WITH ENDOGENOUS FACTORS.
     (pp. 1107-1108; emphasis added)

Although Hendrick and Garand had other goals in mind when making
this statement, their conclusions are relevant to an analysis of
subnational policy-adjustment rates using a dissipative model.

COMPLEXITY AND CHAOS THEORY

     The review of literature now turns to complexity and chaos
theory as they relate to the present research goals.  Thus, a
complete overview of complexity and chaos theory is postponed. 
Rather, the discussion shows general aspects of both, as well as
their social science applications.

     COMPLEXITY AND CHAOS THEORY

     PRIMARY SOURCES: Jantsch, 1980; Prigogine, 1980;
     Mendelbrot, 1982; Prigogine and Stengers, 1984; Holden,
     1986; Gleick, 1988; Glass and Mackey, 1988; Stewart,
     1989; Nicolis and Prigogine, 1989; Krasner, 1990;
     Schroeder, 1991; Hall, 1991; Lewin, 1992; Peitgen et
     al., 1992; Gluck, 1992; Waldrop, 1992; Kaye, 1993; and
     Kellert, 1993.

     AN OVERVIEW

     Chaos theory was popularized by Gleick (1988) in CHAOS: THE
BIRTH OF A NEW SCIENCE.  However, as Loye and Eisler (1987)
pointed out, chaos theory has a long if not well-recognized
heritage.  They stated:

     ..."chaos" theory is not a new or alien notion to
     social science.  ....  In modern times, "chaos" or
     "transformational" theoretical questions lie at the
     core of Hegel's philosophy of history as well as the
     multifaceted dialectical theory of Marx and Engels that
     initiated the formative dialogue for the development of
     modern sociology (Hughes, 1958).  Moreover, Pregogine
     himself cites Auguste Comte, Emile Durkheim,, and
     Herbert Spencer as the forerunners of his concept of
     dissipative structures referring particularly to
     Durkheim's concept of moral density as a precondition
     of the division of labor (Boulding, 1978).

     Defining "complexity" and "chaos theory" can help establish
their orientations and applications.  Thus, Waldrop (1992) points
to four aspects of systems which characterize complexity.  First,
"a great many independent agents are interacting with each other
in a great many ways" (p. 11).  Accordingly, complex systems
function as do "chemically reacting proteins, lipids, and nuclei
acids that make up a living cell, or the billions of
interconnected neurons that make up a brain, or the millions of
mutually interdependent individuals who make up a human society"
(p. 11).
     Second, systemic interactions can lead the system to
spontaneous self-organization.  The familiar "hidden hand" or
"ghost in the machine" are ways of expressing spontaneous self-
organization.  In fact, the concept of a free market or even
grass-roots associations represents some manifestations of
spontaneous self-organization.
     The third element of complexity comes from learning through
feedback--adaptability.  Complex systems "don't just passively
respond to events " (p. 11).  In terms applicable to the present
research, "species evolve for better survival in a changing
environment--and so do corporations and industries" (p. 11).
     The fourth and last aspect of complexity is its distinction
from "complicated" and "unpredictable".  Accompanying this is
complexity's distinctively dynamic aspect.  Waldrop explained
best when he wrote:

     every one of these complex, self-organizing, adaptive
     systems possess a kind of dynamism that makes them
     qualitatively different from static objects such as
     computer chips and snowflakes, which are merely
     complicated.  Complex systems are more spontaneous,
     more disorderly, more alive than that.  At the same
     time, however, their peculiar dynamism is also a far
     cry from the weirdly unpredictable gyrations known as
     chaos. (pp. 11-12)

     The definition of "chaos theory" has been provided by
Kellert (1993).  He stated that "chaos theory is THE QUALITATIVE
STUDY OF UNSTABLE APERIODIC BEHAVIOR IN DETERMINISTIC NONLINEAR
DYNAMICAL SYSTEMS" (p. 2; emphasis in original).  As is evident
in the two definitions, some lines separating complexity from
chaos theory are thicker than others.  Following Waldrop's
definition, complexity pertains to a system with specific
characteristics.  Chaos theory, by contrast, describes "BEHAVIOR
in...systems" (emphasis changed).  Thus, whereas it can be said
that chaos theory may describe some kinds of complex systems, it
is also occasionally hard to distinguish whether a specific
inquiry falls more on one side or the other side of the line
separating complexity and chaos theory.

     APPLICATIONS TO SOCIAL SCIENCE INQUIRY

     PRIMARY SOURCES: Metcalfe, 1974; Jantsch, 1975, 1980,
     1981; Gemmill and Smith, 1985; Loye and Eisler, 1987;
     Savit, 1988; Leifer, 1989; Reisch, 1991; McCloskey,
     1991; Hobbs, 1993; and Gregersen and Sailer, 1993.

     The social sciences can also benefit from complexity and
chaos theory analyses.  Economics in particular has come under
the new scrutiny of chaos theory (Arthur, 1989, 1990).  Other
social science phenomena, such as the international monetary
exchange rates (Eldridge, 1994), organizational change (Germmill
and Smith, 1985; Leifer, 1989), and population size projections
(May, 1974; Li, 1975) have proven ripe for reappraisal using
complexity and chaos theory.  More fields within the social
sciences--international relations and international law in
particular--are bound to follow.
     In exchange for further attention by social scientists,
chaos theory can make a contribution to the field.  Social
science literature is already dotted with pioneering
applications.  Through these efforts, a clearer vision of the
road ahead is emerging.  For instance, Reisch (1991) and
McCloskey (1991) independently applied chaos theory to the study
of history.  Reisch considered the future study of history.  He
favored a chaotic framework.  He explained:

     ...what kind of science should history become?  For in
     recent years there has been a bifurcation of sorts in
     the natural sciences, namely, the emergence of chaos
     theory--the science of physical systems governed by
     nonlinear dynamical laws.  ....I will argue that any
     science of history should fall into this new branch of
     physical theory.  For history, I will show, is chaotic. 
     And, I will demonstrate, it is characteristic of events
     in chaotic systems that they just cannot be explained
     with covering laws and initial conditions as Hempel
     [1965] believed they could. (p. 2)

     Part of the chaotic, nonlinear dynamics of history can be
captured in the linkage of "for want of a nail....the kingdom is
lost".  Thus Reisch puts this cascading feedback into a
historical context.  He wrote:

     Consider for example what the world might look like
     today if a person ONLY SLIGHTLY more combative than
     Khrushchev had been at the Soviet helm during the Cuban
     missile crisis--certainly VERY DIFFERENT if nuclear war
     had ensured.  Pascal probably performed a parallel
     thought experiment when he mused that the course of
     history was shaped by Cleopatra's nose.  And as for the
     history of science we can easily image "Darwin" not
     being an (almost) household name if circumstances had
     not led to his famous voyages [sic.]. (pp. 6-7;
     emphasis added; see also Squire et al. [1931])

In fact, Charles Darwin took only a single voyage.  However, to
show how tenuous events sometimes are, Darwin was not Captain
FitzRoy's first choice.  Darwin's father originally denied his
son permission to go and only a chance conversation with Charles'
uncle brought about his father's change of mind.  FitzRoy, a
scholar of phrenology, almost rejected Darwin because his nose
suggested a lack of commitment.  The start of the voyage was
delayed repeatedly and on two occasions the BEAGLE had to return
to port after starting out.  Last, Darwin's participation in the
voyage was justified as a means for him to prepare for a later
career in the clergy, not as a natural historian (Desmond and
Moore, 1992)!
     Of particular importance to this discussion on subnational
policy-adjustment behavior is Reisch's assessment of history as a
function of social and economic structures.  His discussion of
constraint and determination is particularly relevant for the
discussion on deterministic models of state behavior.  Reisch
concluded:

     ...while there is an appealing simplicity and elegance
     to this notion that certain social and economic
     structures will hold sway over the details of
     historical situations, there are, I think, some
     falsehoods buried under some of its truth.  There
     should be little doubt, for instance, that the course
     of historical events is constrained by social,
     economic, technological, and other factors.  But the
     mistake structuralism makes in rejecting L'HISTOIRE
     VNEMENTIELLE is to confuse CONSTRAINT with
     DETERMINATION.  As Braudel [1980] suggests, geographic,
     economic, social, ideological, and other sorts of
     structures surely limit historical and future
     possibilities for any given people.  But within those
     "envelopes" of possibility all kinds of things happen.
     (p. 8; emphasis in original)

     Finally, history is a science which involves time--lots of
time--making initial conditions critical.  For this reason Reisch
rejected a covering-law approach to history in favor of chaotic
analyses.  He elaborated:

     ...provided the laws which govern a chaotic system are
     known, the greater the temporal distance between
     initial (and intervening) conditions on the one hand
     and the event to be explained on the other, the greater
     the accuracy with which those conditions must be known. 
     This predicament rests on the nonlinearity of the laws
     governing chaotic systems.  But even when we do not
     know what the laws that govern the system might be, if
     that system exhibits extreme sensitivity to initial or
     prior conditions--as does history, I suggest--then we
     can be certain that those laws are nonlinear. 
     Consequently, the temporal scale of the covering-law
     explanations we might wish to make must be diminished
     given the extreme accuracy with which initial and
     intervening conditions must be known. (p. 17)

     McCloskey (1991) also analyzed the role of chaos theory in
history, with particular reference to the U. S. Civil War.  He
noted that small events have major implications as shown in
Lincoln's election and the succession of the South.  These events
being far from inevitable at that point in history, McCloskey
followed Fogel (1989) when he recalled that "like many historians
before him he emphasizes the PRECARIOUS BALANCE of American
politics in the 1850s, which could have been turned one way or
the other by minor events" (p. 26; emphasis added).  The term
"precarious balance" in chaos theory would be rendered as "at the
edge of instability" or "approaching a bifurcation point".
     History reveals that small events have large consequences. 
McCloskey exemplified this point:

     The point is not that great oaks from little acorns
     grow.  They do....  The point here is rather that in
     some modeled worlds an acorn produces by itself a great
     tree in an instant.  Such a world is unstable, as in
     the world of the United States in 1856-1965.  The
     models need not be complicated.  As students of chaos
     theory since Poincar have pointed out, simple models
     can generate astonishingly complicated patterns.  The
     slightest perturbation can yield an entirely different
     history.  (And in catastrophe theories, quickly.) 
     Confederate states depended on recognition by Great
     Britain, which depended on...Confederate success. (p.
     28)

     Finally, amplified, nonlinear feedback may account for
behavior which otherwise seems senseless.  McCloskey explained:

     Chaotic motion is to be distinguished from randomness. 
     Big randomness in models of the economy leads to
     fatalism.  Chaos--which is to say, very strong effects
     generated wholly within the model, but giving random-
     looking results--can lead to activism.  The president
     hoping that his jawboning will end a depression is a
     nonlinear dynamist: he thinks that little actions of
     his own can overwhelm the natural randomness.  (pp. 32-
     33)

     History is not the only social science to receive a second
look through complexity and chaos theory.  They have also been
directly related to issues in political science.  In their
research entitled "Chaos and Transformation: Implications of
Nonequilibrium Theory for Social Science and Society", Loye and
Eisler (1987) use chaos theory to examine

     (1) the realities of social breakdown and potential
     chaos in this late 20th-century world, which present
     the need for new action-oriented social theory; (2) the
     "breakthrough" nature of "chaos" theory in natural
     science; (3) the originating version of social science
     as a tool for social problem solving; (4) the gap
     between social science's formative hopes and
     contemporary performance; (5) the roots in modern
     social science of a social equivalent to natural
     scientific "chaos" theory; and (6) examples of current
     "chaos"/"transformational" theoretical works and works
     in progress. (p. 53)

     Loye and Eisler's objectives are far-reaching. 
Acknowledging "normative requirements" (p. 57), they offer four
benefits which an application to chaos theory would make in
social science.  Thus, they identify:

     1.   benefits of improved forecasting;

     2.   benefits of improved interventional guides;

     3.   benefits of participatory rather than
          authoritarian problem solutions; and

     4.   benefits of providing a clearer sense of
          system goal states or prohuman images of the
          future. (p. 57)

     In a different vein, Gregersen and Sailer (1993) used chaos
theory to argue "that the customary social science goals of
'prediction' and 'control' of systems behavior are sometimes, if
not usually, unobtainable" (p. 777).  Their article, entitled
"Chaos Theory and Its Implications for Social Science Research",
ends by enumerating "six familiar claims [=implications] about
the study of social phenomena for which chaos theory provides new
theoretical arguments".  Their six implications were:

     Implication 1. So long as social sciences continue to
     rely on cross-sectional studies, it is unlikely that
     they will discover and model the chaotic nature of
     social systems.

     Implication 2. Poor analytical results (e.g., low R2
     values and lack of statistical significance) are to be
     expected when analyzing chaotic systems with standard
     statistical methods.

     Implication 3. Use simulation techniques to study
     chaotic or complex social systems, but do not expect to
     be able to mimic any specific actual systems.

     Implication 4. Statistical technology will remain
     useful, but will play a different role in the analysis
     of chaotic systems.

     Implication 5. Qualitative methods will increase in
     importance when studying potentially chaotic social
     systems.

     Implication 6. Social science must develop a definition
     of "understanding" when analyzing chaotic systems. (pp
     793-798, passim)

These six implications are important for how they contrast
current preferences in social science inquiry, public policy
included.  Implications One and Two, in particular, are relevant
for research done on subnational policy-adjustment behavior as
reviewed earlier in this chapter.
     Finally, Savit's (1988) treatment of chaos in stock market
prices made two very important points for social science inquiry.
First, concerning statistical methods and inquiry into the
phenomenon, he wrote:

     it should be clear that there is a great deal of
     important information in the details of a chaotic
     sequence.  This information will sometimes be critical
     for discerning the hidden order in the sequence, and
     for understanding the nature of the sequence and using
     it advantageously.  Many of the common techniques of
     statistical analysis which were developed to deal with 
     "noisy" data involve various methods of smoothing the
     data.  Many such methods are not generally appropriate
     to the study of chaotic systems since too much useful
     information may be lost in the process of rendering the
     data smooth.  Indeed, in some sense, the hallmark of
     chaos is precisely its jumpiness. (p. 289)

     Data which do not easily settle into linear models may be
unfamiliar and unfriendly to researchers in the social sciences
who are accustomed to assumptions of linearity and hence
predictability.  However, an even greater challenge to these
assumptions was suggested by Hobbs (1993).  Hobbs pointed out the
need for ex post facto explanations.  He stated that "if chaotic
systems typically give rise to ex post facto explanations, then
the ubiquity of chaotic systems in nature argues for the ubiquity
of ex post facto explanations as an ordinary part of scientific
practice" (p. 122).  Thus, at the end of his analysis, he
concluded:

     All of the above ex post facto explanations are
     nonproductive because they are based on theories whose
     initial conditions are subject to some sort of
     indeterminacy or time lag.  Once it is known that a
     system is chaotic, virtually everything that happens to
     the system becomes causally relevant to its subsequent
     states, at least over periods of time less than the
     relaxation time, due to sensitivity to initial
     conditions.  (p. 136)

     Methods of investigating chaotic phenomena is only half of
Savit's contribution to the present research.  He also reminded
readers of a fundamental difference between using chaos theory to
study social science and the study of natural sciences.  Although
the tools are converging, in the case of the social system "the
study of the system disturbs the system itself" (p. 290).  He
continued, using examples from the stock market:

     Market prices in a nonlinear system can be very
     sensitive to small changes in the environment and it is
     not at all clear how much systems would respond to this
     additional kind of feedback.  This interference by an
     observer on the phenomenon observed, while not usually
     important in macroscopic physical systems, is
     reminiscent of analogous considerations in quantum
     mechanics which governs the microscopic physical world. 
     STUDYING THIS PROCESS IN A NONLINEAR MARKET IS CERTAIN
     TO YIELD REMARKABLE AND INTERESTING EFFECTS OF GREAT
     PRACTICAL SIGNIFICANCE. (p. 290; emphasis added)

     With the reivew of past research on policy-adjustment
behavior and complexity and chaos theory in place, it is now
possible to take the next step of this investigation--application
of physics noise and biology to inquiry on subnational behavior. 
The next chapter describes these models as well as generates
Hypotheses for empirical analyses.





                          Chapter Three
                                    
                      Models, Hypotheses, and Data


     CHAPTER ABSTRACT: As Chapter Two showed, past analyses
     have failed to build a convincing case using
     deterministic and equilibrium models.  Also, Chapter
     One announced the point of departure for the present
     research: a dissipative system identified by physics
     models and described by biology models.
          The present chapter introduces the physics and
     biology models which will be used to test hypotheses
     concerning data on the record of subnational policy-
     adjustment behavior.  First to be presented are the
     physics models.  Thus, the concept of "noise" is
     introduced and three types of noise--white, pink, and
     brown--are distinguished and described.
          Physics tells what type of system is observed;
     biology describes what the system's agents are doing. 
     Four contesting models--the Red Queen Hypothesis,
     Punctuated Equilibrium, Stationary Model, and Turnover-
     Pulse Model--can be tested to provide insights into the
     dynamics of the system.
          The data which will be used to test these models
     are identified and discussed.


DISSIPATIVE MODELS - PHYSICS AND EVOLUTION

     Six models are used to either identify or characterize
criticality in the subnational system.  The first step is to
determine if subnational policy-adjustment rates shows random
behavior or evidence of a deterministic system.  To achieve this
goal, the "noise" of the system is analyzed with reference to
determining its color.  The next step elaborates on its
predecessor by determining the dynamics of subnational
interaction within the system.  To achieve this second goal, four
models which characterize competition within a system are used.  
     Physics distinguishes three types of "noise", or points
which do not seem to fit expected values ("signal").  Thus,
"white noise" has a log-linear power spectrum slope of zero. 
(White noise is not discussed separately here since it is
subsumed under brown noise as described in Chapter One.)  Self-
organized criticality, also referred to as "pink noise", has a
log-linear power spectrum slope between zero and negative two. 
In contrast, "brown" noise (or "Brownian noise") has a log-linear
power spectrum slope of negative two.

     WHITE, PINK, AND BROWN NOISE  

     As discussed in Chapter One, white and brown noise are
signals of random movement and can be treated jointly.  In
contrast is pink noise which shows deterministic behavior.  The
following discussion describes the two possibities which open up
for further analysis.

     PRIMARY SOURCES: Petersen, 1977; Dutta and Horn, 1981;
     Lawrence et al., 1987; McHardy and Czerny, 1987; Kagan
     and Knopoff, 1987; Bak, Tang, and Wiesenfeld, 1987,
     1988; Tang and Bak, 1988a, 1988b; Carlson and Langer,
     1989; Jaeger et al., 1989; Bak and Tang, 1989; Bak,
     Chen and Creutz, 1989; Kananoff et al., 1989; Hwa and
     Kardar, 1989; Dhar, 1990; Babcock and Westervelt, 1990;
     Held et al., 1990; Bhattacharya and Waymore, 1990;
     Schroeder, 1991; Bak and Chen, 1991; Pietgen et al.,
     1992; and Ross, 1993.

     SELF-ORGANIZED CRITICALITY

     Beginning a discussion of a dissipative phenomenon by
referencing self-organized criticality (SOC) is appropriate.  Bak
et al. (1988), after discussing how dissipative dynamical systems
evolve into a critical state, showed their enthusiasm for this
topic when they wrote, "we believe that the concept of self-
organized criticality can be taken much further and might be THE
underlying concept for temporal and spacial scaling in
dissipative nonequilibrium systems" (p. 373; emphasis in
original).
     Indeed, SOC has a place in the discussion of subnational
behavior; Bak and Chen (1991:46) showed this opportunity.  They
wrote:

     We propose the theory of self-organized criticality:
     many composite systems naturally evolve to a critical
     state in which a minor event starts a chain reaction
     that can affect any number of elements in the system. 
     Although composite systems produce more minor events
     than catastrophes, chain reactions of all sizes are an
     integral part of the dynamics.  According to the
     theory, the mechanism that leads to minor events is the
     same one that leads to major events.  Furthermore,
     composite systems never reach equilibrium but instead
     evolve from one metastable state to the next.

It is the point of departure in studying physics noise that the
fifty states may form a "composite system" which behaves as
described above.
     To clarify SOC behavior, note that two familiar phenomena
exhibit SOC: earthquakes and avalanches (with the latter
represented by the behavior of a sandpile).  As early as 1956
Gutenberg and Richter recognized that earthquake frequencies
could be described using a power-law formula.  Additional work by
Held et al. (1990) has confirmed computer models of a sandpile to
show SOC.
     The interesting aspect of earthquakes is their self-
similarity.  That is to say, not only do a string of individual
earthquake events within a system over time show SOC, but each
earthquake event itself shows SOC.  In reference to this latter
feature, the seismographic analysis of the event "shows the
energy distribution at the stationary critical state.  The
distribution fluctuation indeed fits a power law" (Bak and Tang,
1989:15,636) expected of SOC.
     What is power law?  Consider the sandpile example: When
adding sand at a single grain at a time to a flat circular disk,
it is possible to monitor fluctuations in the weight of the
sandpile.  (A grain of sand weighs about 0.0006 gram.) 
Avalanches--or where sand slides off the side of the disk--are
therefore able to be measured precisely.  Data are collected over
many trials of adding sand until the pile approaches the edge of
instability.  An additional gain will cause a perturbation in the
system, creating an avalanche which is scale invariant (or
"characterized by an infinite correlation length, and so occur on
all length and time scales up to limits determined by the finite
size of the system" [Held et al., 1990:1,120]).  Repeated
perturbations show that avalanches are not predictable in either
individual sizes (weight) nor timing.  The experiment also shows
that a single grain of sand can cause avalanches ranging from
extremely small to (in theory) the system's finite limit.  
     An important conclusion holds that avalanches are
predictable in displaying a long-run pattern.  That is, although
it is impossible to predetermine WHEN a system will reach
instability and HOW MUCH the system will adjust itself before
achieving stability (except on a short-term basis, see Kagan and
Knopoff [1987]), it is possible to say that the overall profile
of the avalanches will display a log-linear relationship.  
     Log-linear relationships are easily described.  The X- and
Y-axes are expressed in logarithmic scale.  Using avalanches as
the example, the X-axis is defined as the size of each avalanche.
The Y-axis is defined as the number of avalanches with the same
X-axis value.  The resulting graph of the ordered pairs will
reveal a straight line.  Thus, Gutenberg and Richter (1956), when
working on earthquakes, found a straight line with a slope
between -1.25 and -1.50, depending on the fault under study. 
Also, the Richter scale uses logarithmic values to calibrate
earthquake power.
     The slope of the line is important in that it must be
between zero and negative two.  Thus, in addition to the graph of
SOC showing log-linearity, the slope   must be 0>-BETA>-2.
     Held et al. (1990) inserted one note of caution.  They found
that for larger sandpiles, the distribution of avalanches became
sharply peaked and the scale invariance broke down.  They
conclude that "the presence or absence of self-organized
criticality may therefore be related to a damping length scale
which we do not fully understand" (p. 1122).  Although their
comments were written in 1990, this issue has not been pursued in
the literature.  
     Also, since SOC phenomena exhibit a range of slope values
between zero and negative two, it may be the case that different
slopes indicate different classes of SOC (Tang and Bak, 1988*). 
However, this point has not been demonstrated.
     Based on the preceding discussion of 1/f noise, Hypothesis
One applies SOC to the behavior of subnational agents:

     Hypothesis One: Subnational policy-adjustment behavior
     exhibits features of self-organized criticality. 

     Test for Hypothesis One: Subnational policy-adjustment
     behavior conforms with expectations of power-law
     linearity showing a slope BETA such that 0>-BETA >-2.

Hypothesis One is tested in Chapter Four.

     BROWN NOISE

     Brownian noise (or Brownian motion) refers to a phenomenon
observed by the botanist Robert Brown around 1827.  Brown
observed the movement of molecules.  Movement is due to very
light collisions with other molecules.  As clarified by
Bhattacharya and Waymire (1990:17)

     Perhaps the simplest way to introduce the continuous-
     parameter stochastic process known as BROWNIAN MOTION
     is to view it as the limiting form of an unrestricted
     random walk.  To physically motivate the discussion,
     suppose a solute particle immersed in a liquid suffers,
     on the average, f collisions per second with the
     molecules of the surrounding liquid.  Assume that a
     collision causes a small random displacement of the
     solute particle that is independent of its present
     position.  For simplicity, consider displacements in
     one particular direction, say the vertical direction,
     and assume that each displacement is either +  or - 
     with probabilities p and q = 1-p, respectively.  The
     particle then performs a one-dimensional random walk
     with step size  . (emphasis in original)

This movement (defined above as either + DELTA or - DELTA), when
plotted along the same logarithmetic X- and Y-axes of SOC, shows
log-linearity with a slope of negative two.
     Brown noise can be created by summing independent random
numbers.  It also exhibits a Gaussian distribution with means at
zero.  The random movement of the agent means it has a
probability of 1.0 of returning to a point which it had earlier
occupied.
     Although Brownian movement is random, random does not mean
insignificant.  For instance, it has been applied to "such areas
as statistical testing of goodness of fit, analyzing the price
levels of the stock market, and quantum mechanics" (Ross,
1993:458).  Further, the presence of Brownian noise in diverse
phenomena may suggest a dynamic similar to molecular movement. 
Specifically, other phenomena exhibiting Brownian motion may be
influenced by actions from adjacent units.  For instance, if
states exhibit policy-adjustment rates which suggest Brownian
movement, the behavior may be due to "collisions" with behavior
in neighboring states.
     Brownian randomness can also create deterministic shapes. 
Thus, Peitgen et al. (1992:297-299) describe a process for
creating a Sierpinski gasket via the random process of throwing a
die.  They explain

     We have just seen the generation of the Sierpinski
     gasket by a RANDOM PROCESS, which is amazing because
     the Sierpinski gasket has become A PARAGON OF STRUCTURE
     AND ORDER for us.  In other words, we have seen how
     RANDOMNESS CAN CREATE A PERFECTLY DETERMINISTIC SHAPE. 
     To put it still  another way, if we follow the time
     process step by step, we cannot predict where the next
     game point will land because it is determined by a
     throwing die.  But nevertheless, the pattern which all
     game points together leave behind is absolutely
     predictable.  This demonstrates an interesting
     INTERPLAY BETWEEN RANDOMNESS AND DETERMINISTIC
     FRACTALS. (p. 299; emphasis added)

     Based on the preceding discussion, Hypothesis Two applies
Brownian noise to the behavior of subnational agents:

     Hypothesis Two: Subnational policy-adjustment behavior
     exhibits features of Brownian noise.  

     Test for Hypothesis Two: Subnational policy-adjustment
     behavior conforms with expectations of power-law
     linearity showing a slope -BETA  equal to -2. 

Hypothesis Two is tested in Chapter Four.

EVOLUTIONARY MODELS

     "It's a jungle out there", not only for species seeking to
survive to the next generation, but also for states seeking self-
maximization in an environment of limited economic development
resources.  Species and states compete for access to, and control
over, limited resources of value to themselves and other species
or states.  However, to say that these agents COMPETE is to leave
open the options of how competition takes place.  For purposes of
this research, "competition" is conceptualized as behavior
falling between the extremes of "combat" and "cooperation". 
Competition includes selfish acts of self-maximization leading to
a loss by other system agents.  Such competitive behavior is
close to combat.  However, competition may also produce behavior
which recognizes joint outcomes for system agents or helps other
system agents as a means of helping one's own self in the
process.  This behavioral outcome of competition is closer to
cooperation.  It is expected that subnational agents exhibit the
full range of competitive behavior with greater and lesser
degrees of combat and cooperation.  This full range has been
described by Waldrop (quoted above) as showing concurrent "mutual
accommodation and mutual rivalry".
     To advance the study of subnational behavior, four models of
competition, each drawn from evolutionary biology, provide a
means of comparing how agents compete within a system.  These
four models are:
     1.  Van Valen's RED QUEEN HYPOTHESIS;
     2.  Eldredge and Gould's PUNCTUATED EQUILIBRIUM;
     3.  Stenseth and Maynard Smith's STATIONARY VIEW; and
     4.  Vrba's TURNOVER-PULSE.
     Using biological models in the present research is governed
by their attention to the systemic level of analysis, a failure
already noted in deterministic models traditionally used for
analyzing subnational policy-adjustment behavior.  Unlike models
used in the studies cited in Chapter Two, evolutionary biology
models assess the joint influences of agents and the environment
as they act together within the system.  First, agents are
recognized in their capacity to create and react to stimuli, thus
generating systemic feedback which can either escalate or
diminish (both as non-linearities).  Second, environmental
perturbations are recognized in the biological models as
important stimuli for change.  In short, biological models of
competition come the closest in recognizing and operationalizing
a complex, dissipative system on the edge of chaos (Stanley,
1979, Conrad, 1983, Jantsch, 1986).
     At the outset of this discussion, it must be stressed that
the differences between the four biological models can sometimes
get lost in the similarities.  Indeed, each author is careful to
mark the distinguishing features and claim a unique fief. 
Nevertheless, Vrba (1993:442) showed how easily one model shifts
into the next.  She wrote:

     One might well ask whether the Red Queen could lead to
     punctuated equilibria.  Perhaps the metaphorical lady
     is not running constantly without getting anywhere, but
     instead is an occasional sprinter to new places where
     she rests at length before the next sprint (which would
     of course not be Lewis Carroll's Red Queen).  Or
     perhaps in between rare sprints she makes short runs
     hither and thither, in different directions in
     different populations of the same species without net
     directional gain.  The latter is what Eldredge
     (personal communication) argues, and the former
     resembles Stenseth and Maynard Smith's (1984)
     stationary outcome in local ecosystems, although not
     necessarily across whole species.... 

     The following discussion clarifies the contending
interpretations of competition and sets up a case for using these
models in the evaluation of states in a self-organized system. 
The goal is to find whether any of these four models capture
fundamental dynamics of the subnational system.  With this goal
realized, further analysis can evaluate the reasons for the
failure of some models and the success of others.

     PRIMARY SOURCES: Eldredge and Gould, 1972; Van Valen,
     1973, 1974; Foin et al., 1975; Van Valen, 1975;
     Stanley, 1975; Hallam, 1976; Van Valen, 1976; Gould and
     Eldredge, 1977; Stanley, 1979; Mccune, 1982; Vrba,
     1982; Hoffman and Kitchell, 1984; Stenseth and Maynard
     Smith, 1984; Vrba, 1985; Boyce, 1990; Vrba, 1993;
     Kauffman, 1993.

     THE RED QUEEN HYPOTHESIS

     Van Valen's controversial model of evolutionary biology has
clear applications to business and public policy.  His model--the
Red Queen Hypothesis (RQH)--explains how species evolve primarily
in reference to each other within the ecosystem.  Van Valen
explains the dynamics of species living in a system in the
following manner:

     The amount of resources is fixed and can be thought of
     as an incompressible gel neutrally stable in
     configuration, supporting the peaks and ridges.  If one
     peak is diminished there must be an equally total
     increase elsewhere, in one related peak or more
     uniformly.  Similarly, increase in a peak results in an
     equal decrease elsewhere.
          Species occupy this landscape and can be thought
     of as trying to maximize their share of whatever
     resource is scarcest relative to its use and
     availability.  This resource will take the role of the
     gel, and the MOMENTARY FITNESS of a species will be
     proportional to the amount of gel under its area (the
     amount of the limiting resource it controls).  To a
     sufficiently close approximation this momentary fitness
     seems to be what natural selection maximizes. (Val
     Valen, 1973:19; emphasis in original)

     In simplest terms, RQH describes a zero-sum situation where
one species' adaptive gain results in an equal loss distributed
to all other species in the system.  Thus, "the sum of absolute
fitness in a community is constant" (p. 577).  For instance, in
an ecosystem of fifty species, one species may evolve by three
units of advantage.  However, each of the 49 remaining species
will thereby be disadvantaged by 3/49 darwins such that the sum
of the disadvantage will be three.  The same mechanism applies
whether a SINGLE species evolves between t1 and t2 time, or
whether MULTIPLE species evolve in the time interval.  
     RQH is a biotic model of evolution; that is, it attributes
evolutionary stimuli on a species to the changes by other species
in the system.  Significantly, and in contrast to other models
which follow, RQH does not attribute evolutionary stimuli to
changes in the physical environment.  "The Red Queen does not
need changes in the physical environment, although she can
accommodate them" (p. 19).  Other models, Vrba's Turnover-Pulses
in particular, see it differently (and depend on environmental
changes to the exclusion of biotic stimuli).
     RQH also holds that evolution takes place as an incremental
series of steps, without pauses and variations in the pace.  RQH
differs from other evolutionary models, in particular punctuated
equilibrium, in seeing change as continuous.  However, as
indicated by the name, RQH also believes that the overall outcome
of all species' change is a species' constant-level momentary
fitness.
     Based on the preceding discussion, Hypothesis Three applies
Van Valen's Red Queen Hypothesis to the behavior of subnational
agents:

     Hypothesis Three: Subnational policy-adjustment
     behavior displays the coevolutionary characteristics of
     Van Valen's Red Queen Hypothesis.

     Test for Hypothesis Three: Subnational policy-
     adjustment behavior displays evidence of each state's
     momentary fitness which fluctuates in the short-term in
     reference to all other systemic agents but shows no net
     advancement over the long-term relative to other
     subnational agents.

Hypothesis Three is tested in Chapter Five.

     PUNCTUATED EQUILIBRIUM

     Unlike the RQH, punctuated equilibrium (PE) sees
evolutionary change, in particular speciation, as a periodic and
major event.  PE sees changes in species as periodic in that
evolution does not proceed incrementally (i.e., the accumulation
of small steps over time).  Rather, PE sees changes occurring in
leaps, contrary to the Linnean maxim "natura non facit saltus"
("Nature doesn't make jumps").  Not only is it periodic, but
evolution is also a major event according to PE.  Changes
constitute morphological discontinuities to the previously-
existing conditions.  Thus, to cite an example used by Eldredge
and Gould, brain size in Homo sapiens did not increase
incrementally through the ancestral stock.  Instead, increased
brain size came about as a major change at a single moment in the
geologic/biologic time scale.  According to Eldredge and Gould,
the fossil evidence supports this view by the absence of
intermediate forms (species stasis).
     Support coming from the absence of intermediate forms, as
stated immediately above, is an essential difference between PE
and other models of evolution.  Eldredge and Gould point to the
absence of intermediate forms as an accurate reflection of fossil
ecosystems.  This interpretation differs from earlier
interpretations (including Darwin's) which saw the lack of
intermediate forms as an inevitable bias in the fragmentary
fossil record.  According to Eldredge and Gould, STASIS (OR LACK
OF EVIDENCED CHANGE) IS DATA.  In fact, PE predicts stasis and
lack of change as an important aspect of what happens in
evolution.
     Inherent in the PE model of evolution, and a colorrary of
stasis and rapid evolutionary activity, is the rejection of
evolution which occurs by the accumulation of gradual, random
steps.  Rather, PE holds that evoltuion occurs as a non-random
(directional) and non-incremental (sudden) process.
     Based on the preceding discussion, Hypothesis Four applies
Eldredge and Gould's Punctuated Equilibrium model to the behavior
of subnational agents:

     Hypothesis Four: Subnational policy-adjustment behavior
     displays features of Eldredge and Gould's model of
     punctuated equilibrium.

     Test for Hypothesis Four: Subnational policy-adjustment
     behavior exhibits stasis and abrupt evolutionary
     change, both at the state level.  Subnational policy-
     adjustment behavior is not expected to exhibit
     incremental changes or meandering adaptation
     strategies.

Hypothesis Four is tested in Chapter Five.

     STATIONARY VIEW

     Stenseth and Maynard Smith (1984) have proposed the third
model of species competition which can be used for understanding
subnational policy-adjustment behavior.  Their point of departure
is based on the interaction of populations of species in a
community. 
     In their analysis, Stenseth and Maynard Smith held that the
only possible models for evolutionary competition are RQH and SV
(thus rejecting PE and Turnover-Pulses, discussed below).  They
rejected the suggestion that their conceptualization is
punctuated equilibrium in disguise.  They summarized and
distinguished the major differences:

     The Red Queen equilibrium is dominated by biotic
     interactions; that is, THE MAIN FEATURE OF THE
     ENVIRONMENT OF EACH SPECIES CONSISTS OF THE OTHER
     SPECIES IN THE COMMUNITY.  In contrast, if the
     Stationary picture is correct, EVOLUTION IS DRIVEN BY
     PHYSICAL CHANGES.  It is tempting to suggest that the
     two pictures correspond respectively to a gradualist
     and to a stasis plus punctuation interpretation of the
     fossil record (Eldredge and Gould, 1972; Gould and
     Eldredge, 1977; Stanley, 1979).  We cannot support this
     interpretation strongly. (p. 877; emphasis added)

Thus, species respond to other species' evolution in RQH.  In
contrast, for SV, species respond to changes in the physical
environment.  
     A further difference between RQH and SV is the latter's
rejection of the zero-sum outcome of evolution.  RQH is based on
one species' evolutionary gain creating an equal amount of loss
in other species.  Fitness advantage is momentary and no species
pulls ahead.  In contrast, SV holds that the objective for each
species is to reach its specific evolutionary maximum.  The
researchers introduced the idea of "evolutionary lag" to define
the distance a species is from its maximum fitness.  Evolutionary
lag therefore defines how fit a species is in absolute terms, not
in relative terms as done in RQH.
     Stenseth and Maynard Smith also reference the rate of
evolution.  Differences in evolutionary stimuli translate to
expected differences in the rate of evolution.  Rate in RQH, as
already explained, is incremental and results in fluctuation of
the species' fitness levels.  As for SV, rate of evolution shows
bursts of activity which is associated with "major changes in the
physical environment" (p. 877).  Rate of evolution would then
slow until the physical environment is again disturbed.  
     Based on the preceding discussion, Hypothesis Five applies
Stenseth and Maynard Smith's Stationary View model to the
behavior of subnational agents:

     Hypothesis Five: Subnational policy-adjustment behavior
     displays features of  Stenseth and Maynard Smith's
     Stationary View model. 

     Test for Hypothesis Five: Subnational policy-adjustment
     behavior shows evidence of bursts of evolutionary
     activity associated with perturbations to the physical
     environment followed by a slowing down of evolutionary
     change and eventual stasis until the system is again
     perturbed. 

Hypothesis Five is tested in Chapter Five.

     TURNOVER-PULSES

     Vrba's Turnover-Pulses (T-P) model provides the fourth and
final variation on the relation between species and the
environment, the system, and other species.  Essential in T-P is
the role of the environment.  Vrba explained the relationship
between changes in species and the environment.  She wrote:

     Evolution is normally conservative at least in relation
     to speciation and extinction.  Speciation does not
     occur unless forced by changes in the physical
     environment.  Similarly, forcing by the physical
     environment is required to produce extinctions and most
     migration events.  Thus, MOST LINEAGE TURNOVER IN THE
     HISTORY OF LIFE HAS OCCURRED IN PULSES, NEARLY
     SYNCHRONOUS ACROSS DIVERSE GROUPS OF ORGANISMS, AND IN
     PREDICTABLE SYNCHRONY WITH CHANGES IN THE PHYSICAL
     ENVIRONMENT. (p. 428; emphasis added)

Thus, unlike RQH which was a biotic model, T-P is a physical
environmental model.
     For T-P, the ecosystem is of secondary importance (unlike
RQH).  Vrba wrote "the habitat requirements basically depend on
physical environmental variables and not on the particular set of
biotic characteristics in any one ecosystem" (p. 430).  Thus, T-P
is robust enough to occur in any variety of physical
environments.  
     According to T-P, all species in the ecosystem evolve in
tandem, creating the pulse.  To clarify the process and
implications, Vrba wrote:

     The model does emphasize that, when lineages undergo
     turnover at all, they do so in concert with others.  To
     use a metaphor, Turnover Pulse dictates to lineages:
     WHETHER you respond or not is none of my business, but
     IF you do then you must do it together.  This
     environmental change has a deterministic role in that
     turnover would not occur without it.  But other,
     internal factors contribute to the PROBABILITY of
     speciations and together with local biotic interactions
     must provide strong proximal causes of the NATURE of
     speciation change, and geographic factors determine the
     spatial distribution of turnover events during a pulse.
     (pp. 431-432; emphasis in original)

The discussion will return to elaborate on Vrba's references to
the "deterministic role" and "internal factors contribute to the
PROBABILITY of speciations" (see Chapter Six).  In the meantime,
note that whereas the physical environment perturbs the system,
the biotic environment influences how the system will evolve
thereafter.
     Based on the preceding discussion, Hypothesis Six applies
Vrba's Turnover-Pulse model to the behavior of subnational
agents:

     Hypothesis Six: Subnational policy-adjustment behavior
     displays features of Vrba's turnover-pulse model.

     Test for Hypothesis Six: Subnational policy-adjustment
     behavior shows evidence of rapid evolutionary movement
     associated with initiating events originating from
     environmental perturbations.  The system is otherwise
     conservative and cannot initiate change without an
     environmental stimulus.

Hypothesis Six is tested in Chapter Five.

DATA

     Data on subnational policy-adjustment behavior are available
from the Conway publication INDUSTRIAL DEVELOPMENT.  This
publication was targeted to professionals involved in industrial
site-location decisions.  Realizing that the practitioner needed
a clearer idea of differences between state policy packages, in
1966 INDUSTRIAL DEVELOPMENT started a systematic listing of
economic development incentives and the states which offer them. 
The list was updated annually with some options dropped from the
original listing and others added.  Other research has utilized
the Conway data, notably Grady (1987), Ambrosius (1989), and
Brace (1991).
     The Conway list can be used for empirical research after
correcting some inconsistencies in its presentation.  In order to
assure accuracy in findings, a panel of clean data was prepared
using only incentives which were carried from 1970 to 1986.  In
all, 46 incentives can be used (see Table 1).  By using a
consistent list over the years it is possible to accurately
reflect state changes in policy adjustment.  For 17 years, 16
yearly changes are  given.  Thus, 800 data observations are used
to test the hypotheses.

                                Table One
                            List of Policies

FINANCIAL ASSISTANCE FOR INDUSTRY (16 policies)

     State sponsored industrial development authority

     Privately sponsored development credit corporation

     State authority or agency revenue bond financing

     State authority or agency general obligation bond
     financing

     City and/or county general obligation bond financing

     State loans for building construction

     State loans for equipment machinery

     City and/or county loans for building construction

     City and/or county loans for equipment and machinery

     State loan guarantees for building construction

     State loan guarantees for equipment and machinery

     City and/or county loan guarantees for equipment and
     machinery

     State financing aid for existing plant expansions

     State matching funds for city and/or county industrial
     financing programs

     State incentive for establishing industrial plants in
     areas of high unemployment

     City and/or county incentive for establishing
     industrial plants in areas of high unemployment


SPECIAL SERVICES FOR INDUSTRIAL DEVELOPMENT (12 policies)

     State financed speculative building

     City and/or county financed speculative building

     State provides free land for industry

     State-owned industrial park sites

     City and/or county-owned industrial park sites

     State funds for city and/or county development-related
     public works projects

     State funds for city and/or county master plans

     State funds for city and/or county recreational
     projects

     State funds for private recreational projects

     State program to promote research and development

     State supported training of "hard core" unemployed

     State science and/or technology advisory council


TAX INCENTIVES FOR INDUSTRY, OTHER LAWS (18 policies)

     Corporate income tax exemption

     Personal income tax exemption

     Excise tax exemption

     Tax exemption or moratorium on land and capital
     improvements

     Tax exemption or moratorium on equipment and machinery

     Inventory tax exemption on goods in transit (freeport)

     Tax exemption on manufacturers' inventories

     Sales/use tax exemption on new equipment

     Tax exemption on raw materials used in manufacturing

     Tax credits for use of specified state products

     Tax stabilization agreements for specified industries

     Tax exemption to encourage research and development

     Accelerated depreciation of industrial equipment

     State right-to-work law

     State minimum wage law

     State fair employment practice code

     Statewide uniform property tax evaluation law

     Statewide industrial noise abatement law




          
                             Chapter Four
                                    
                         The Physics of "Noise" 
                                    
                                    
     CHAPTER ABSTRACT: Following what has been said in Chapter
Three, the opportunity exists to describe state policy-adjustment
rates using quantitative measures.  Chapters Four and Five
explore this opportunity, here with physics noise and later with
biology.  As for now, two options--self-organized criticality
(SOC) and Brownian motion--can be empirically tested.  The
distinguishing factor between them is the slope of the log-linear
line.  
          In this chapter, Hypothesis One (for SOC) and
Hypothesis Two (for Brownian motion) are tested.  Findings
suggest confirming Hypothesis One.  This implies that states
behave in a manner which conforms to expectations of
self-organized criticality. 
     The implications of this finding are discussed.
          
          A QUESTION OF INFLUENCE

               This chapter poses a fundamental question: how
          greatly are states influenced by other states during
          the policy-adjustment process for economic development? 
      As suggested by the latter half of the question, there
          is no assumption that states are influenced in the same
          manner for all policy domains (contra Walker, 1969,
          1973).  Indeed, policy domains vary greatly--taxes,
          education, transportation, welfare--and characteristics
          of each domain are expected to determine the nature and
          extent of inter-state influences.  Thus, answers to the
          question about state influences concerning economic
          development policies cannot be applied to other policy
          domains without appropriate caution.

          HYPOTHESES AND TESTING

               Chapter Three introduced self-organized
          criticality (SOC) as evidenced in earthquakes and
          sandpiles.  "By a change of language (and a little bit
          of imagination), one can transform the sandpile or the
          earthquake model into many situations."  The words are
          from Bak and Chen (1991:52) and point to the
          possibility of using SOC to analyze U. S. subnational
          behavior (see also Levin, 1992:61).  Just as a system
          of tectonic plates evolves, so states adopt policies
          over time.  Every year roughly half the states adjust
          their economic development policy offerings.  Adjusting
          policies assures the state neither looses revenues by
          needlessly offering too many incentives nor becomes
          uncompetitive by offering too few incentives to achieve
          its goal. 
               To show the potential relationship between SOC and
          policy-adoption behavior, recall Bak and Tang's
          (1989:15,635) remarks:

               It is essential that the systems are
          dissipative (energy is released) and that
          they are spatially extended with an
          "infinity" of degrees of freedom. ENERGY IS
          FED INTO THE SYSTEM IN A UNIFORM WAY, either
          directly into the bulk or through the
          boundaries.  The crust of the earth,
          subjected to the pressure from tectonic plate
          motion, may be viewed as a system of this
          kind. (emphasis added) 

               Elaborating on the phrase "energy is fed into the
          system", individual subnational agents, acting in their
          own self-interest, create energy through policy
          adoption and this energy (pressure) accumulates against
          other states.  Eventually--as more and more states
          adjust their policies--states both continuously respond
          to the earlier actions by other states while they
          continuously create future pressure on those same
          states.  The system never reaches stasis nor returns to
          equilibrium; the individual states produce feedback and
          internal perturbations at the system level.  The
          overall result is a legislative year for each of the
          fifty states showing how states adjust their policy
          package to varying degrees.  When the tally of all
          states in a given year is assembled, it can be analyzed
          empirically.
               Have states evolved spontaneous organization as
          suggested in SOC?  Or are states independent agents
          acting randomly as modeled by Brownian motion? 
          Hypotheses One and Two ask these questions in a
          different form.  To begin testing the Hypotheses, the
          criteria for empirical results must be stated.  Thus, 

               Test for Hypothesis One: Subnational policy-
               adjustment behavior conforms with
               expectations of power-law linearity showing a
               slope such that 0>-BETA>-2.
          
               Test for Hypothesis Two: Subnational policy-
               adjustment behavior conforms with
               expectations of power-law linearity
               showing a slope -BETA equal to -2. 
          
          DATA AND ANALYSIS     
   
               To test these Hypotheses, the Conway data are
          tallied on a yearly basis for each state.  The number
          of policies offered in each state is compared to the
          previous year's policies for that state.  Over the
          years studied, annual changes in the number of policies
          offered ranged from one to eleven.  Thus the values one
          through eleven become the X-values in the ordered
          pairs.  The Y-value is determined by how many times
          each X-value appeared in the tally of annual changes.  
               With the data prepared as identified, the
          distribution of yearly changes, and their frequency,
          are noted in Table Two.  
               A graph of these yearly data (not presented) shows
          a familiar Poisson curve.  Indeed, if based solely on
          this observation, the data might be said to be random
          and in support of Hypothesis Two.  However, such a
          conclusion would be hasty.
          
          
                           Table Two
                               
                   Tallies of Annual Policy
                    Adjustments, 1970-1986 
                                
          
                Year    0   1  2  3  4  5  6  7  8  9 10 11
          
               70-71   25  16  3  1  1  2  2
               71-72   30  10  5  3  2 
               72-73   22  15  5  2  3  3 
               73-74   14  23  7  2  1  2  0  0  1
               74-75   19  14  8  6  1  1  0  0  1 
          
               75-76   24  14  4  1  5  1  0  1
               76-77   26   9 10  2  1  1  1
               77-78   27  11  4  5  2  1
               78-79   24  11  5  4  1  2  1  1  1
               79-80   36   7  3  3  0  0  0  0  1
          
               80-81   20  17  9  1  0  2  0  0  0  1
               81-82   21  17  4  4  1  2  0  0  0  0  0  1
               82-83     *
               83-84     *
               84-85   23   9  7  3  6  1  0  0  0  0  0  1
          
               85-86   27  11  6  2  3  0  1 
          
          
               
               Testing the data must proceed further as suggested
          in observations from Glass and Mackey (1988) concerning
          the apparent similarities between graphs of random
          noise and graphs of chaos.  In their Chapter Three
          ("Noise and Chaos"), the authors point out that the
          Poisson process is the simplest model for a random
          walk.  However, random data have long been confused
          with chaotic data.  Adding to he confusion is the fact
          that "random numbers" are generated from a
          deterministic logistic equation.  To stress the
          difference, however, is to insist on the fact that
          although random data and chaotic data can appear
          similar, chaotic data are the result of deterministic
          output.  This feature of chaotic data--earlier pointed
          out in Chapter Two--should be the deciding factor
          separating random and chaotic data.
               The apparent similarities between random and
          chaotic data have implications for testing between the
          two.  In examining the processes of identifying random
          data from chaotic data, Glass and Mackey leave no doubt
          that "OBSERVATION OF EXPONENTIAL PROBABILITY DENSITIES
          IS NOT SUFFICIENT TO IDENTIFY A PROCESS AS A POISSON
          PROCESS" (emphasis in original).  Some other test is
          needed which can successfully distinguish random data
          from chaotic data.  As it works out, the slope of a
          log-linear power spectrum line distinguishes between
          random and chaotic data.
               To test Hypotheses One and Two using log-linear
          power spectrum slopes, the annual policy-adjustment
          record for all states from 1970 to 1986 is regressed
          with logarithmic values.  The slope, R2, and
          Y-intercept data are shown in Table Three.
               The yearly data showing annual changes in all
          state's economic development policy packages conforms
          well to expectations for self-organized criticality
          (Hypothesis One).  In each case the slope is within the
          range of zero to negative two (with a minimum slope of
          -0.90 and maximum of -1.60).       
               In addition to slope values which cluster well
          within the anticipated range, the R2 values range from
          very high to middle confidence.  The average R2 value
          is *0.79.
               How much confidence should these findings be given
          in terms of Hypotheses One and Two?  Certainly the high
          R2 value suggests one source of confidence.  However,
          results may be strengthened, refined, or questioned by
          reference to other tests.
          
          
                         Table Three
                               
                     Power Law Regression
                           1970-1986
                                
          
               Year      Slope      R2    Y-intercept
          
               70-71     -1.20     .60       0.95
               71-72     -1.15     .99       1.01
               72-73     -1.07     .76       1.07
               73-74     -1.56     .86       1.24 
               74-75     -1.51     .82       1.22
          
               75-76     -1.28     .65       1.03
               76-77     -1.51     .82       1.09
               77-78     -1.32     .84       1.07
               78-79     -1.24     .87       1.04
               79-80     -0.90     .96       0.82
          
               80-81     -1.60     .69       1.21
               81-82     -1.50     .82       1.17
               82-83     *
               83-84     *
               84-85     -1.04     .56       1.04
          
               85-86     -1.28     .89       1.07
          
          
          
               The literature does not identify an alternative
          test for self-organized criticality.  However, random
          behavior has several tests.  Thus, although Hypothesis
          One cannot further be tested, Hypothesis Two can be
          further explored.
               First, brown noise should show a Gaussian
          distribution with a mean value of zero.  These
          conditions are not observed (see Table Four); the
          distribution shows a 14% occurrence of downward
          adjustments (from -9 to -1) and a 38% occurrence of
          upward adjustments (from 1 to 11).  Further, the values
          in Table Four translate to a loss of 166 policies and a
          gain of 600 policies, or a 3.61-fold gain over losses. 
          These findings show a clear lack of symmetry in the
          distribution; rather, the distribution is skewed in a
          specific direction and this thus inconsistent with a
          random walk and Brownian motion.  Given these
          observations, it is unlikely that a state would return
          to a previously-existing location in its policy
          package.  By observation, Brownian motion and
          Hypotheses Two are not supported.

          DISCUSSION ON PHYSICS NOISE

               What influences operate on states considering
          policy adjustment for economic development objectives? 
          The data suggest a process consistent with self-
          organized criticality.  Further, results of tests
          presented in this chapter have closed out one important
          option for interpreting state behavior.  The case
          against Brownian motion (random movement) suggests that
          some factors are indeed serving to guide behavior. 
          Policy makers, like God, do not throw dice (Stewart,
          1989); deterministic rules are used when adjusting a
          state's policies.  Further, it appears that these
          deterministic rules are most relevant at the level of
          the system of subnational agents, not the agent in
          isolation.
          
          
                         Table Four
                               
              Distribution of Policy Adjustments
                (Additions and Deletions Noted)
                                
            changes         -9  -8  -7  -6  -5  -4  -3   
            frequency        1   0   0   1   2   1  12  
          
            changes         -2  -1   0   1   2   3   4     
            frequency       19  63  338 121 61  27  26
          
            changes          5   6   7   8   9  10  11
            frequency       16   4   2   4   0   0   2
          
          
          
               With empirical findings in favor of SOC, it is
          necessary to return to the literature on this
          phenomenon to extract parts which will help us
          understand subnational policy-adjustment behavior.  To
          begin, Held et al. (1990:1120) opened their analysis of
          sandpiles by observing

          Recent theoretical investigations have shown
          that when certain spatially extended
          nonequilibrium systems are driven, they
          naturally evolve into a critical state. This
          state is barely stable and...the occurrence
          of such critical states in nonequilibrium
          systems is spontaneous; it does not require,
          as in equilibrium systems, the tuning of
          experimentally adjustable parameters to
          particular values or critical points.
          
               The importance of this conclusion (reinforced
          throughout the literature on chaos theory and SOC) can
          be applied to states in the federal system.  One
          undeniable fact about economic development policy in
          the 1970's and 1980's has been the lack of direction
          and rules for engagement at the federal level.  Indeed,
          Eisinger (1988) pointed out that "the word 'national'
          almost never modifies 'economic development' in
          American politics" (p. 4).  The flip side of this
          absence at the federal level is the predominant role of
          the states.
               It would be easy to explain agent organization and
          a system-level orientation as the result of federal
          coordination.  However, finding evidence of SOC in an
          unstructured environment is indicative of expectations
          from chaos theory.  Also consistent with the
          expectations from chaos theory is that the agents
          behave within a systemic context, aware of how their
          individual actions affect other agents in the system.
               The quote from Held et al. also calls attention to
          the stimulus-response relationship of such a "barely
          stable" system.  Unlike deterministic models which have
          a proportionality of stimulus to response, the
          dissipative system--by definition being far from
          equilibrium, dynamic, and on the edge of chaos--needs
          only a slight perturbation to set off enormous results.
          Like the stock market crash of Black Monday, 1987, the
          system was SLIGHTLY perturbed but SUFFICIENTLY
          perturbed to set off a chain of events completely
          aproportional to the stimulus.
               This discussion--whether of tectonic plates, stock
          markets, or policy-adjustment rates--leads to a note of
          caution: "Warning! This system may explode at any
          time!"  This potential hazard is absent in
          deterministic and equilibrium perspectives, and its
          absence there highlights a corollary: Analyzing
          dissipative subnational behavior in a deterministic or
          equilibrium perspective will understate the possible
          influence which one more perturbation can create. 
          Truer to reality, the subnational system (like any
          dissipative system) may be like the proverbial camel's
          back--one more straw and, instead of binding under a
          little more weight, the back breaks entirely.  Policy
          makers need to recognize this important feature of
          dissipative systems not only to avoid letting it cross
          into instability, but also to understand how to address
          matters should the system exceed the edge of chaos. 
               Part of the policy makers' understanding of
          dissipative systems can be summed up in the
          unmathematical conclusion: the whole is greater than
          the parts.  Along these lines, Bak and Chen (1991:46)
          pointed out the necessity of caution when they stress
          that large, interactive systems are not just scaled-up
          versions of small systems.  They wrote:

               Traditionally investigators have analyzed
               large interactive systems in the same way
               they have small, orderly systems, mainly
               because the methods developed for simple
               systems have proved so successful.  They
               believed they could predict the behavior of a
               large interactive system by studying its
               elements separately and by analyzing its
               microscopic mechanisms individually. For
               lack of a better theory, they assumed that
               the response of a large interactive system
               was proportional to the disturbance.  It was
               believed that the dynamics of large
               interactive systems could be described in
               terms of an equilibrium state that is
               disturbed now and then by an external force.
          
          In fact, as Bak and Chen made clear, analyses should
          stress the differences between the deterministic,
          equilibrium, and dissipative systems, not their
          apparent similarities.  One system is not a "poor man's
          version" of the other.  One should not be "simplified"
          to the other, nor automatically labled as linear with
          extraneous noise.
               An additional point to emerge from this discussion
          is emphasized by Carlson and Langer (1989:2,635). 
          Their work on slip-slide models (earthquakes) shows a
          potential hazard.  They wrote

               Somewhat alarmingly, our results suggest that
               small events lead to smoothing on a larger
               scale, resulting in larger events later on.
          
          Examples which might explain "larger events later on"
          include sudden falls in the stock market which are out
          of equilibrium with rational expectations or routein
          events which suddenly get out of hand.  However, this
          observation has other applications: what may appear as
          stable working systems may be hiding potential
          cataclysmic events.  Such events could be more
          disruptive in a single event than would many smaller
          events over a longer period of time.  It follows that
          government policies which seek to dampen needed
          adjustments in a dissipative system may be creating a
          situation where a larger and more disruptive single
          adjustment occurs in the system.  In writing on
          organizational behavior, both Gemmill and Smith (1985)
          and Leifer (1989) call attention to this very important
          point.  To quote Leifer (p. 905)

               an organization that chooses to use its
               natural tendencies to dampen change and
               restabilize equilibrium, will, in the long
               run, become increasingly misaligned with its
               environment.  The greater the misalignment,
               the less the organization can depend on the
               environment for the resources and skills
               necessary to renew itself, leading eventually
               to a deterioration in coping mechanisms and
               eventually to substantial decline.
          
               Looking back at the discussion in this chapter,
          results of tests on data of subnational policy-
          adjustment behavior provides strong evidence for SOC
          and against random behavior.  Also excluded as
          candidates for modeling subnational behavior are
          deterministic and equilibrium systems.  It is not
          surprising, therefore, that the review of the
          literature presented in Chapter Two identified so many
          unsuccessful attempts to explain this aspect of policy
          behavior.  Nor is it surprising that Grady (1987) found
          greater support for the equilibrium model over the
          deterministic models he tested.  Recall that
          equilibrium models are closer to dissipative models
          than are deterministic models.
               Although conclusive in their own right, these
          observations are most relevant for the discussion which
          follows concerning biology models of competition
          (Chapter Five).  Of the four models to be tested, none
          is based consciously on a dissipative approach. 
          However, two models suggest movement (Red Queen and
          Turnover-Pulses) in contrast to the remaining two which
          emphasis stasis (Punctuated Equilibrium and Stationary
          View).  Thus, it might be expected as a preliminary
          assessment that the models which stress movement would
          be more successful in explaining the dissipative
          subnational system than are equilibrium models
          stressing stasis.
               Chapter Five explores the relevance of biology to
          subnational behavior.  In an environment of individual
          agents seeking their survival, which strategies and
          patterns of interaction prevail?  





                         Chapter Five
                               
                     Biological Models of
                    an Evolutionary System
                               
                               
                                
               CHAPTER ABSTRACT: The present chapter builds
               on work presented above.  As shown in Chapter
               Four, state policy-adjustment behavior is
               consistent with self-organized criticality. 
               Having identified a system, it is appropriate
               to inquire into the specific behavior which
               takes place within the system.  Biological
               models of evolution--agents within an
               ecosystem--can help clarify subnational
               behavior.  Empirical tests of four contending
               models (described in Chapter Three) help
               identify the dynamics which take place
               between subnational agents.
          
          MODELS OF COMPETITION

               How do states behave when adopting economic
          development polices?  A partial answer has been
          provided in Chapter Four.  However, this answer can be
          supplemented by reference to other models which
          recognize individual agents within a system. 
          Candidates for inclusion at this juncture include
          analyses of cooperation, competition, avoidance,
          learning, and/or combat (to cite a partial list). 
          While recognizing that many other models are legitimate
          contenders, the present research limits analysis to
          four models of competition.
               Discussed here are four models drawn from
          evolutionary biology which can give insight into the
          behavior driving subnational policy adjustment.  Each
          model stresses a different aspect of how agents
          compete.  Hence each model identifies a different
          fingerprint which is expected to appear in the data. 
          The biological model which most closely matches
          subnational policy-adjustment behavior may also best
          capture its dynamics.  Further, models which fail to
          explain subnational policy-adjustment behavior would
          seem to be the least similar to the phenomenon under
          discussion.
               In testing Hypotheses Three through Six, states
          are compared to species, the latter of which evolve
          through genetic mutations in the chromosomes (see
          discussion in Chapter One of Alchain's [1950, 1953] and
          others.  States, however, evolve through policy
          adjustment.  The competitive advantage which genetic
          mutation brings to species is comparable to the
          competitive advantage which state leaders hope will
          accrue to their state through policy adjustment.  Each
          state seeks to use policy options to become a more fit
          competitor for its goal of obtaining economic
          development opportunities.
               Morphological characteristics in species reflect
          each individuals' phenotype.  The same holds true for
          states; the "pro-business image" of the state (its
          morphological characteristics) is a direct reflection
          of that state's policy adjustment record (its
          phenotype).  Given this linkage, comparison of species
          can be undertaken by shifting analyses from
          morphological characteristics to the genetic structure
          of the species.  In fact, much work in biology has
          focused on this approach (see Ayala [1976] and its
          bibliography).  This direct comparison of phenotypes
          also has its place in comparing states: the present
          research examines the "genetic" structure of the states
          by looking at the policy-adoptioon record of each
          state.  The list of 46 policies (see Table One) tracked
          from 1970 to 1986 is the state's evolving chromosome. 
          Mutation occurs with policy adjustment.
               Gingerich (1993) outlined his own approach to
          analyzing evolutionary rates.  His comments, even if
          not his intent, show the place of complexity and chaos
          theory in this application.  Gingerich wrote:

               Evolution is a gradual step-by-step process
               in which each generation inherits a
               developmental program that determines its
               form.  The program is inherited from the
               preceding generation, and thus evolutionary
               sequences are appropriately studied as
               GENERATION-BY-GENERATION TIME SERIES.  THIS
               APPROACH PROVIDES A MEANS FOR DISTINGUISHING
               SIGNIFICANT NON-RANDOM BEHAVIOR FROM             
               STOCHASTIC RANDOM CHANGE. (p. 459; emphasis
               added)
          
          Thus Gingerich picks up on aspects of the preceding
          discussion through reference to distinguishing random
          from non-random behavior.  Further, the biological
          "generation-by-generation" time series equates in
          public policy to yearly legislative adjustments in each
          state's public policy package.  
               In the following four sections, each biological
          model is further described with reference to the
          Hypotheses which were developed in Chapter Three.  The
          goal of this discussion is to determine which model(s)
          capture the dynamics of the subnational policy-adoption
          record.  The Conway data are again used to match the
          expectations of the models and draw conclusions.

          TEST OF HYPOTHESIS THREE
          THE RED QUEEN HYPOTHESIS

               The RQH has its place in a dissipative systems
          approach as well as an application to subnational
          behavior.  As for dissipative systems, consider Lewin's
          remarks:

               The Red Queen effect is particularly
               pertinent in biology, because it is a
               reminder that species do not lead isolated
               ives but instead are linked inextricably
               with others.  The evolutionary success of one
               species may therefore be as much a function
               of what other species do as what the species
               itself does.  Some biologists go so far as to
               argue that the Red Queen is THE driving force
               in evolutionary history, with environmental
               change playing only a minor part.  The notion
               is clearly resonant with the dynamics for
               complex systems, an internal rather than an
               external engine for change for the species as
                a community. (p. 58; emphasis in original)
          
               The dynamics of the RQH can be applied to states
          in the federal system.  In this application, states
          "evolve" (with the goal of becoming more fit
          competitors) by adopting economic development policies.
          The RQH suggests that individual goals may not be
          inconsistent with systemic goals.
               Chapter Three introduced the RQH.  The Hypothesis
          which was generated from this discussion, and the Test
          for the Hypothesis, stated

               Hypothesis Three: Subnational policy-
               adjustment behavior displays the
               coevolutionary characteristics of Van Valen's
               Red Queen Hypothesis.
          
               Test for Hypothesis Three: Subnational
               policy-adjustment behavior displays evidence
               of each state's momentary fitness which
               fluctuates in the short-term in reference to
               all other systemic agents but shows no net
               advancement over the long-term relative to
               other subnational agents. 
          
               Given that all states' fitness levels can be
          measured between 1970 and 1986, the RQH can easily be
          tested empirically for its application to economic
          development policy adjustment.  As such, any state's
          fitness level (measured vis--vis all other states in
          the system) is expected to fluctuate over time as it
          both gains fitness (through its own policy adjustment)
          and looses fitness (through policy adjustment of all
          other states).  Fluctuation over the years, however,
          will be articulated around a specific point of fitness
          (see Van Valen, 1974:299), the value of which is not
          relevant to the RQH.  
               Van Valen (1973:18) noted the relationship between
          a series of momentary fitness levels.  He wrote:

               Therefore the effects of these minor
               perturbations are not independent of each
               other over time.  If they were, they should
               not be distributed about a constant mean. 
               The rate of deterioration of the effective
               environment has what we can descriptively
               call the property of homeostasis.  A large
               change in one interval is compensated for by
               a small one later, on the average, and vice
               versa; the mean itself does not undergo a
               random walk.
          
               With these observations, it is possible to show
          whether state momentary fitness levels fluctuate around
          a fixed point using a linear regression.  For each
          state, a linear regression (where the X-axis signifies
          the years between 1970 and 1986 and the Y-axis shows
          the year's fitness level for each individual state)
          should produce a line with a slope equal to zero. 
          Further, the R2 value will verify movement
          (fluctuation) in the system.  (The Y-intercept values
          are unimportant for the RQH although they add important
          information to the discussion of state behavior as be
          discussed in Chapter Six.)
               Start with the Conway data such that the yearly
          difference in each state's policies can be shown.  Sum
          the differences for the fifty states to give the total
          annual change (TAC).  Observe also the value for each
          state's annual change (VS1, VS2,...VS50).  Using the
          same formula for all years studied, each state's
          momentary fitness at any year is defined as

                         VSN-[(TAC-VSN)/49].
                                
          The value VSN represents an unspecified state's
          individual and self-initiated change (if any) in its
          policy package for the one-year interval.  This change
          is positive (i.e., it increases the state's momentary
          fitness) and must be added to the negative impact on
          the momentary fitness resulting from the actions of all
          other states.  The value TAC-VSN shows the annual
          change of all states in the same year except the state
          whose fitness level is being determined.  There being
          forty-nine other states, the value TAC-VSN is divided
          by forty-nine to show the level of disadvantage which
          the other forty-nine states have imposed on SN.
               Table Five shows the regression values for the
          slope, R2, and Y-axis intercept.  Of interest to the
          Red Queen Hypothesis is the values of the slope.  Graph
          One shows the distribution of slopes for all states.
               Before analyzing the results, two cautionary
          points must be made.  First, the RQH itself predicts
          that all states, when analyzed as above, will not
          exhibit an OLS slope exactly at zero.  The case could
          be that a state's momentary fitness is high or low at
          both the starting year and final year of analysis. 
          This condition would result in a positive or negative
          slope which is more an artifact of the sample than a
          result of the phenomenon under study.  Nevertheless,
          RQH predicts that states should show a zero-value slope
          outside of this condition.
          
          
                         *Table Five
                               
                      Regression Data on
                     State Fitness Levels*
                                
               State           Slope     R2     Y-intercept
          
               Tennessee      -0.23     0.22       1.34
               Texas          -0.19     0.10       0.94
               New Jersey     -0.18     0.11       1.87
               Vermont        -0.17     0.07       0.92
               Massachusetts  -0.15     0.19       1.29
               
               Connecticut    -0.15     0.15       1.29     
               Indiana        -0.14     0.13       1.01
               Minnesota      -0.14     0.17       1.14
               Hawaii         -0.13     0.12       0.30
               Washington     -0.13     0.04       1.45
               
               New Mexico     -0.13     0.11       0.49
               Georgia        -0.12     0.14       0.55
               New Hampshire  -0.11     0.14       0.65
               North Dakota   -0.10     0.16       0.65
               Nevada         -0.10     0.15       0.34
               
               Montana        -0.08     0.05       0.65
               Oregon         -0.08     0.02       0.93
               South Dakota   -0.08     0.03       1.15
               New York       -0.07     0.08       0.00
               Oklahoma       -0.04     0.01       0.07
               
               Virginia       -0.03     0.01       0.07
               Kansas         -0.02     0.01       0.20
               Wisconsin      -0.02     0.01      -0.05
               Idaho          -0.02     0.01      -0.20
               Delaware       -0.02     0.00       1.03
          
               West Virginia  -0.01     0.00       0.41
               North Carolina -0.00     0.00      -0.45
               Michigan       -0.00     0.00       0.69
               Maine          -0.00     0.00       0.32
               Maryland        0.03     0.01      -0.04          
               
               Utah            0.04     0.05      -0.54
               Wyoming         0.04     0.02      -1.03
               Ohio            0.05     0.01      -0.35
               Arizona         0.07     0.04      -0.78
               Illinois        0.07     0.03       0.09
               
               Alabama         0.07     0.06      -0.66
               Pennsylvania    0.07     0.05      -1.12
               Arkansas        0.09     0.11      -0.65
               Rhode Island    0.09     0.11      -0.90
               Florida         0.09     0.08      -0.62
               
               Nebraska        0.12     0.11      -1.08
               Alaska          0.13     0.08      -1.48
               South Carolina  0.13     0.08      -1.05
               Kentucky        0.15     0.22      -1.05
               Mississippi     0.18     0.29      -1.23
               
               Iowa            0.19     0.16      -0.85
               Missouri        0.23     0.11      -1.22
               California      0.24     0.16      -0.91
               Colorado        0.26     0.43      -1.86
               Louisiana       0.31     0.15      -1.24
               
          *data ranked in ascending order based on slope
                                
                    
               Second, a high number of regression slopes at zero
          can reflect stability in the system rather than Red
          Queen dynamics.  To eliminate the possibility of this
          condition, the regression slope must show a near-zero
          value in the presence of agent movement.  As such, the
          R2 values reflect fluctuation.
          
          
          
                         Graph One
                               
                     Distribution of Data 
                        from Table Five
                                
                 Slope           R2         Y-intercept
          
               -0.25   0      0.00   0       -2.0  0
               -0.20   1      0.03  16       -1.7  1
               -0.15   3      0.06   6       -1.4  1
               -0.10  10      0.09   5       -1.1  4
               -0.05   5      0.12   8       -0.8  7
               -0.00  10      0.15   4       -0.5  5
                0.05   4      0.18   6       -0.2  4
                0.10   7      0.21   1        0.1  5
                0.15   4      0.24   2        0.4  4
                0.20   2      0.27   0        0.7  7
                0.25   2      0.30   1        1.0  3
                0.30   1      0.33   0        1.3  6
                0.35   1      0.36   0        1.6  2
                0.40   0      0.39   0        1.9  1
                              0.42   0
                              0.45   1
          
          
          
               To show a near-zero value of the slopes, refer to
          Table Five and Graph One.  The slopes range from ****
          to **** with a variance of ****.  Graph One also shows
          that the distribution peaks near the predicted zero-
          slope value.  These findings are consistent with RQH
          expectations.
               To show a dynamic system, refer to the R2 values. 
          Notice that the values do NOT show a good fit.  This
          outcome is consistent with RQH and shows that the
          states are in movement.  A good fit would suggest
          little movement; a poor fit suggests greater levels of
          movement.  Because R2 is low and the slopes are near
          zero, the RQH is supported.
               The analysis of data is consistent with
          expectations of Van Valen's RQH.  Although the
          subnational system exhibits fluctuations and movement
          (i.e., it is a dynamic system), individual agents'
          policy-adoption rates fluctuate around a constant mean
          value which is particular to the agent.  To return to
          the Red Queen in Carroll's text, the agents are all
          running but staying together.  No agent breaks from the
          pack.

          TEST OF HYPOTHESIS FOUR 
          PUNCTUATED EQUILIBRIUM

               Unlike RQH (which models continuous activity and
          species turnover even in the absence of perturbations),
          the remaining three models of evolution stress the role
          and importance of stasis as an essential characteristic
          of evolution.  Nevertheless, none of these three models
          sees stasis in the same manner nor due to the same
          causes.  PE is distinct from SV and T-P in seeing
          stasis at a species level.  According to PE, species
          evolve in bursts and these bursts are separated by long
          periods of inactivity.  In contrast, SV and T-P both
          evaluate stasis at the system level.  
               The discussion of punctuated equilibrium in
          Chapter Three generated Hypothesis Four and Test for
          Hypothesis Four.  They stated

               Hypothesis Four: Subnational policy-
               adjustment behavior displays features of
               Eldredge and Gould's model of punctuated
               equilibrium.
          
               Test for Hypothesis Four: Subnational policy-
               adjustment behavior exhibits stasis and
               abrupt evolutionary change, both at the state
               level.  Subnational policy-adjustment
               behavior is not expected to exhibit
               incremental changes or meandering adaptation
               strategies.
          
               Hypothesis Four can be tested for stasis and
          abrupt evlutionary change through reference to the
          Conway data.  As for stasis, Hypothesis Four expects to
          find evidence in the policy-adoption record that each
          state has made no change in its net economic
          development policy package.  Thus, when comparing years
          in a time-series analysis, states should regularly show
          no difference.  
               Table Six shows each state's stasis in terms of
          consecutive years without a net change in its policy
          package.  In reference to these data, a case can be
          made that states exhibit periods of stasis.  Policy-
          adjustment behavior is not observed in every yearly
          increment.  In fact, for each single-year increment,
          **% of the states do not make policy adjustments.  This
          percentage drops as the length of time increases:
          ******.
               In addition to showing stasis, states must exhibit
          abrupt evolutionary activity in a specific direction. 
          Put differently, states are expected not to show
          incremental evolutionary changes or policy-adjustment
          patterns suggestive of a lack of a specific direction.
          
          
                            Table Six
                               
                     Stasis in Subnational
                  Policy-Adjustment Activity
                                
                         Number of Years/Frequency of Stasis
          State          1    2    3    4    5    6    total
          
          Alabama
          Alaska
          Arizona
          Arkansas
          California
          
                    ...........................
          
          Virginia
          Washington
          West Virginia
          Wisconsin
          Wyoming
          
          totals
          
          
          
               Data on this second aspect of state behavior are
          already in evidence from Tables Two and Four.  It can
          be seen that rapid evolutionary changes are rare
          events.  Further, the high number of policy retractions
          is contrary to expectations of policy-adjustment
          behavior following a specific direction.
               An additional expectation of PE--although by no
          means an essential expectation of the theory--refers to
          "living fossils", or species which have not evolved
          over millions of years (an extremely long stasis).  Are
          there any "living fossils" in the subnational list?  In
          fact, all states have been active in evolving as
          competitors for economic development opportunities. 
          The greatest change from 1970 to 1986 was *** and the
          least ***.  Thus, another expectation of PE is not
          observed in the subnational record.
               In sum, inspection of the policy-adoption records
          for each state suggests the presence of stasis at the
          state level but fails to find abrupt evolutionary
          activity to the exclusion of incremental and
          undirectional adjustments.  It is therefore not
          possible to suggest that Hypothesis Four goes very far
          in capturing the dynamics of subnational behavior.  The
          dynamics of PE do not adequately match the observed
          policy-adoption record of subnational agents to permit
          extrapolation of agents' behavior according to the PE
          model of competition.

          TEST OF HYPOTHESIS FIVE 
          STATIONARY VIEW MODEL

               Stenseth and Maynard Smith have been the primary
          advocates in favor of the Stationary View (SV) model, a
          non-biotic variant on RQH.  They agree that advances by
          one species lead to deteriorations in competitiveness
          by others.  However, SV takes exception with the zero-
          sum assumption which is integral to RQH.  This
          difference, as it works out, leads to a set of
          evolutionary possibilities which strongly contrast SV
          to other biological models of evolution.
               Stenseth and Maynard Smith (1984:870) recognize
          only RQH and SV as viable alternatives (although they
          acknowledge PE).  As such, they wrote

               The main conclusion to emerge is that
               ecosystems are expected to approach one of
               two evolutionary modes.  The first mode,
               corresponding to the Red Queen condition, is
               a steady state of change (even in a constant
               physical environment) characterized by
               continuing evolutionary change, extinction
               and speciation.  The other is one of
               evolutionary stasis, which zero rate of
               evolution and no extinction or speciation;
               evolutionary change occurs only in response
               to changes in the physical environment. 
          
               Chapter Three introduced the general aspects of
          SV.  The Hypothesis which was generated from this
          discussion, and the Test for the Hypothesis, stated

               Hypothesis Five: Subnational policy-
               adjustment behavior displays features of
               Stenseth and Maynard Smith's Stationary View
               model. 

               Test for Hypothesis Five: Subnational policy-
               adjustment behavior shows evidence of bursts
               of evolutionary activity associated with
               perturbations to the physical environment
               followed by a slowing down of evolutionary
               change and eventual stasis until the system
               is again perturbed. 

               Stenseth and Maynard Smith base their analyses on
          an EVOLUTIONARY LAG analysis (Maynard Smith, 1976b). 
          In a series of articles (Lawlor and Maynard Smith,
          1976; Maynard Smith, 1976a, 1976b; and Stenseth and
          Maynard Smith, 1984), the theoretical arguments for SV
          were developed.  This model expects certain physical
          evidence.  As pointed out,

               on the Stationary view we would expect bursts
               of evolution, extinction and speciation
               associated with, and caused by, major changes
               in the physical environment.  Following such
               bursts, there would be a slowing down of          
               evolutionary change and species turnover,
               until a new physical catastrophe kicked the
               system into motion again. ....  Species are
               expected to remain phenotypically fairly
               constant for extensive periods. (Stenseth and
               Maynard Smith, 1984:878)

          Importantly, SV has three characteristics: evolutionary
          bursts in conjunction with environmental perturbations;
          slowing evolutionary rates; and stasis at the system
          level.
               How best can the data be presented to test
          Hypothesis Five?  The "Adjustments" column in Table
          Seven shows the following tallies marked by the letters
          shown:

               A. number of states deleting policies;
               B. number of policies deleted;
               C. number of states adding policies;
               D. number of policies added;
               E. number of states with no adjustment;
               F. annual gain/loss in policies (D-B); and
               G. annual activity (D+B).

          Although variable F references the possibility of gains
          or losses in policies, the yearly totals show only
          gains.
               SV projects an environmental perturbation which
          causes an evolutionary burst.  For each variable
          tracked in Table Seven, an evolutionary burst shall be
          defined as follows:

               A. more than ten states;
               B. more than twenty policies;
               C. more than twenty states;
               D. more than fifty policies;
               E. fewer than fifteen states;
               F. greater than or equal to forty policies; and
               G. greater than sixty policies.
          
          
                         Table Seven
                               
                    Policy-Adjustment Rates
                           1970-1986
                                
          
          Year Interval     Adjustments        Burst Criterion
                        A, B; C, D; E; F, G
          
          1970-1971     7,17; 6,34;25;17,51
          1971-1972     4, 4;16,33;37;29,37
          1972-1973     9,14;19,44;22;30,58       
          1973-1974    15,23;21,42;14;19,65     A,B,C,E,G
          1974-1975    12,22;19,43;19;21,65     A,B,G   
          
          1975-1976     7, 8;19,49;24;41,57     F
          1976-1977     6,12;12,38;26;26,50
          1977-1978     4, 6;19,41;27;35,47
          1978-1979     3, 3;23,65;24;62,68     C,D,F,G
          1979-1980     4, 9;10,21;36;12,30
          
          1980-1981     9,21;21,36;20;15,57     C
          1981-1982     8,11,21,51;21;40,62     C,D,F,G
          1982-1983           *
          1983-1984           *
          1984-1985     5,10;22,62;23;52,72     C,D,F,G
          
          1985-1986     6, 6;17,41;27;35,47
          
         rounded means: 7,11;17,41;25;31,55
          
          
               
               Based on these variables and the accompanying
          definition of evolutionary burst, the strongest case
          for an evolutionary burst is found in 1973-1974, 1978-
          1979, 1981-1982, and 1984-1985.
               SV predicts that high levels of activity will be
          attributed to an environmental stimulus.  Referring to
          the four yearly intervals above, the OPEC oil embargo
          may account for activities in 1973-1974 and again in
          1978-1979.  Curiously, however, the states seem to have
          followed alternative strategies with each shock: in
          1973-1973, states reduced policies more than in 1978. 
          Nevertheless, SV doesn't ask how evolution is achieved;
          it looks for activity.  Therefore SV may explain 1973-
          1974 and 1978-1979 activity by referencing the
          influence of OPEC.  Further, the 1974-1975 activity--
          higher than average in three categories--may also be a
          result of OPEC.
               Under assumptions of SV, harder to explain using
          the SV model is the activity in 1981-1982 and 1984-
          1985.  SV requires an environmental perturbation; the
          system cannot initiate change sponte sua.  However,
          what environmental shock can be identified for these
          years?  In fact, the 1981-1982 and 1984-1985
          evolutionary bursts are not problematic for a
          dissipative system.  Although these two bursts seem
          uncorrelated to large external perturbations, a
          dissipative system on the edge of chaos can be
          purturbed by small events which may not be identifiable
          in an analysis of the external environment.
               Even accepting the influence of OPEC and some
          unidentified environmental perturbation in 1981-1982
          and again in 1984-1985, an essential aspect of SV
          predicts a tapering off of evolutionary (policy-
          adjustment) rates and eventual stasis.  It makes sense
          to indentify statis in reference to two variables: the
          number of states making no adjustment in their policy
          package as well as the number of total policy changes
          (variables E and G of Table Seven).  "Stasis" at the
          system level is then defined as when 1) forty of more
          states make no policy changes in a year (given the
          average number at 25) and 2) the total policy change is
          less than 30 (given the average at 55).
               Given these variables and definitions, at no point
          does the subnational system suggest stasis as required
          by SV; policy-adjustment rates continue to be strong
          throughout the sixteen-year interval of data.  The
          lowest adjustment rate--thirty adjustments in 1979-
          1980--is immediately followed by high values of 57 and
          62 policy adjustments.  There is no definitive evidence
          between OPEC shocks nor after the 1978 shock to suggest
          an evolutionary trend toward stasis.
               Overall, the data suggest that subnational policy-
          adoption rates may be sensitive to external
          perturbations but that the system does not function
          under the assumptions essential to SV.  In fact, the
          chief characteristic of the model--that the system will
          settle into stasis after a perturbation--is the most
          problematic aspect of Hypothesis Five to support. 
          Indeed, the closest the system moved to stasis was in
          1979-1980 when only fourteen states modified their
          policy package to delete nine policies and add twenty-
          one.
               Based on the data and analysis presented here,
          Hypothesis Five cannot be supported.  Although it is
          possible to find evidence of environmental
          perturbations on the system, this aspect alone is not
          sufficient.  Indeed, Vrba's Turnover-Pulse Hypothesis
          (considered in Hypothesis Six), as well as the RQH, can
          incorporate environmental shocks into the model without
          arriving at an stationary interpretation.
               In conclusion, the criterion of identifiable
          environmental perturbations is not applicable to a
          dissipative system.  It therefore cannot be used to
          test the model's application to the subnational policy-
          adjustment record.  However, the essential role of
          stasis serves as a decisive factor in evaluating SV. 
          Stasis is not found and Hypothesis Five cannot be said
          to explain this example of subnational behavior.

          TEST OF HYPOTHESIS SIX 
          TURNOVER-PULSE MODEL

               The final model to be tested against subnational
          policy-adoption records is Vrba's Turnover-Pulse model
          (1982, 1985).  She originally suggested it as a variant
          of PE.  However, in its 1993 version in AMERICAN
          JOURNAL OF SCIENCE the model is different enough to
          merit individual attention.
               Like SV, T-P emphasizes environmental
          perturbations as the sole source of evolutionary
          events.  "Evolution is normally conservative...," Vrba
          (1993:428) wrote, "Speciation does not occur unless
          forced by changes in the physical environment." 
               Vrba described the expectations for this model
          such that

               most lineage turnover in the history of life
               has occurred in pulses, nearly synchronous
               across diverse groups of organisms, and in
               predictable synchrony with changes in the
               physical environment.  MOST OF THESE 
               TURNOVER-PULSES ARE SMALL PEAKS INVOLVING FEW
               LINEAGES AND/OR RESTRICTED GEOGRAPHIC AREAS. 
               SOME OF THEM ARE MASSIVE AND OF GLOBAL EXTENT
               (Vrba, 1982, 1985). (p. 428; emphasis added)
          
          Notice the first half of this quote and how close this
          condition (also stated on page 432 of her 1993 article)
          is to the SV criterion for evolutionary bursts
          (Stenseth and Maynard Smith, 1984:878; cited above). 
          Indeed, "evolutionary bursts" and "turnover-pulses" may
          reduce to the same phenomenon.  However, turnover
          pulses cannot be predicted even given an environmental
          stimulus.  Vrba states "WHETHER you [=the species]
          respond or not is none of my business, but IF you do
          then you must do it together" (1993:431-432; emphasis
          in original).
               Also similar between SV and T-P is stasis.  Vrba
          showed this similarity between the two models when she
          wrote

               in a particular ecosystem all changes in
               species composition should be concentrated in
               pulses initiated by physical environmental
               changes, while ecosystem stasis, at least in
               species composition, should obtain between
               pulses (as predicted also by Stenseth and
               Maynard Smith's 1984, stationary hypothesis...."   
               (p. 428)
          
          However, Vrba points out T-P "does not REQUIRE stasis"
          (p. 446; emphasis added).  When present, each model
          attributes stasis to different factors: in SV it is
          attributable to reduced evolutionar y lag leading to
          reduced evolution; in T-P stasis results from
          evolutionary conservatism in the absence of physical
          perturbations.
               The importance of this discussion can be further
          seen in reference to a further dividing line between SV
          and T-P: Vrba's concept of INITIATING EVENTS, or causes
          which initiate evolutionary activity even if those same
          causes do not carry through the activity once begun. 
          Vrba wrote

               all INITIATION of turnover is exclusively
               physical.  It allows the possibilities of
               gradual microevolution, co-evolution,
               competition, and other proximal biotic
               interactions but reserves the INITIATION of
               turnover for physical changes. (p. 428;
               emphasis in original)
          
               Vrba also described the data-testing process for
          the theory:

               testing for prevalence of turnover-pulses
               depends on statistical analyses of whether or
               not a significant majority of turnover events
               in diverse groups of organisms is involved in
               pulses. (p. 432)
          
               Drawing on the discussion of T-P, Hypothesis Six
          and Test for Hypothesis Six stated

               Hypothesis Six: Subnational policy-adjustment
               behavior displays features of Vrba's
               turnover-pulse model. 
          
               Test for Hypothesis Six: Subnational policy-
               adjustment behavior shows evidence of rapid
               evolutionary movement associated with
               initiating events originating from
               environmental perturbations.  The system is
               otherwise conservative and cannot initiate
               change without an environmental stimulus.
          
               How well do the Conway data match expectations for
          T-P?  Much of the information needed to analyze T-P has
          already been introduced in the discussion of SV.  As
          such, it becomes a question of interpreting the same
          observations for T-P expectations.
               As already stated, the data show that subnational
          policy-adoption rates jumped during the OPEC shocks as
          well as on two later years.  However, failure to find
          an apparent stimulus for the latter two pulses may be
          more a reflection of a dissipative system than the
          failure of the model to meet expectations.  This
          feature of T-P (as with SV) cannot be used as a
          criterion to evaluate the model.
               However, the criterion of stasis can serve as a
          test for the model.  As for stasis, unlike SV, T-P does
          not predict that evolutionary rates must go to stasis
          in the absence of perturbations.  Recall Vrba states
          that T-P can accommidate stasis but stasis is not
          necessary (cited above).  Therefore the primary problem
          with SV--the lack of stasis--does not disquality T-P. 
          Further, the distribution of peaks (pulses) is
          consistent with Vrba's expectations.  The subnational
          system shows "most...small peaks...[and] some...massive
          and of global extent" (Vrba, 1993:428; cited above).
               In evaluating T-P against the data, it must be
          emphasized that the data fit the model only under the
          assumption of a dissipative system.  Such an assumption
          is at odds with Vrba's fundamental interpretation of
          evolutionary systems as "normally conservative" and
          "forced" to change (see citations above). 
          Consequently, although the four observed fluctuations
          (pulses) can be explained in a dissipative system, such
          an assumption is a contradiction to T-P.  For this
          reason it follows that subnational behavior does not
          correspond well with the behavioral dynamics modeled in
          T-P.
               As for the four models which have been discussed,
          the Red Queen Hypothesis best captures the dynamics of
          the subnational system.  This match, as it works out,
          is not simply the best of four possible tests; RQH
          shows a strong match between the empirical expectations
          of the model and data analyses.
               As a result of empirical testing, subnational
          policy-adjustment behavior can be said to be a
          dissipative system characterized by competition by
          agents which are engaged in a zero-sum strategy for
          resources.  This is a non-trivial conclusion which can
          be supported for the first time by appeal to data,
          models, and testing against alternative explanations.  
          The significance of this outcome is the topic of the
          last chapter, Chapter Six, "Conclusions".



                               Chapter Six
                                    
                               Conclusions


     CHAPTER ABSTRACT: The final chapter of this
     dissertation presents an overview of the research
     presented above.  Spontaneous order and behavior are
     examined as means of describing the history of
     subnational economic development strategies.  Having
     shown that a system exists (Hypothesis One), additional
     insight can be gained from economics, boids, and
     percolation analyses.
          Physics noise demonstrated the existence of a
     dissipative system.  This step made, chaos theory also
     describes how a dissipative system changes.  These
     principles are applied to Eisinger's (1988) study of
     state policy adoption.  They are found to be consistent
     with the facts but at odds with his implicit assumption
     of a non-dissipative dynamic among decision makers. 
     The alternatives of order mandated through rules
     (forced order) and order arising spontaneously though
     agent interaction (fractal order) are examined for
     strengths and weaknesses.
          The observations from biology are consistent with
     those from physics; of the four biology models tested,
     the Red Queen Hypothesis alone satisfies the evidence
     of the subnational data.  Comments are made concerning
     resource abundance and its influence on the type of
     completion (see Rosenzweig et al., 1987).  Further, it
     is shown that a dynamic dissipative system can perturb
     itself (contra Vrba's assumptions) and need not rely on
     external or environmental perturbations.
          Last, based on the findings presented in this
     dissertation, some topics for further research are
     identified for public policy and political science
     inquiry.

SPONTANEOUS ORDER
AND BEHAVIOR
     
     At the start of this research, a fundamental question was
asked: What variables influence state policy-adjustment behavior?

At the conclusion of this research, a clearer answer has emerged.
     In the preceding pages it has been suggested that the
deterministic and equilibrium models are inappropriate for
evaluating this public policy phenomenon.  A deterministic system
is based on linearities and proportional cause-effect.  With such
a model, prediction is possible well into the future. 
Equilibrium systems are characterized by coping mechanisms which
restore the system to its original dynamic.  Such systems do not
undergo change to reach a new equilibrium.
     Dissipative systems, by contrast, exhibit features which
better match the policy process by capturing the dynamics of
routine discontinuity in government and the plurality of concerns
resulting from conflicting policy preferences.  The dissipative
model defines itself by establishing spontaneous organization in
the midst of rapidly changing conditions.  Deterministic and
equilibrium models fail to match the environment of public policy
decisions.  
     There are clear differences between the three systems
described here, of which only the deterministic and equilibrium
models are represented in the research on subnational behavior. 
The absence of the dissipative model may be due to the
relatively-recent popularity of complexity and chaos theory as
research orientations.  Whatever the reason for its absence, this
dissertation as provided the first analysis of the subnational
policy process using an approach drawn from complexity and chaos
theory and emphasizing dissipative systems.  A review of the
sources shows no dissertation (as of June, 1994) has been written
linking chaos theory to public policy.
     Of course, the tools available through complexity and chaos
theory are abundant and wide ranging. Therefore, no suggestion is
made that the present inquiry has exhausted the research
potential of this approach.  Instead, the discussion focused on
two options to explore the subnational system: physics answered
the questions concerning what type of system, if any type at all,
was represented by the subnational agents; also, biology answered
questions about the dynamics of the system once identified.
     Using physics to describe subnational behavior has confirmed
that the total output of all subnational economic development
activity is not random.  Rather, state activities--although
undertaken on an individual basis and for self-maximizing
purposes--continue to produce a collective output which is
organized at the system level.
     Rejecting random behavior at the system level requires a
critical and informed look at the data.  What might seem to be a
random distribution of state activities shows the classic
confusion between "random" and "chaotic".  Too often "random" and
"chaotic" are used synonymously, and used to denote "non-linear"
(Glass and Mackey, 1988; Savit, 1988, McCloskey, 1990).  In fact,
whereas both are non-linear, only chaotic data are the product of
deterministic rules.  Random data are not governed by
deterministic rules.
     The data analysis of subnational policy strategies presented
in Chapter Four shows the deterministic origin of a self-
organized critical system.  Such a system has specific behavior:
because it is at the edge of chaos, small perturbations can
trigger large, system-wide responses.  Further, the size and
frequency of system adjustments occur in a log-linear manner with
a slope of the line falling between zero and negative two.
     Finding self-organized criticality in subnational behavior
allows further insights into the organization of agents at a
systemic level.  However, least the most obvious benefit of
analyzing physics noise go overlooked, note that the
investigation into white, brown, and pink noise has shown that
THERE IS A SYSTEM in the first place.  What happens in one state
is relevant beyond its borders.  State leaders may be autonomous
agents seeking their own best outcome (either for themselves or
the state); still, they are also agents in a fifty-player system
which functions according to set rules.
     What are these rules?  There is no assertion here that state
decision makers are even aware of the systemic significance of
their actions.  In fact, it would be precarious to argue that
decision makers overcome cognitive limits and endure high costs
of information to consider their actions in depth and the impact
of those actions at the systemic level.  Rather, the contention
is that state decision makers scan their local environment--
contiguous states, national leaders, other idiosyncratic
influences (Walker, 1969:883)--and act on information and motives
which are decidedly not systemic.  Surprisingly, the outcome of
fifty local self-maximizers in an iterative process (such as
annual adjustments) reaches beyond the agent, the yearly time
increment, and the immediate goal.  The outcomes of local
behavior spread to influence the entire system.
     How can behavior which is regional shape a systemic outcome?
In terms borrowed from economics, the description of individual
state behavior is similar to the behavior of individual consumers
and the formation of a market.  The outcome in both examples is
one of a self-organized system of agents.  Consider, however, the
assumption of the consumers' "perfect knowledge" and the
criticism of this assumption from work on cognitive limits
(Waldrop, 1992; see also Chattoe [1994a] for references to Baumol
and Benhabib, 1989; Futia, 1977; Tesfatsion, 1980a, 1980b). 
Increasingly, evidence is surfacing which suggests that perfect
knowledge is unnecessary for the formation of a market.  Rather,
the desired outcome is produced through iterated purchases, SOME
knowledge of product quality leading to purchasing differentials,
and supply-side competition (Chattoe [1994a] cites Anderline
[1990], Binmore and Samuelson [1990], Eaton and Slade [1989],
Fudenberg and Maskin [1990], Hansen and Samuelson [1988], Kreps
[1990], and Nachbar [1992]).  For both policy adjustment and
market formation, perfect knowledge is unnecessary and
alternative behavior may lead to the same result.
     A second example might help show how else behavior which is
regional can shape a systemic outcome.  Consider Craig Reynolds's
cellular automata experiments with "boids" (for bird-like, or
"birdoid").  This experiment leads to understanding how only a
few rules can create a powerful structure when agents iterate
behavior.  Reynolds (1987, 1992, in press) gave the boids three
basic behavioral rules (as summarized by Waldrop):

Rule 1.   maintain a minimum distance from other
          objects in the environment, including other
          boids;
Rule 2.   try to match velocities with the boids in the
          neighborhood; and 
Rule 3.   try to move toward the perceived center of
          mass of boids in the neighborhood.

     The behavioral outcome of these rules is significant. 
Reynolds observed that regardless of where the boids started,
they always formed a flock (personal communication, 1994).  Yet
at no time were the boids told to form a flock!  SYSTEMIC-LEVEL
FLOCK FORMATION WAS A BY-PRODUCT OF SIMPLE, LOCALIZED RULES.
     The relation between boid behavioral rules and state
decision making practices are close enough to invite comparison. 
Rule One: States evaluate their competitive position (distance)
in reference to "other objects in the environment", such as
contiguous states, national leaders, and other haphazard input. 
Due to cost restraints, availability of information, and
cognitive limits, states do not evaluate the entire system.  Nor
did boids.
     Rule Two: States also seek to remain in the competition for
economic development opportunities.  Consider Jones's application
of the disequilibrium-adjustment model as suggested by Plaut and
Pluta (1983).  Jones (1990:221) explains

     The model assumes that differences among states in the
     supply of public policies attract (or repel) economic
     growth.  Economic growth will occur in the more
     attractive states until equilibrium is restored
     (through, for example, the diffusion of policy
     innovations made in the more attractive states to other
     states), or until basic economic circumstances change. 
     The model is estimated by relating economic growth
     (i.e., change) to LEVELS of policies across states.
     (emphasis in original)

The disequilibrium-adjustment process involves "matching
[=adjusting] velocities with the boids [=states] in the
neighborhood".  It makes no difference whether "neighborhood" is
defined in a strictly geographic sense, a function of cognitive
impact, or a function of predominance in the field.  Rather, it
is significant that the decisions are not made on an evaluation
of the entire system.
     Last, Rule Three: What of the behavior to "move toward the
perceived center of the mass...in the neighborhood"?  This rule
for boid behavior may point to an aspect of state behavior which
has heretofore gone under-noticed.  Moving to the center means,
by implication, not being located at the front or the rear of the
flock.  What significance would this have for states?  To start,
being located at the front could produce two negative outcomes. 
First, the state may be paying too much for economic development.
Leading the states in economic development policies suggests a
policy package with is too generous.  States may want a SLIGHT
ADVANTAGE over competitors; outright leadership may be a
disadvantage on a cost basis.  Second, consider the influence of
a mobile workforce.  Leaders in economic development may end up
attracting the unemployed and underemployed out of other states. 
A reputation that "State X is doing everything it can--and doing
it better than all the other states--to employ its citizens" may
backfire by attracting greater number of job seekers.
     Obvious cost and effectiveness considerations exist for not
assuming the leadership position.  What of the laggards, or the
states which adopt policies the last?  Why are they drawn into
the center?  First, as Walker (1969:890) points out, 


     In fact, once a program has been adopted by a large
     number of states it may become recognized as a
     legitimate state responsibility, something which all
     states ought to have.  When this happens it becomes
     extremely difficult for state decision makers to resist
     even the weakest kinds of demands to institute the
     program for fear of arousing public suspicions about
     their good intentions; once a program has gained the
     stamp of legitimacy, it has a momentum all its own.

Every state needs jobs for the voters; not offering incentives to
business would be a disastrous strategy.     
     Second, a state strategy which rejected offering competitive
incentives to business would mean the state would lose industry,
jobs, and taxes to other states.  Policy makers would be voted
out of office.  The laggard position is inconsistent with
contemporary economic development strategies and political
realities.
     Analysis of potential hazards of leader and laggard roles in
economic development policy-adoption records suggests that the
middle ground may be the safe place to remain.  By rejecting
leader and laggard positions, states come to position themselves
closer to the center of the pack of competitors, thereby
maximizing themselves by avoiding negative outcomes although
receiving fewer positive outcomes.  
     Like Reynolds' cellular automata programs, computer models
using lattice analyses of percolation hold another means of
understanding state behavior.  Percolation, like diffusion,
calculates changes in cells based on events in the surrounding
environment.  Percolation is commonly represented by a forest-
fire simulation.  Albinet et al. (1986:1-2) describe the salient
features of percolation simulation:

     The model is dynamic and its evolution can be observed
     as a function of time.
     The interaction between contaminated and non-contaminated
     sites (heat transfer in the case of a fire):
     a) depends on the nature of each site (type of "tree")
     b) is not limited to first neighbor interactions,
     c) takes into account latency (a "tree" will only
     ignite when it has received enough "heat"),
     d) allows for cooperative contamination (a "tree may be
     "heated" by several neighbours simultaneously), and,
     e) may be non-isotopic (effect of "wind" or "slope" for
     example).

     As was seen for boids, using the above rules for percolation
in a forest fire is both a matter of SUBSTITUTING terms for
forest fires to state diffusion and/or REWRITING THE PROGRAM to
use deterministic rules which more accurately reflect those found
in the subnational system.  Percolation holds the promise of
rewarding future work in understanding diffusion dynamics.  As
for a first step in interpreting results, Albert et al. observed
three regions in a 200 X 200 forest.  They wrote

     In region I, we are in a transitory state where the
     ignition of the whole of the first two rows gives a
     quick advance of the front
     (the system "remembers" the initial conditions). 
     Region II is associated to a steady propagation where
     the scaling laws work: this is the interesting region
     for us.  In zone III, the finite dimension of the
     system is the prevailing factor: the propagation is
     restricted by the border of the forest, giving a
     decrease of the speed of the mean position of the
     front. (p. 4)

This suggests a dynamic which diffusion has been unable to fully
capture and describe.  Percolation analyses may provide important
insights into the spread of policies.

ANSWERING THE QUESTION:
BUT WHY "NOISE"
AND BIOLOGY??

     Finalizing the research outline presented in Chapter One
required a choice to use physics noise and biology models when
analyzing subnational policy-adjustment behavior.  The benefits
of this approach can be seen both by their match to the empirical
record as well as in their ability to bring insight into the
subnational process beyond the testimony of the data. 
The Dissipative System
     There are lessons to be learned beyond the empirical
discussion of physics noise described with reference to
Hypotheses One and Two in Chapter Four.  Having confirmed a
dissipative system, what more is thereby known?  To answer,
Gemmill and Smith (1985) described CHANGE IN A DISSIPATIVE SYSTEM
in terms of 1) disequilibrium; 2) symmetry breaking; 3)
experimentation; and 4) reformation.  Leifer's (1989) four-
component change sequence is slightly different.  He emphasized
1) the point of singularity; 2) transformation utilizing radical
strategies; 3) inefficient acting and experimentation; and 4)
resynthesis.  These two sets of terms emphasized different
aspects of the same change process. 
     Ineffective and inefficient systems must change.  How is
this goal achieved?  As a starting point, the system experiences
disequilibrium.  Rules for coping are ineffective.  The system
must divert increasing amounts of attention, energy, and capital
to maintaining itself rather than achieving its production or
service goals.  Such a system is also fragile.  Even a minor
shock can cause it to experience inordinately high levels of
instability.  
     The more serious the disequilibrium, the less strong a
disturbance it takes to send the system out of control.  At a
certain point the system will experience a perturbation from
which it cannot recover.  Such a perturbation is called a "point
of singularity" and it triggers cascading, non-linear outcomes
within a system.  The point of singularity is comparable to "the
straw that broke the camel's back" or the match which starts a
vast forest fire.  Both analogies show how even a small event can
trigger a large outcome.  Thus, what might have started as just
another political rebellion or revolt could end up as a full-
scale revolution.
     Experiencing the point of singularity is like crossing the
Rubicon.  Once it's occurred, there can be no turning back.  For
the organization, two changes can result: entropy or
revitalization.  Entropy leads to the demise of the system.  By
contrast, a dissipative system may revitalize itself and produce
order out of chaos.  Revitalized organizations have achieved a
new equilibrium.    
     Arriving at a new equilibrium is a difficult process. 
Systems must work toward a twin processes of jettisoning old
rules and expectations while replacing them with new rules and
expectations.  Neither half is easy.  The old rules which
performed a disservice to the organization may be protected,
deeply entrenched, psychologically safe, or simply too hard to
break.  Further, new rules are neither apparent nor assured.  The
period following the point of singularity is therefore filled
with ineffective and inefficient experimentation, trial and
error, individual initiatives, and unconventional problem-solving
approaches.  The system, in other words, is in chaos.
     Chaos rules.  However, a fundamental tenet of chaos theory
holds that order emerges spontaneously from chaos.  When an
organization is in chaos, some experimentation will be
successful.  What doesn't work is eliminated and other options
are brought forth.  Trial and error eventually shows what types
of solutions the system prefers and what it rejects.  The next
round of solutions draws on the stability and learning of the
first set of trials.  Experimentation and unconventional problem-
solving approaches continue.  However, the levels of chaos
diminish and the organization finds higher levels of order.
     Change within a dissipative system will eventually lead to
order which is built on new rules, new coping mechanisms, and
around a new equilibrium.  Further, the emergence of order out of
chaos "just happens" without design or intervention.  Order also
leads to organizational revitalization which is the final outcome
of change in a dissipative system.  The system is in tune with
the environment.  Of course, nothing lasts forever.  The system
may again experience growing levels of disequilibrium.  Such a
system, if it experiences a point of singularity, will again
undergo the chaos-to-order sequence or it will experience entropy
and terminate. 
     History of Economic Development
     How can a dissipative system help us understand the behavior
of American states in the economic development policy domain? 
Although no study has set out to answer this question, Eisinger's
(1988) review of the history of state economic development
efforts from 1930 to 1986 shows characteristics of chaos theory
models.  Even without his using the jargon or references found in
the literature of chaos theory itself, Eisinger has described a
dissipative system.
     As background to the study, Eisinger begins his survey of
subnational economic development efforts by noting that policies
can be either supply-side oriented or demand-side oriented. 
Without entering into the full scale of Eisinger's argument, note
that he holds that the supply-side orientation is generally
ineffective and inefficient whereas the demand-side constitutes
an improvement to subnational economic development efforts.
     As for the point of singularity which ushers in chaos, the
initial stimulus came when Mississippi adopted the Balance
Agriculture With Industry (BAWI) program in the 1930's.  It was
the match in the forest, or, put differently, it "lifted the
curtain on an era of more competitive subsidization and broader
state and local involvement in industrial development efforts"
(Cobb, 1982:5; cited in Eisinger, p. 32).  Whatever the analogy,
Mississippi's innovation became every state's invitation to
follow its lead.  
     According to the dissipative model, this point of
singularity is followed by chaos in the form of trial and error,
local initiatives, and unconventional decision making.  Again
Eisinger provides evidence to show the presence of this "complex
and simultaneous interaction" (Gemmill and Smith, 1985). 
Consider the following citations (emphases added):

     (1) ...economic development is an increasingly
     important government activity....  THERE ARE NO
     ESTABLISHED FRAMEWORKS.  THE FIELD LACKS AN ANALYTICAL
     SYNTHESIS AND A CRITICAL INTERPRETATION. (p. 4)

     (2) the putative value of these INDUCEMENTS, WHOSE
     EFFICACY ECONOMISTS HAVE DOUBTED FROM THE BEGINNING,
     have diminished as more and more states have offered
     them. (p. 11)

     (3) MUCH PROGRAM DESIGN IS FRANKLY EXPERIMENTAL, AS
     STATES TRY ALTERNATIVE WAYS OF WIELDING LEVERAGE in the
     economy. (p. 24)

     (4) State and local officials still devote SUBSTANTIAL
     ENERGIES AND RESOURCES to the competition for footloose
     firms, DESPITE INDICATIONS...THAT SUCH BUSINESSES DO
     NOT CONSTITUTE THE MAIN SOURCE OF EMPLOYMENT GROWTH IN
     A STATE'S ECONOMY.  (p. 25)

     (5) To the extent that states have relied on RANDOM
     INQUIRIES from out-of-state firms and saturation
     prospecting trips, THE PURSUIT OF ECONOMIC DEVELOPMENT
     HAS BEEN AN ESSENTIALLY UNPLANNED ACTIVITY. (p. 26)

     (6) A number of communities and states also produce
     lists of development activities and goals IN MORE OR
     LESS RANDOM FASHION that they label "economic
     development plans". (p. 26)

     (7) MOST OF THE HISTORY OF ECONOMIC DEVELOPMENT IS
     NOTABLE FOR THE ABSENCE OF PLANNING. (p. 29)

     (8) From an economic point of view IT IS SIMPLE,
     THEREFORE, TO MAKE THE ARGUMENT THAT LOCATION
     INCENTIVES ARE INEFFECTIVE as devices for the
     generation of new investment and new jobs.  (p. 337)

     To summarize evidence for a dissipative system in economic
development, the argument can further cite Eisinger who speaks of
a "lack of a strong consensus about what works and what does not"
(p. 24) and a "continual search for whatever appears to work" (p.
31).  
     The final component of a dissipative system is
revitalization, or the emergence of order out of chaos. 
Revitalization is also suggested by Eisinger's observations. 
First, consider the following quote (emphasis added): 

     The thesis of this book is that subnational economic
     development policy has undergone A RECENT SHIFT from an
     almost exclusive reliance on supply-side location
     incentives to stimulate investments to an approach that
     increasingly emphasizes demand factors in the market as
     a guide to the design or invention of policy. (p. 10)

     Second, as if to underscore revitalization in this policy
domain, Eisinger starts his Chapter One by noting the high levels
of "consensus and expectation" for economic development. 
According to Eisinger, this consensus emerged "sometime after the
mid-1970's", or forty years after Mississippi's adoption of the
BAWI policy.  Thus Eisinger distinguishes between the first forty
years of subnational policy adoption aimed at economic
development (from 1930 to 1970) and the years that follow.  His
description of the latter's "ineffective", "random", and
"experimental" policy initiatives is replaced by references to
the former's "design" and "consensus" approach.
     As pointed out in the discussion of dissipative systems,
order must arise from chaos spontaneously; in a dissipative
system, order cannot be imposed.  Eisinger suggested that the
revitalization in subnational economic development policy
preferences was spontaneous rather than a result of federal
intervention.  As pointed out earlier, Eisinger stated briefly
that "the word 'national' almost never modifies 'economic
development'" (p. 4). He elaborated, noting that 

     The invention and implementation of economic
     development policies are the responsibilities chiefly
     of numerous state and local governments....  Most of
     these actors operate independently of one another. 
     There is no federal department of economic development
     to offer guidance, nor is there a national development
     plan or coordinating body.  Federal programs do exist,
     but their exploitation and implementation are left
     mainly to state and local officials and citizens'
     groups. (p. 23)

     Do subnational economic development efforts from 1930 to
1986 (and possibly beyond) show the expected four components of a
dissipative model?  Based on the evidence presented, the answer
is in the affirmative.
     Discussion
     If there is one point where Eisinger parted company with the
discussion of a dissipative system, it is when he identified
stimuli which may have brought about the change from supply-side
policies to demand-side policies in the '70's.  According to
Eisinger, the change was due to "shifts in the distribution of
people and jobs across the landscape, changes in the federal
political arrangement, and deep transformations in the nature of
the American economy" (p. 10).  Perhaps Eisinger was responding
to a felt need to find a causal stimulus for switching from
supply-side policies to demand-side policies.  Indeed, in non-
chaotic analyses, it is natural to think in terms of cause-
effect: if a change occurs, something must have caused it; and
big changes require big causes.  Chaos theory rejects both the
need for a cause as well as the matched scale of cause-outcome. 
This rejection has interesting implications for Eisinger's
conclusions.  
     First, if the argument of this research is accurate--that
is, if subnational economic development is best described as a
dissipative system--it is not necessary to produce a cause-effect
reason for the change from supply-side policies to demand-side
policies.  Chaos theory postulates that order emerges
spontaneously out of chaos.  Thus, unlike a deterministic
argument where both the cause and effect must be identified,
there is no reason in the dissipative study of subnational
economic development to produce a cause for the new order.  This
spontaneous revitalization is sometimes hard to accept. 
Something out of nothing?  Order out of chaos?  It defies our
traditional ways of thinking.  Nevertheless spontaneous order is
confirmed by observations of nature (see Nicolis and Prigogine,
1989) and organizations (see Gemmill and Smith, 1985; and Leifer,
1989).  Thus, researchers must increase their comfort level with
order emerging from chaos and learn to accept spontaneous
revitalization without needing to search for agents of change. 
     Second, Eisinger suggests that three stimuli--shifts in
distribution of population and jobs, federal changes, and a
transformation of the American national economy--led the states
to RESTORE ORDER out of their inefficient acting and
experimentation with supply-side policy adoption preferences. 
The alternative interpretation is that these three influences
DESTROYED THE ORDER of the supply-side policy options which were
out of equilibrium with the environment or, put differently, that
the stimuli INITIATED CHAOS (the direct opposite of "restore
order") in a system in disequilibrium.  In other words, these
three stimuli might have acted individually or collectively as
points of singularity which ushered in experimentation leading to
demand-side options.  In fact, Eisinger himself suggested the
point of singularity option when he stated that "the entrepreneur
[or entrepreneurial state using demand-side policies] encounters
resistance, however, for such activity has a DISEQUILIBRATING
IMPACT..." (p. 8; emphasis added).  
     There is a fine but important distinction to be made between
looking at stimuli as either (1) producing order out of a system
in disequilibrium or (2) producing chaos within the system which
then develops spontaneous order.  The first outcome is "forced
order" and the second outcome is "fractal order".  Diagram One
shows the two alternative sequences.  In Diagram One-A, the
system in disequilibrium regains stability due to the stimulus
acting on the system.  Thus, the level of disequilibrium drops
following the stimulus.  In Diagram One-B, the system in
disequilibrium experiences greater instability (chaos) due to the
stimulus acting on the system.  Thus, the level of disequilibrium
rises following the stimulus.  Eventually, the system finds
greater levels of stability.
     Understanding this distinction between forced and fractal
order must precede any assessment of a system's or an
organization's return to order.  On one hand, the forced order
system shown in Graph One is clearly moving toward a better
equilibrium following the stimulus.  On the other hand, the
fractal order system in Graph Two IS ALSO MOVING TO A BETTER
EQUILIBRIUM although it must pass through a chaotic period. 
Decision makers, therefore, should be cautious to understand the
presence and role of chaos in restoring order to systems.  A
leader who sees greater chaos following a stimulus must
understand that order will eventually arise and that additional
stimuli many be unnecessary and/or harmful to the emerging order.
Nor should chaos be damped or "blamed" on organizational members.
     Comparing patterns of forced order to fractal order, it is
noted that both systems eventually arrive at order.  However, the
differences are important here: for the forced order option,
order is imposed, as through organizational mandates, laws, or
authority, all with the explicit assumption of a deterministic or
stable environment.  In contrast, for the fractal order option,
order arises naturally from the experimentation and innovation by
the agents in the system.  This difference in origin of order
raises the question: which alternative route to order is better? 
Said differently, is it better to allow a system to experience
naturally-occurring chaos and develop its own rules leading to an
equilibrium (bottom up fractal order), or should rules and
objectives be stipulated and the organizational members made to
conform (top down forced order)?  It would of course be nave to
suggest that one answer fits all options.  In some instances, the
system cannot experience high levels of chaos without incurring
high costs to those involved.  This is especially true for
political organizations such as nations where the cost of chaos
is citizen death or misery (starvation, riot, war, dislocation,
etc).  Conversely, even these examples may benefit from fractal
order under the recognition that forced order is not a panacea. 
Forced order may fail to reflect the dynamics of the system and
its members, thus perpetuating disequilibrium.  Forced order may
also fail to keep in tune with the changing time, thereby
creating increasing obstacles to the organization's goals.
     Unfortunately, when deciding between fractal order and
forced order, the issue may be determined for all the wrong
reasons.  Consider the time factor involved: arriving at order
out of chaos may be a lengthy process.  Some organizational
objectives may not have the time to allow order to evolve. 
Private-sector organizations are in a highly competitive
environment.  Chaos could lead to stock-holder dissatisfaction
and customer defection.  Also, if the public sector is in
disequilibrium, the politicians who waits for order to emerge
spontaneously will be vulnerable at the next election.  Thus,
there is a tendency to mandate order so as to forestall possible
criticism for indecisiveness, incompetency, or preference of the
status quo.  Action makes for good headlines, shows leadership,
and increases personal levels of authority and power.  One thing,
however, is certain: No politician or organizational leader ever
received credit for what happened spontaneously.
     Although no a priori logic can affirm whether forced order
or chaos-derived fractal order is better, some broad statements
about relative advantages can be made.  As such, forced order
tends to be less ambiguous.  Organizational members have clear
marching orders.  Joined to clarity is the psychological lift
that the problem will soon be better.  Last, forced order is
instantaneous and directives come from a well-defined source of
authority.  Fractal order, however, is a lengthy process,
undirected, a joint undertaking, and may require greater
sacrifices before arriving at the benefits.  Personnel are
psychologically ill-disposed to "wait out" the distant arrival of
order (like waiting for Godot).  The shifting equilibrium makes
job performance difficult to define and evaluate.  Personnel may
experience difficulties with the high levels of uncertainty
during the organization's search for order.
     In contrast, although chaos-derived fractal order is slower
and uncertain, it may be absolutely necessary.  Chaos-derived
order recognizes that no one knows how to mandate solutions.  The
bounded rationality of the decision makers cannot foresee all the
possibilities of organizational needs.  Nobody has the necessary
knowledge, analytic abilities, or imagination to solve vast and
complicated problems.  Further, top-down authority may alienate
some of the organization's personnel.  Chaos-derived order arises
from the interplay of the organization's needs and objectives and
it may better recognize the role of personnel as important
elements in the organization.  
     As for whether fractal order or forced order is better,
consider the short-term versus long-term outcomes.  If bringing
an organization back to equilibrium in the short term means
forcing it back through mandated rules or coercion, the objective
may be realized but there is no reason to believe (as often seems
the case) the results will be satisfactory in the long term. 
Imposed order may temporarily mask the problem without ever
addressing the fundamental causes.  The small advantage of quick
order through mandates may mean a greater loss later.
     Nevertheless, the choice between either forced or fractal
order need not be bipolar.  Even if a business or political
situation cannot wait for order to emerge out of chaos, leaders
can note how the system moves spontaneously from chaos to order
and formulate rules for returning to equilibrium based on the
signs given by the system.  The leader can thus facilitate or
accelerate organizational change by watching where order emerges
and by building around those loci of order in ways the
organization seems to prefer.  Further, it would be advisable
that the alert leader also destroy order when such order may lead
to future entrenched interests or sub-agenda behavior.  In any
event, no assumption should be made that "one size fits all" when
it comes to organizational change.
     Conclusions
     The goal in this research has been to show that the history
of economic development policy adoption from 1970 to 1986 can be
described as a dissipative system.  This goal has been greatly
advanced: By giving a mixture of qualitative and quantitative
evidence, it was possible to identify characteristics of a
dissipative system.  Further, this analytic approach supports
Eisinger's conclusions concerning the relation between the states
and Washington.  Eisinger wrote:

     A...reason for this study is to suggest the importance
     of attending to subnational policy developers in the
     United States, for to overlook these generally in favor
     of a focus on Washington, particularly in the field of
     economic policy-making, is to come away with only a
     partial grasp of a nature of American political
     impulses and possibilities. (p. 5)

     This discussion also argues that models based on
deterministic or equilibrium assumptions have had only limited
success in explaining subnational economic development records
because they are the wrong models.  To replace these models, a
dissipative system emphasizing "profound reformation of system
parts [states] and a total alteration of forms, relationships,
and/or the process of maintenance and growth" (Gemmill and Smith,
1985) is suggested.
     The discussion of the three models--deterministic,
equilibrium, and dissipative--and their use for analyzing policy-
adoption records suggest an important point: Linear models of
analysis are inappropriate for describing and explaining non-
linear events.  When linear models are (mis)applied to non-linear
phenomena, they produce incorrect results.  Worse, these
incorrect results have the appearance of credibility and
authority due to the power of statistical testing methods and
empirical analyses.  Such an appearance of authority can be hard
to dispel.  Clearly, the right tool for the job is an absolute
necessity. 
     Additionally, this analysis has shown that research must
look at phenomena at the macro-level.  Seemingly in every field
of science the object of research has been dissected into smaller
and smaller pieces.  Valid as such an approach is, it is not the
only approach.  In some cases, emphasizing the small components
means losing sight of the system as a whole.
Biology Models
     The analysis of physics noise affirms the existence of a
system.  It also opens up multiple questions about what the
system does.  There are many possible avenues of inquiry at this
point, including those stressing cooperation among agents,
competition, learning, diffusion, percolation, coevolution, and
so forth.  Again a limit is imposed on the research to focus only
on competition within the system without suggesting that this
choice is better than any others or that it is correct. 
Additional research can return to this juncture and follow
different lines of inquiry.
     Understanding how systems evolve can bring insights into the
origin, life, and fate of the system.  Biology identifies four
principle models of evolution, each placing an emphasis on
different relations among agents in the system, which can answer
this question.  To the extent that the subnational system shows
characteristics expected by one or another model, the subnational
system may be influenced by the same factors as the biology
models.  Importantly, the present discussion does not analyze the
validity of these models to biology; rather, the present research
tests how expectations of each model are represented in the
policy-adjustment history of states.
     Coronation of the Red Queen
     The Red Queen Hypothesis showed the best match between
expectations and observations in the subnational system. 
Further, the match is not only the best when ranked against the
other three models.  The Red Queen Hypothesis shows an excellent
match between expectations and the data from policy adjustment.
     What does the success of the Red Queen Hypothesis mean for
public policy analyses?  To answer, recall that Red Queen
describes a dynamic system of zero-sum competition.  No state
pulls ahead despite constant policy adoption.  At any time the
states may exchange relative competitive positions due to
fluctuations in adoption.  However, the short-term pattern is not
matched by the long-term outcome.  In the long-term, the
fluctuations of states' policy adoptions level out such that the
long-term relative positions of states remain constant.  Despite
all the policy-adjustment activity in the states, no state in the
long-term in better off in competing for economic development
opportunities.
     The Red Queen Hypothesis is also a biotic model which can
also facilitate external perturbations.  Considering that it is a
biotic model, changes in the system can originate from within the
system itself.  If this is correct, states are not solely driven
by external factors as under a Turnover-Pulse or Stationary View
model.  For self-maximizing states, the biotic model holds an
important possibility: states are at least in part masters of
their own fates.  Further, this feature of the subnational system
suggests that the states determine both the PRESENT dynamic of
the system as well as the FUTURE dynamic.  States may be able to
change the rules of engagement for system-wide economic
development strategies.
     Solid findings in favor of the Red Queen Hypothesis are only
possible if the fifty states form a system and if their
collective behavior is not random.  Although the other biology
models make a similar requirement of the data, verification of
the Red Queen Hypothesis points to a second means of verifying
the conclusions of Hypothesis One.  There is consistency between
the self-organized criticality tested in Hypothesis One and the
Red Queen Hypothesis tested in Hypothesis Three.  
     The collective failure of Punctuated Equilibrium, Stationary
View, and Turnover-Pulses suggests that the dynamics they tried
to model were not present in the subnational system.  Notably
these models represent conservative systems showing stasis or
equilibrium.  However, the data of state policy adoption are
clear in showing a yearly strategy which meanders (contra
Punctuated Equilibrium) by both increasing and decreasing and/or
altering the economic development policy package (see Table
Four).  Sudden bursts of changes followed by long-term stasis are
inconsistent with subnational policy-adjustment records.
     The Stationary View also fails to meet the expectations of
the data.  All states compete by continuously enlarging their
policy packages; the result is continuous movement, not
stationary equilibrium.
     Further, the Turnover-Pulse model describes a system which
is driven to change by exogenous influences; the population of
agents does not itself initiate change.  Such thinking is
consistent with systems in equilibrium or stasis; however, the
subnational data show that the system is not conservative. 
Further, the dissipative system demonstrated in Hypothesis One
does not need an initiating cause to generate change.
     Implications of Findings
     The discussion of biology models is rich in potential for
additional applications to subnational policy adjustment in a
dissipative system.  To demonstrate this potential, two research
opportunities are examined: competitive strategies based on
resource abundance, and the dynamics of systems which appear to
be conservative but may be dissipative.
     Important insights from Rosenzweig, Brown, and Vincent
(1987) suggest the dynamics which contrast Red Queen evolutionary
patterns from the Stationary Model.  Rosenzweig et al. observed
that resource limits influence which evolutionary pattern the
system will display.  Thus, in the face of abundant resources--
incentives in this case--the system evolves according to Red
Queen expectations.  In contrast, limited resources lead to an
equilibrium system consistent with Stenseth and Maynard Smith's
Stationary View.
     This observation, distinguishing Red Queen and Stationary
View origins, seems to fit the history of subnational economic
development policy strategies.  States have been described as
"well-stocked candy stores" for industry.  Further, the
innovation, diffusion, and acceptance of a wide variety of state-
sponsored policy incentives points to an abundant supply which
would favor Red Queen evolution.
     However, the correlation between resource abundance and
resulting competitive strategies leads to the conclusion that the
Red Queen competitive model may not be a fixed aspect of
collective subnational strategies.  Said differently, states
could alternatively compete without adopting the abundant list of
policies.  Such competition would recall an equilibrium system,
not a self-organized critical system.
     If COMPETITION is the goal but the FORM OF COMPETITION is
not fixed, which competitive strategy would better maximize
outcome of policy adjustment?  If the answer to this question is
Red Queen, then the states have already found the best solution
to meet their goals.  However, if the answer is closer to the
Stationary View, what might be done to alter the competitive
field to be closer to the Stationary View?
     The above question addresses the option of transforming a
Red Queen strategy to a Stationary View strategy.  Might not this
transformation come about spontaneously when otherwise-abundant
resources are depleted or gradually become scarce?  Does the
competitive strategy within the system change as resources
dwindle or enlarge?  If so, how might the subnational system
change?
     Again a question, and again further questions which solicit
an answer.  If the states agree to limit the resources they
allocate to economic development, would competition switch from
Red Queen to Stationary View?  Would all states thereby capture
as much economic development benefits for reduced costs?  Would
competition still drive states to innovate and strive for
superior results?  These are questions worth further analysis. 
     Also worth a more detailed examination are Vrba's
observations concerning evidence of the Turnover-Pulse model. 
One aspect of this model deserves special attention.  Recall Vrba
pointed out that "Most of these turnover-pulses are small peaks
involving few lineages and/or restricted geographic areas.  Some
of them are massive and of global extent (Vrba, 1993:428)".  This
observation sounds very much like the empirical expectations of
self-organized criticality.  Compare the pattern shown in Table
Two which produce a slope between zero and negative two on a log-
linear graph.  The self-organized pattern also shows the majority
of events in the lowest range with few events in the highest
range.  
     Surprisingly, however, Vrba uses her observation in support
of a conservative system which does not change except when forced
to.  In contrast, the same observation could be made about the
data describing a dissipative system which is dynamic (not
conservative) and which can change in large ways due to small
perturbations (or far from being forced to).  
     In the absence of further analysis, it is impossible to say
whether Vrba's system is conservative or dissipative.  The
relative frequency of small perturbations to large perturbations
may suggest a conservative system if emphasis is placed on the
low frequency of large events.  However, the small events should
not be emphasized (and the large events declared exceptions);
rather, the analysis of these events must show the relationship
of all events within the system.  Self-organized criticality
incorporates all the data by showing that the log-linear line is
formed by numerous small events and fewer larger events.
     The Turnover-Pulse model also falters under Vrba's
assumption that the system (being conservative) cannot perturb
itself.  For some systemic dynamics--deterministic and
equilibrium among them--such a conclusion is reasonable.  Thus,
Vrba is consistent in her analysis of the habitat as a
conservative system when she concludes it cannot perturb itself. 
However, what if the system isn't conservative?  What if it is
dissipative (as suggested by the relative frequency of small and
large events)?  Under this assumption, the logic that the system
cannot perturb itself is flawed.  For a dissipative system, the
stimulus which perturbs the system can arise from within the
system itself.  Reference to the predator-prey relationship and
logistic population models helps make this point.
     In the predator-prey relationship, the predator and prey
evolve so that over-predation is bad for both predator and prey. 
Over-predation leads to the extinction of the prey and the loss
of resource for the predator.  The logistic population curve of a
prey shows that its population fluctuates in inverse proportion
with a predator population.  Ideally, the population of neither
predator nor prey will decrease to zero although either may reach
very small numbers.
     However, even under the normal, long-term predator-prey
relationship, what if something happened to the prey when its
numbers were at the low range?  What if the species (or
population) went extinct due to a random event?  Clearly, the
prey would be absent from the system.  But its absence would have
implications for the predator as well.  The predator would have
to adjust, possibly relying more heavily on other food sources
previously used by other species.  Whatever the outcome, the
affects of the prey's absence and the resulting survival strategy
of the predator will create a set of perturbations which may
amplify through the entire food chain.  Thus, a small event--the
accidental extinction of a prey--creates feedback and adjustment
which move throughout the entire system.  The system, in short,
has been perturbed not by an environmental modification but by a
small internal event which produces large consequences.
     No doubt additional reflection will show how routine or
otherwise-insignificant events in a dissipative ecosystem can
create outcomes leading to the endogenous perturbation of the
system (contra Vrba).  Likewise, additional reflection will also
show the relation of these comments about ecosystems to the
dissipative subnational system.  Can events which are otherwise
routine eventually lead the subnational unstable system into
chaos?  At the very least, there is no reason to reject the
possibility.  For instance, suppose a state which is facing
financial difficulties, cutbacks, and a hostile electorate (thus,
on the edge of chaos) looses a primary source of employment to a
rival state....  This scenario sounds like the formula for
driving a state over the edge of chaos, thereby perturbing the
system of states in a manner which could show amplifying
iterations.  As a first step, the state may adopt a high number
of economic development policies which, in the manner described
by Red Queen, would kick the entire system into a higher level of
action.
     Synthesis of Models
     Having examined the physics models and biology models
independently of each other, it is necessary to show how they
complement each other.  As for physics, self-organized
criticality describes a dynamic system which never settles into a
long-term stasis or equilibrium.  The system is constantly
reorganizing itself along new lines and according to
deterministic laws.
     Of the four biology models, only the Red Queen describes
dynamics which are consistent with a self-organized critical
system.  Consider the elimination of all others: Punctuated
EQUILIBRIUM and STATIONARY View show by their very names the
equilibrium and stasis which characterize these models.  Further,
although Turnover-Pulses stresses movement, it describes movement
as special events occurring only when a conservative system is
forced to respond.  
     Only the Red Queen model comes close to capturing the
dynamics of a dissipative system characteristic of self-organized
criticality.  Importantly, the empirical test for both models is
different enough to avoid the suggestion that one model
necessarily follows from the other.  Recall that the test for
self-organized criticality (Hypothesis One) looked at data at
yearly intervals with all states aggregated.  This approach is
notably different from the test for the Red Queen Hypothesis
(Hypothesis Three) where data were separated out by each
individual state.
     In conclusion, the sum of the two models is greater than the
parts.  One model from physics and one model from biology have
been shown to describe the history of subnational policy behavior
with such satisfaction as to make these models acceptable bases
for analyzing the subnational system.  Further, the self-
organized system of physics is consistent with the dynamics of
the Red Queen competition.  This consistency between the physics
and biology models is also an important source of verification
for applying these models to subnational policy-adoption
behavior.




          Bibilography of Literature Cited
                                
          Ambrosius, Margery Marzahn (1985) "Olson's Thesis and
          Economic Growth in the States: The Role of Interest
          Group Strength in State Economic Polymaking, 1969-1980"
          Paper presented at the annual meeting of the Midwest
          Political Science Association, Chicago.
          
          Ambrosius, Margery Marzahn (1989) "The Role of
          Occupational Interests in State Economic Development
          Policy-Making" Western Political Quarterly March:53-68.
          
          Arthur, W. Brian (1990) "Postive Feedbacks in the
          Economy" Scientific American Feb.:92-99.
          
          Arthur, W. Brian (1989) "Competing Technologies,
          Increasing Returns, and Lock-In by Historical Events"
          The Economic Journal 99(March):116-131.
          
          Ayala, F. J. 1976 Molecular Evolution Sunderland, MA:
          Sinauer Associates.
           
          Gould, Stephen Jay and Niles Eldredge (1977)
          "Punctuated Equilibria: The Tempo and Mode of Evolution
          Reconsidered" Paleobiology 3:115-151.
          
          Babcock, K. L. and R. M. Westervelt (1990) "Avalanches
          and Self-Organization in Cellular Magnetic-Domain
          Patterns" Physical Review Letters 64:2168-2171.
          
          Bak, Per and Chao Tang (1989) "Earthquakes as a Self-
          Organized Critical Phenomenon" Journal of Geophysical
          Research 94(B11):15635-15637.
          
          Bak, Per, Kan Chen and Michael Cruetz (1989) "Self-
          Organized Criticality in the 'Game of Life'" Nature
          342(6251):780-782.
          
          Bak, Peter and Kan Chen (1991) "Self-Organized
          Criticality" Scientific American Jan. 46-53.
          
          Bak, Per, Chao Tang and Kurt Wiesenfeld (1988) "Self-
          Organized Criticality" Physical Review A 38(1):364-374.
          
          Bak, Per, Chao Tang and Kurt Wiesenfeld (1988) "Self-
          Organized Criticality" Physical Review A 38:364-374.
          
          Bak, Per, Chao Tang and Kurt Wiesenfeld (1987) "Self-
          Organized Criticality: An Explanation of 1/f noise"
          Physical Review Letters 59:381-384.
          
          Bhattacharya, Rabi N. and Edward C. Waymire (1990)
          Stochastic Processes with Applications New York: John
          Wiley and Sons, Inc. 
          
          Boulding, Kenneth (1978) Ecodynamics: A New Theory of
          Societal Evolution Beverly Hills: Sage Publications.
          
          Boyce, Mark S. (1990) "The Red Queen Visits Sage Grouse
          Leks" American Zoologist 30:265-270.
          
          Brace, Paul (1985) "The Effects of Organized Interests
          on State Legislatures" Paper presented at the meeting
          of the American Political Science Association, New
          Orleans.
          
          Brace, Paul (1991) "The Changing Context of State
          Political Economy" Journal of Politics 53(2):297-317.
          
          Brace, Paul and Youssef Cohen (1989) "How Much Do
          Interest Groups Influence State Economic Growth?"
          American Political Science Review 83:1297-1301.
          
          Brace, Paul (1988) "Legislatures and Economic
          Performance" Paper presented at the annual meeting of
          the American Political Science Association, Washington,
          D. C.
          
          Braudel, *** (1980) "History and the Social Sciences:
          The Longue Dure" IN Ecrits sur l'Histoire, transl.
          Sara Mathews Chicago: ****.
          
          Brierly, Allen Bronson and Richard C. Feiock (1993)
          "Accounting for State Economic Growth: A Production
          Function Approach" Political Research Quarterly (Sept.)
          46(3):657-669.
          
          Carlson, J. M. and J. S. Langler (1989) "Properties of
          Earthquakes Generated by Fault Dynamics" Physical
          Review Letters 82:2632-2635. 
          
          Choi, Kwang (1979) "A Study of Comparative Rates of
          Economic Growth" Ph.D. diss., University of Maryland.
          
          Claggett, William, William Flanigan, and Nancy Zingale
          (1984) "Nationalization of the American Electorate"
          American Political Science Review 78:77-91.
          
          Cnudde, Charles F. and Donald J. McCrone (1969) "Party
          Competition and Welfare Policies in the American
          States" American Political Science Review 63:858-866.
          
          Conrad, Michael (1983) Adaptability: The Significance
          of Variability from Molecure to Ecosystem New York:
          Plenus Press.
          
          Darwin, Francis (ed.) (1892/1958) The Autobiography of
          Charles Darwin and Selected Letters New York: Dover
          Publications.
          
          Dawson, Richard E. and James A. Robinson (1963) "Inter-
          Party Competition, Economic Variables, and Welfare
          Policies in the American States" The Journal of
          Politics 63:265-289.
          
          Desmond, Adrian and James Moore (1992) Darwin: The Life
          of a Tormented Evolutionist New York: Warner Books.
          
          Dhar, Deepak (1990) "Self-Organized Critical State of
          Sandpile Automation Models" Physical Review Letters
          64:1613-1616.
          
          Dutta, P. and P. M. Horn (1981) "Low-Frequency
          Fluctuations in Solids: 1/f Noise" Reviews of Modern
          Physics 53(3):497-516.
          
          Dye, Thomas R. (1984) "Party and Policy in the States"
          Journal of Politics 46:1097-1116.
          
          Dye, Thomas R. (1980) "Taxing, Spending, and Economic
          Growth in American States" Journal of Politics 42:1085-
          1107.
          
          Dye, Thomas (1966) Politics, Economics, and The Public
          Chicago: Rand McNally.
          
          Easton, David (1953)  The Political System New York:
          Alfred A. Knopf.
          
          Eisinger, Peter (1988) The Rise of the Entrepreneurial
          State: State and Local Economic Development Policy in
          the United States Madison, WI: University of Wisconsin
          Press.
          
          Fogel, Robert (1989) Without Consent or Contract: The
          Rise and Fall of American Slavery Boston: ****.
          
          Foin, T. C., J. W. Valentine and F. J. Ayala (1975)
          "Extinction of Taxa and Van Valen's Law" Nature
          257:514-515.
          
          Fry, Brian R. and Richard F. Winters (1970) "The
          Politics of Redistribution" American Political Science
          Review 64:508-522.
          
          Garand, James C. (1992) "Changing Patterns of Relative
          State Economic Growth Over Time: Limitations on Cross-
          Sectional Tests of Olson's Thesis" Western Political
          Quarterly 45:469-83.
          
          Gemmill, Gary and Charles Smith (1985) "A Dissipative
          Structure of Organization Transformation" Human
          Relations 38:751-766.
          
          Gingerich, Philip D. (1993) "Quantification and
          Comparisioon of Evolutionary Rates" American Journal of
          Science 293-A:453-478.
          
          Glass, Leon and Michael C. Mackey (1988) From Clocks to
          Chaos: The Rhythms of Life Princeton, N.J.: Princeton
          University Press.
          
          Gleick, James (1987) Chaos: Making a New Science New
          York: Viking.
          
          Grady, Dennis O., 1987. State Economic Development
          Incentives: Why do States Compete? State and Local
          Government Review 1987:86-94.
          
          Gray, Virginia and David Lowery (1989) "Reply to 'How
          Much Do Interest Groups Influence State Economic
          Growth?'" American Political Science Review 83:1302-
          1308. 
          
          Gray, Virginia and David Lowery (1988) "Interest Group
          Politics and Economic Growth in the United States"
          American Political Science Review 82:109-31.
          
          Gregersen, Hal and Lee Sailer (1993) "Chaos Theory and
          its Implications for Social Science Research" Human
          Relations (July) 46(7):777-802.
          
          Gulick, Denny (1992) Encounters with Chaos New York:
          McGraw-Hill, Inc.
          
          Hall, Nina (ed.) (1991) Exploring Chaos: A Guide to the
          New Science of Disorder New York: W. W. Norton &
          Company.
          
          Hallam, A. () "The Red Queen Dethroned"  :12-13.
          
          Harris, Dorothy L. (1994) "A. B. Cambel: Applied Chaos
          Theory--A Paradigm for Complexity" Behavioral Science
          39:72-75.
          
          Harrison and Kanter (1978)
          
          Held, G. A., D. H. Solina Jr., D. T. Keane, W. J. Haag,
          P. M. Horn, and G. Grinstein (1990) "Experimental Study
          of Critical-Mass Fluctuations in an Evolving Sandpile"
          Physical Review Letters 65(9):1120-1123.
          
          Hempel, Carl (1965) "The Function of General Laws in
          History" IN Aspects of Scientific Explanation and Other
          Essays ***** New York: *****.
          
          Hendrick, Rebecca M. and James C. Garand (1991)
          "Variation in State Economic Growth: Decomposing State,
          Regional, and National Effects" The Journal of Politics
          (Nov.) 53(4):1093-1110.
          
          Hobbs, Jesse (1993) "Ex Post Facto Explanations" The
          Journal of Philosophy (March) 90(3):117-136.
          
          Hofferbert, Richard I. (1966) "The Relation between
          Public Policy and Some Structural and Environmental
          Variables in the American States" The American
          Political Science Review 60:73-82.
          
          Hofferbert, Richard I. (1964) "Classification of
          American State Party Systems" The Journal of Politics
          26:550-567.
          
          Hoffman, A. and F. A. Kitchell (1984) "Evolution in a
          Pianktic System: A Paleobiologic Test of Models of
          Multispecies Evolution" Paleobiology 
          
          Holden, Arun V. (1986) Chaos Princeton, NJ: Princeton
          University Press.
          
          Hwa, Terence and Mehran Kardar (1989) "Dissipative
          Transport in Open Systems: An Investigation of Self-
          Organized Criticality" Physical Review Letters
          62(16):1813-1816.
          
          Iwai, K. (1981b) "Schumpeterian Dynamics II:
          Technological Progress, Firm Growth and 'Economic
          Selection'" Discussion Paper Number CFDP #603, Cowles
          Foundation, New Haven, August.
          
          Iwai, K. (1981a) "Schumpeterian Dynamics I: An
          Evolutionary Model of Innovation and Imitation"
          Discussion paper Number CFDP #602, Cowles Foundation,
          New Haven, August.
          
          Iwai, K. (1984a) "Schumpeterian Dynamics I: An
          Evolutionary Model of Innovation and Imitation" Journal
          of Economic Behavior and Organization 5(2):159-190.
          
          Iwai, K. (1984b) "Schumpeterian Dynamics II:
          Technological Progress, Firm Growth and 'Economic
          Selection'" Journal of Economic Behavior and
          Organization 5(3-4):321-351.
          
          Jacob (1986)
          
          Jaeger, H. M., Chu-heng Liu, and Sidney R. Nagel (1989)
          "Relaxation at the Angle of Repose" Physical Review
          Letters 62:40-43.
          
          Jantsch, E. (1980) The Self-Organizing Universe New
          York: George Braziller Publishers.
          
          Jennings, Edward T., Jr. (1979) "Competiton,
          Constituencies, and Welfare Policies in American
          States" American Political Science Review 73:414-429.
          
          Jones, Bryan (1990) "Public Policies and Economic
          Growth in the American States" The Journal of Politics
          52:219-33.
          
          Kadanoff, Leo P., Sidney R. Nagel, Lei Wu and Su-min
          Zhou (1989) "Scaling and Universatality in Avalanches"
          Physical Review A 39:6524-6537.
          
          Kagan, Y. Y and L. Knopoff (1987) "Statistical Short-
          term Earthquake Prediction" Science 236:1563-1567.
          
          Kauffman, Stuart A. (1993) The Origins of Order: Self-
          Organization and Selection in Evolution New York:
          Oxford University Press.
          
          Kaye, Brian (1993) Chaos and Complexity: Discovering
          the Surprising Patterns of Science and Technology New
          York: VCH.
          
          Kellert, Stephen H. (1993) In the Wake of Chaos:
          Unpredictable Order in Dynamical Systems Chicago: The
          University of Chicago Press.
          
          Key, Jr., V. O. (1951) Southern Politics in State and
          Nation New York: Alfred A. Knopf.
          
          Krasner, Saul (ed.) (1990) The Ubiquity of Chaos
          Washington: American Association for the Advancement of
          Science.
          
          Lasswell, Harold D. (1951) "The Policy Orientation" IN
          The Policy Sciences Daniel Lerner and Harold Lasswell
          (eds.) Stanford, Calif: Stanford University Press.
          
          Lawlor, Lawrence R. and John Maynard Smith (1976) "The
          Coevolution and Stability of Competing Species"
          American Naturalist 110:79-99.
          
          Lawrence, A., M. G. Watson, K. A. Pounds and M. Elvis
          (1987) "Low-frequency divergent X-ray Variability in
          the Seyfert Galaxy NGC4051" Nature 325:694-696.
          
          Leifer, Richard (1989) "Understanding Organizational
          Transofrmation Using a Dissipative Structure Model"
          Human Relations 42:899-916.
          
          Lewin, Roger (1992) Complexity: Life at the Edge of
          Chaos New York: MacMillan.
          
          Lowi (1963)
          
          Loye, David and Riane Eisler (1987) "Chaos and
          Transformation: Implications of Nonequilibrium Theory
          for Social Science and Society" Behavioral Science
          32:53-66. 
          
          Mandelbrot, Benoit B. (1982) The Fractal Geometry of
          Nature New York: W. H. Freeman and Company.
          
          May, Robert M. (1975) "Deterministic Models with
          Chaotic Dynamics" Nature 256:165-166.
          
          May, Robert M. (1974) "Biological Populations with
          Nonoverlaping Generations: Stable Points, Stable
          Cycles, and Chaos" Science 186:645-647.
          
          Maynard Smith, John (1976b) "What Determines the Rate
          of Evolution?" The American Naturalist 110:331-338.
          
          Maynard Smith, John (1976a) "A Comment on the Red
          Queen" The American Naturalist 110:325-330.
          
          McCloskey, Donald N. (1991) "History, Differential
          Equations and the Problem of Narration" History and
          Theory (Jan.) 30(1):21-36.
          
          McCune, A. R. (1982) "On the Fallacy of Constant
          Extinction Rates" Evolution 39:610-614.
          
          McHardy, Ian and Bozena Czerny (1987) "Fractal X-ray
          Time Variability and Spectral Invariance of the Seyfert
          Galaxy NGC5506" Nature 325:696-698.
          
          Metcalfe, J. Les (1974) "Systems Models, Economic
          Models and the Causal Texture of Organizational
          Environments: An Approach to Macro-Organization Theory"
          Human Relations 27(7):639-663.
          
          Mueller, Dennis C. (ed.) (1983) The Political Economy
          of Growth New Haven, CT: Yale University Press.
          
          Musha, T. and H. Higuchi (1977) in Proceedings of the
          Symposium on 1/f Fluctuations (Institute of Electric
          Engineers, Tokyo, Japan)
          
          Nelson, R. R., S. G. Winter, Jr., and H. L. Schuette
          (1976) "Technical Change in an Evolutionary Model"
          Quarterly Journal of Economics 90(1):90-118.
          
          Nelson, R. R. and Winter, Jr., S. G. (1982) An
          Evolutionary Theory of Economic Change Cambridge, MA:
          Belknap/Harvard University Press.
          
          Nelson, R. R. and Winter, Jr., S. G. (1973) "Towareds
          an Evolutionary Theory of Economic Capabilities"
          American Economic Review (Papers and Proceedings)
          63(2):440-409.
          
          Nicolis, Grgoire and Ilya Prigogine (1989) Exploring
          Complexity: An Introduction New York: W. H. Freeman and
          Company.
          
          Olson, Mancur (1982) The Rise and Decline of Nations:
          Economic Growth, Stagflation, and Social Rigidities New
          Haven, CT: Yale University Press.
          
          Peitgen, Heniz-Otto, Hartmut J
rgens, and Dietmar Saupe
          (1992) Chaos and Fractals: New Frontiers of Science New
          York: Springer-Verlag.
          
          Peretz (1986)
          
          Petersen, Karl E. (1977) Brownian Motion, Hardy Spaces
          and Bounded Mean Oscillation New York: Cambridge
          University Press.
          
          Prigogine, Ilya (1980) From Being to Becoming: Time and
          Complexity in the Physical Sciences San Francisco: W.
          H. Freeman and Company.
          
          Prigogine, Ilya and Isabelle Stengers (1984) Order Out
          of Chaos: Man's New Dialogue with Nature New York:
          Bantam Books.
          
          Reisch, George A. (1991) "Chaos, History, and
          Narrative" History and Theory (Jan.) 30(1):1-20.
          
          Reynolds, Craig W. (1987) "Flocks, Herds, and Schools:
          A Distributed Behavioral Model" Computer Graphics
          21(4):25-38.
          
          Rogers, Everett M. (1962) Diffusion of Innovations New
          York: Free Press of Glencoe.
          
          Ross, Sheldon M. (1993) Introduction to Probability
          Models (fifth edition) New York: Academic Press, Inc.
          
          Savit, Robert (1988) "When Random is Not Random: An
          Introduction to Chaos in Market Places" Journal of
          Future Markets 8:271-290.
          
          Schlesinger (1966)
          
          Schroeder, Manfred (1991) Fractals, Chaos, Power Laws:
          Minutes from an Infinite Paradise New York: W. H.
          Freeman and Company.
          
          Schubert, Glendon and Charles Press (1964) "Measuring
          Malapportionment" The American Political Science Review
          58:302-327.
          
          Schuette, H. L. and S. G. Winter, Jr. (1975) "SSIR2: A
          Computer Simulation Model Based on Satisficing,
          Selection and the Innovating Remnant" Discussion Paper
          Number 51, Institute of Public Policy Studies,
          University of Michigan (Ann Arbor).
          
          Schuette, H. L. (1980) "The Role of Firm Financial
          Rules and a Simple Capital Market in an Evolutionary
          Model of Industry Growth" Doctoral Dissertation,
          University of Michigan (Ann Arbor).
          
          Sharkansky, Ira (1967) "Economic and Political
          Gorrelates of State Government Exenditures: General
          Gendencies and Deviant Cases" The Journal of Politics
          29:173-192.
          
          Sornette, A. and D. Sornette (1989) "Self-Organized
          Criticality and Earthquakes" Europhysics Letters
          9(3):197-202.
          
          Squire, J. C. et al. (1931) If: Or, History Rewritten
          New York: ****.
          
          Stenseth, Niles and John Maynard Smith (1984)
          "Coevolution in Ecosystems: Red Queen Evolution or
          Stasis?" *****.
          
          Stewart, Ian (1989) Does God Play Dice? The Mathematics
          of Chaos Cambridge, Mass.: Blackwell.
          
          Tang, Chao and Per Bak (1988a) "Mean Field Theory of
          Self-Organized Critical Phenomena" Journal of
          Statistical Physics 51(5/6):797-802
          
          Tang, Chao and Per Bak (1988b) "Critical Exponents and
          Scaling Relations for Self-Organized Critical
          Phenomena" Physical Review Letters 60:2347-2350.
          
          Tichy, Noel and David N. Ulrich (1984) "SMR Forum: The
          Leadership Challenge--A Call for the Transformational
          Leader" Sloan Management Review Fall 59-68.
          
          Toffler, Alvin (1984) "Science and Change; Forward to
          Order out of Chaos: Man's New Dialogue with Nature New
          York: Bantam Books.
          
          Van Valen, Leigh (1974) "Two Modes of Evolution" Nature
          252:298-300.
          
          Van Valen, Leigh (1976) "The Red Queen Lives" Nature
          260:575.
          
          Van Valen, Leigh (1975) "Extinction of Taxa and Van
          Valen's Law- Reply" Nature 257:515-516.
          
          Van Valen, Leigh (1973) "A New Evolutionary Law"
          Evolutionary Theory 1:1-30.
          
          Vertz, Laura, John Frendreis, and James Gibson (1987)
          "Nationalization of the Electorate in the United
          States" American Political Science Review 81:961-966.
          
          von Bertalanffy, Ludwig (1988) General System Theory:
          Foundations, Development, Applications (revised
          edition) New York: George Braziller.
          
          Vrba, Elisabeth S. (1993) "Turnover-Pulses, The Red
          Queen, and Related Topics" American Journal of Science
          293-A:418-452.
          
          Vrba, Elisabeth S. (1985) "Environment and Evolution:
          Alternative Causes of the Temporal Distribution of
          Evolutionary Events" South African Journal of Science
          81:229-236.
          
          Vrba, Elisabeth S. (1982) "Southern African Hominid-
          Associated Assemblages: Biostratigraphy, Chronology and
          Evolutionary Pulses" Address to 1982 Congres
          International de Paleontologie Humaine, Unione
          Internationale des Sciences Prehistoriques et
          Protohistoriques; Nice, France.
          
          Waldrop, M. Mitchell (1992) Complexity: The Emerging
          Science at the Edge of Order and Chaos New York: A
          Touchstone Book.
          
          Walker, Jack L. (1969) "The Diffusion of Innovations
          Among the American States" The American Political
          Science Review 63:880-899.
          
          Winter, Jr., S. G. (1987*5?) "Natural Selection and
          Evolution" IN J. Eatwell, M. Milgate, and P. Newman The
          New Palgrave: A Dictionary of Economics vols. 1-4,
          London: Macmillan, pp. 614-617.
          
          Winter, Jr., S. G. (1964) "Economic 'Natural Selection'
          and the Theory of the Firm" Yale Economic Essays
          4(1):225-272.
          
          Winter, Jr., S. G. (1971) "Satisficing, Selection, and
          the Innovating Remnant" Quarterly Journal of Economics
          85(2):237-261.
          
          Winter, Jr., S. G. and U. G. Rothblum (1982) "Behavior
          of Market Shares for a Stochastic Growth Model" Journal
          of Economic Theory 36(2):352-366.
          
          Winter, Jr., S. G. (1975) "Optimization and Evolution
          in the Theory of the Firm" IN R. H. Day and T. Groves
          (eds.) Adaptive Economic Models New York: Academic
          Press, pp. 73-118.
          
          Alchian, A. A. (1950) "Uncertainty, Evolution and
          Economic Theory" Journal of Political Economy
          58(3):211-222.
          
          Alchian, A. A. (1953) "Comment: Biological Analogies in
          the Theory of the Firm" American Economic Review
          43(3):600-603.
          
          Axelrod, R. M. (1986) "An Evolutionary Approach to
          Norms" American Political Science Review 80(4):1095-
          1111.
          
          Axelrod, R. M. (1984) The Evolution of Cooperation New
          York: Basic Books.
          
          Axelrod, R. M. (1980a) "Effective Choice in the
          Prisoner's Dilemma" Journal of Conflict Resolution
          24:3-35.
          
          Axelrod, R. M. (1980b) "More Effective Choice in the
          Prisoner's Dilemma" Journal of Conflict Resolution
          24:379-403.
          
          Axelrod, R. M. (1991) The Evolution of Cooperation
          Harmondsworth: Penguin.
          
          Axelrod, R. M. (1987) "The Evolution of Strategies in
          the Iterated Prisoner's Dilemma" IN L. D. Davis (ed.)
          Genetic Algorithms and Simulated Annealing London:
          Pitman, pp. 32-41.
          
          Axelrod, R. M. and D. Dion (1988) "The Further
          Evolution of Cooperation" Science 242:1385-1390.
          
          Axelrod, R. M. and W. D. Hamilton (1981) "The Evolution
          of Cooperation" Science 211:1390-1396.
          
          Enke, S. (1951) "On Maximising Profits: A Distinction
          between Chamberlin and Robinson" American Economic
          Review 41(3):566-578.
          
          Enke, S. (1953) "Comment: Biological Analogies in the
          Theory of the Firm" American Economic Review 43(3):603.
          
          Hayek, F. A. (1967) "Notes on the Evolution of Systems
          of Rules of Conduct" IN Hayek, F. A. Studies in
          Philosophy, Politics and Economics London:Routledge and
          Kegan Paul, pp. 66-81.
          
          McCain, R. A. (1992) A Framework for Cognitive
          Economics Praeger.
          
          Meier, A. and S. Haury (1990) "A Cognitive-Evolutionary
          Theory of Economic Policy" IN K. Dopfer and K.-F.
          Raible (eds.) The Evolution of Economic Systems: Essays
          in Honour of Ota Sik London: Macmillan.
          
          Penrose, E. T. (1952) "Biological Analogies in the
          Theory of the Firm" American Economic Review 42(4):804-
          819.
          
          Penrose, E. T. (1953) "Rejoinder: Biological Analogies
          in the Theory of the Firm" American Economic Review
          43(3):603-607.
          
          Penrose, E. T. (1959) The Theory of the Growth of the
          Firm Oxford: Blackwell.
          
          Peyton Young, H. (1991) "The Evolution of Conventions"
          Working Paper 91-10-043, Economics Research Program,
          Santa Fe Institute, New Mexico.
          
          Peyton Young, H. (1992) "An Evolutionary Model of
          Bargaining" Working Paper 92-02-009, Economics Research
          Program, Santa Fe Institute, New Mexico.
          
          Jantsch, E. (1975) Design for Evolution New York:
          George Braziller Publishers.
          
          Jantsch, E. (1981) The Evolutionary Vision Colorado:
          Westview Press.