Weingast B.R., Wittman D. The Oxford Handbook of Political Economy
Подождите немного. Документ загружается.
12 introduction: the reach of political economy
democracies differs greatly from this ideal. Voters appear to be notoriously unin-
formed (and, indeed, have little incentive to become informed). Some voters base
theirchoiceofcandidatesolelyonpartylabel,whileothervotersrelyonbiased
sources of information, including information provided by pressure groups.
Does this apparent lack of information imply that democracy will fall far from
its ideal? Possibly not, if the lack of information is more apparent than real. In
the following pages, we show how voters can make logical inferences so that their
behavior is similar to perfectly informed voters.
We start with an easy example to illustrate how information revelation arises.
A number of articles study the endogenous timing of elections in parliamentary
systems.
10
Because it has access to information, the ruling party is able to forecast
future economic performance and other events that are likely to impact on voters’
welfare. This information is not likely to be available to the voters. The party in power
has an incentive to call an election when it is at the height of its popularity.
However, this decision-theoretic analysis does not consider the voter response to
an early election call. Voters can infer from an early election call that the ruling party
expects to do worse in the future. Voters can therefore infer that there is likely to be
badnewsinthefuture.
11
The ruling government realizes that voters will act this way.
As a result, governments are less likely to call early elections than they would other-
wise be, and when they do, voters will take this information into account and be less
positively inclined towards the government. Smith (2003, 2004) provides empirical
evidence in support of this argument. Polls taken after the announcement of an early
election show a decline from polls taken before the announcement. Incorporating the
voters’ response made obsolete much of the earlier research on endogenous timing of
elections that did not consider the possibility that voters could make inferences.
Let us now consider another area where earlier research assumed mechanical,
uninformed voters, but more recent research assumes uninformed but rational voters,
oftenwithstarklydiffering results. Starting with Ben Zion and Eytan (1974)and
continuing on into the recent past (see Baron 1994; Grossman and Helpman 1996),
an extensive literature has assumed that the more money a candidate spends on
advertising, the more votes the candidate receives from uninformed voters. Sources
of money tend to come from interests on the extremes of the political distribution. To
get contributions that pay for such advertising, candidates move their policies away
from the median voter toward a pressure group’s most preferred position.
Let us look at the Grossman and Helpman model in greater detail.
12
Grossman and
Helpman assume the following: voters, candidates, and pressure groups are arrayed
along a one-dimensional issue space. Each voter has a most preferred position with a
concave utility function over policy; this means that voters are risk averse.
13
There are
¹⁰ See for example, Cargill and Hutchison 1991 and the long list of citations found in Smith 2003,n.6.
¹¹ Here, we ignore other reasons for calling an early election, in particular the desire of the ruling
party in a coalition government to strengthen its hand.
¹² For heuristic purposes, we simplify their model.
¹³ Risk aversion means that voters prefer a sure thing over a lottery having the same expected value as
the sure thing.
barry r. weingast & donald a. wittman 13
two types of voters: informed voters who know the positions of the candidates and
uninformed voters who have no knowledge of the candidates’ or pressure group’s
positions. Informed voters vote for the candidate closest to the voter’s most preferred
position. Uninformed voters respond only to political advertising—the more money
spent on advertising by one of the candidates, the greater the percentage of unin-
formed voters voting for the candidate.
Each candidate wants to maximize the percentage of votes that he or she receives.
There is one pressure group (say, on the extreme right). The pressure group is willing
to donate money to one of the candidates if the candidate moves right from the
median voter.
The election proceeds as follows:
1. The pressure group makes a one-time take-it-or-leave-it offer to one of the
candidates. If the candidate agrees to move right of the median informed voter,
then the pressure group provides funds to the candidate for political advertising.
If the agreement is accepted, it is binding on both sides.
2. The candidate receiving the offer decides whether to accept or reject it.
If the candidate accepts the offer, then the other candidate knows the position
of the candidate accepting the offer. The other candidate will then choose a
position between the candidate and the median informed voter to capture as
many informed voters as possible.
If the candidate rejects the offer, then the pressure group is out of the picture.
Per the standard Downsian (1957) model, both candidates will then choose to
be at the median of the informed voters.
3. The positions of the candidates are then made public to the informed voters.
The candidate who received the donation then advertises.
4.Thevoterschoose.
Given the set-up of the model, it is not hard to see that the candidate will be willing
to move right from the median of the informed voters as long as the advertising
from the campaign funds sufficiently increases the number of uninformed voters to
compensate for the loss of informed voters caused by the movement to the right and
away from the median informed voter.
The model seems to imply that pressure groups are likely to undermine the political
process. But is it rational for uninformed voters to act in the way postulated? Let us
consider the model more carefully.
In the above model, the candidates, pressure group, and informed voters are all
rational, but not the uninformed voters. As already mentioned, uninformed voters
vote mechanically. But being uninformed does not mean being irrational. Suppose
instead that the uninformed do not vote mechanically but can make logical infer-
ences. We consider two variants with different characterizations of uninformed voter
behavior.
First, let us continue to assume that the uninformed voters know neither the
positions of the candidates nor the position of the pressure group (the pressure group
being equally likely to be on the left or the right). Campaign advertising is, by its very
14 introduction: the reach of political economy
nature, public so that an ordinary person can infer which candidate received the most
contributions by observing which candidate has the most political advertising. The
uninformed voters can simply watch television and passively observe the candidate
who has the most advertisements. Given the logic of the model, the uninformed
can infer that the candidate doing the advertising is further away from the median
informed voter than the candidate not doing the advertising.
Given our assumption that the uninformed voter does not even know whether the
pressure group is on the left or the right, the uninformed voter faces a greater risk
from the candidate who is doing the advertising. Both candidates will on average be
at the median informed voter’s most preferred position, but the candidate receiving
the campaign funds will be more extreme. Thus the risk-averse uninformed voter
should vote for the candidate not doing the advertising! The rational voter does not
act like the mechanical voter in this case. Of course, if this behavior characterizes
uninformed voters, then the candidate will not accept campaign donations from the
pressure group in the first place.
Now suppose that the uninformed know something. For example, they may know
that the National Rifle Association supports one of the candidates, and as a conse-
quence these voters can infer that the candidate receiving the funds is closer than the
median informed voter is to the position of the NRA. More generally, the uninformed
voter may know whether the pressure group is on the right. If the uninformed voter
also knows where he or she stands relative to the median voter, the uninformed voter
totheleftofthemedianvotercaninferthatheorsheshouldvotefortheother
candidate, while those uninformed voters to the right will be inclined to vote for the
candidate receiving funds from the right-wing pressure group.
Consider two cases: if uninformed voters tend to be to the right of the median
informed voter, then the candidate may accept funds from the right-wing pressure
group and even advertise this to be the case. This occurs when the candidate gets
sufficiently more votes from the uninformed voters on the right than she loses from
the uninformed voters on the left to make up for the reduced vote share from
informed voters. Alternatively, if there are more uninformed voters to the left of the
median informed voter, the candidate would lose if she accepted the deal from the
pressure group. Hence she would not do so in the first place.
In this version of the model, pressure group contributions help the uninformed
voters. If the mass of uninformed voters is to the right of the median of the in-
formed voters (and hence, the overall median is to the right of the median of the
informed voters), then one candidate will accept the funds and the effect of cam-
paign donations will be to move the candidate to the right from the median of the
informed voters. On the other hand, if more uninformed voters lie to the left of the
median, neither candidate will accept funds from a pressure group on the right.
14
In
short, pressure groups aid the political process, rather than undermine it!
We have just modeled the case where some voters are uninformed about the candi-
dates’ positions. Another possibility is that voters are informed about the candidates’
¹⁴ The exact result requires more technical specification and can be found in Wittman 2005a.
barry r. weingast & donald a. wittman 15
positions but not about their relative quality. Again, the voters are rational and the
pressure group has private information (in this case, about the relative quality of
the candidates). A number of recent papers consider this case but employ differing
subsidiary assumptions: advertising has content (see Coate 2004; Wittman forth-
coming); advertising has no content, but expenditures on advertising signal infor-
mation (Prat 2002, this volume); pressure groups make the offers (Coate, Prat),
candidates make the offers; there is one pressure group, there are multiple pressure
groups; the candidates are only interested in winning, and the candidates have policy
goals. These various modeling efforts do not all come to the same positive conclusion
as the previous paragraph. In general, the results depend on whether the value of
the revealed information is outweighed by the loss from inferior candidate positions
when the candidates compete for pressure group funds. In turn, this balance depends
to a great degree on the number of pressure groups and whether it is the candidates or
the pressure groups that make the offer. All these various modeling efforts take into
account that information valuable to uniformed voters is revealed by the pressure
group’s donation or endorsement and all of them assume rationality of the voters.
This is the key methodological advance—how voters can incorporate information
that others might want to distort or hide (see Prat, this volume).
We have shown how uninformed voters can make inferences from behavior and
thereby become more informed. Because all of this is embedded in a game, all
other players take this behavior and information into account when they make their
decisions; and of course the uninformed take the other players’ strategies into account
when they make their own inferences.
15
The final example for this section considers aggregation of information in the
context of voting. Suppose a set of voters face a decision about how much money
to spend. To gain intuition, we begin with an exceedingly simple example. Suppose
that there are five voters with identical preferences: three have unbiased estimates of
the correct action to take, while two are fully informed. The voters know whether they
are informed or not. The uninformed know that there are informed voters, but not
how many. Suppose further that the correct action is to spend $7 million and that,
with equal probability, the uninformed players receive a signal that it should be 5, 7,
or 9 million dollars. Assuming that the voters cannot communicate with each other,
howlikelyisitthatthemajorityruledecisionisnot7? The answer is zero if the voters
are rational: all the uninformed voters will rationally abstain. By doing so, they know
that only informed voters will participate and that these informed voters will make
the correct decision.
This example illustrates two important but related issues. First, the more informed
people will choose to vote (here, at least, the argument does not go against conven-
tional wisdom). Second, the potential voter asks: given that he will be pivotal, should
¹⁵ There are other ways in which voters can be informed despite an apparent lack of information.
Parties create brand names so that party labels are in fact informative about a candidate’s position (Cox
and McCubbins 1993, 2005). Relying on biased information can be rational for voters who have strong
priors in favor of one of the parties (Calvert 1985). And, uninformed voters can learn from polls of
informed voters (McKelvey and Ordeshook 1986).
16 introduction: the reach of political economy
he vote, and if he votes, how should he vote. In other words, the decision to vote and
how to vote does not just depend on whether the person will be pivotal, but also on
the preferences and information structure of all of the voters. Our understanding of
the problem is no longer in terms of decision theory but of game theory.
To illustrate this idea in terms of a more complicated, but more realistic model,
assume that there are three voters (or three groups of voters), labeled V, V
i
,andV
u
;
and two states of the world, labeled 1 and 2. Assume that voter V votes for candidate
D regardless of the state of the world. The second voter is independent but informed
and will be labeled V
i
. This voter knows the state of the world and votes for D when
the state of the world is 1 and votes for R when the state of the world is 2:givenV
i
’s
preferences, D makes a better president if the state of the world is 1 while R makes a
better president if the state of the world is 2. The third voter, V
u
,isalsoindependent
(with the same preference structure as V
i
), but is uninformed. However, V
u
knows
the preferences and information sets of the other two voters.
How should V
u
vote if the probability of state 1 (where D is V
u
’s preferred can-
didate) is more likely than the probability of state 2? The decision-theoretic model,
where V
u
’s vote is based on the mostly likely state of the world (state 1), suggests that
V
u
vote for D. But the game-theoretic pivot model argues that V
u
should vote for R.
The reasoning is as follows. If the state of the world is 1, then both voter V and voter
V
i
vote for D, and D will win regardless of V
u
’s vote. When the state of the world is
2, then the other two voters will split their vote and V
u
will be pivotal. Under such
circumstances, V
u
shouldvoteforRsincesheprefersRtoDinstate2.SoV
u
always
votes for R.
16
Behaving in this way allows the informed voter, V
i
, whose preferences
are similar to V
u
’s, to be pivotal in all circumstances. This behavior results in better
outcomes for V
u
than those suggested by the decision-theoretic perspective. Because
the latter tells V
u
always to vote for D, V
u
votes for D even in the state 2 when she
prefers R.
These two examples show that uninformed voters can make inferences about how
to behave that make them better off, even when they remain ignorant of critical
aspects of the election. Now this particular example requires V
u
to know a lot about
the other voters, but the conceptual apparatus can be incorporated into other models
where the information requirements are not so high. To get back to our earlier dis-
cussion of pressure groups where some voters are uninformed about the candidates’
positions, Wittman (2005) shows that the uninformed voters to the right (left) of the
median voter should always employ the following rule of thumb: vote for (against) the
candidate endorsed by the right-wing pressure group.
17
Sometimes this could result
in some of the uninformed voters on the right voting for the wrong candidate, and at
other times this could result in some of the uninformed voters on the left voting for
the wrong candidate. However, even if all the mistaken votes for one candidate were
reversed this would not change the outcome. So fully rational but uninformed voters
consider the effect of their behavior when pivotal even if their likelihood of being
¹⁶ For a more extended discussion see Fedderson and Pesendorfer 1996, 1999.
¹⁷ The actual model employs additional assumptions that assure that certain pathologies do not arise.
barry r. weingast & donald a. wittman 17
pivotal is small. Indeed, in this example, by their rule of thumb, the uninformed make
the informed median over all voters the pivot.
To summarize: democratic theory has long held that ignorant voters harm the
operation of democracy. The force of this section is to demonstrate that uninformed
voters and uninformed actors more generally can make inferences based on the
behavior of others, the structure of their strategic situation, and signals received from
other actors. These inferences make uninformed voters better off than predicted by
decision-theoretic models; and they improve the workings of democracy more than
predicted by traditional democratic theory.
18
We believe that in the next decade the
aggregation and revelation of information will continue to be a very important mode
of research and that it will continue to overturn received wisdom (see Moulin, this
volume; and Ledyard, this volume; for further examples).
3 Evolutionary Models of Human
and Political Behavior
.............................................................................
Both economics and evolutionary models of human behavior employ the concepts of
survival and equilibrium (Alchian 1950). Nonetheless, the implications of economic
and biological models at times conflict. Although people who are capable of achieving
their goals because of either their physical or mental prowess are more likely to
survive and produce offspring, it is not clear that fitness would accrue to those who
maximized utility and were happier. Furthermore, at times evolutionary fitness may
be gained by being less rational; for example, the emotional may serve as a useful
commitment device (see Hirshleifer 2001).
Humans are pre-eminently social animals. Political structures are one kind of social
structure, and such structures need to be compatible for better or worse with the
biology of human behavior. Are people naturally xenophobic, vengeful, or generally
limited in their capacity for empathy? One can generate all kinds of hypotheses about
human behavior. But economics and/or evolutionary biology demand that the hy-
pothesized behavior survives in a competitive equilibrium. The principle of survival
in equilibrium imposes discipline on modeling efforts because not all hypotheses
satisfy this criterion.
Models of pure self-regarding preferences have generated considerable insight into
the political process—as this Handbook attests—yet such an assumption is not req-
uisite for rational behavior. People may be other regarding in that they care about
their children or feel altruistic or vengeful towards others.
19
How other-regarding
¹⁸ See Ansolabehere (this volume) who makes this point.
¹⁹ The phrase “other regarding” gets around the problem that altruistic behavior may make the
person feel good and therefore altruism could be termed selfish behavior.
18 introduction: the reach of political economy
behavior survives in equilibrium is a major research question that still has not been
fully answered.
Let us start with an easy question. Why are human parents altruistic to their
children? Infants and small children need care in order to survive. Parental altruism
helps to ensure the genetic transmission. But genetic relatedness rapidly approaches
zero as the population increases in size. So this simple explanation for altruism falters
when we want to extend it to the population as a whole.
A significant number of researchers seek to understand the role of vengeance.
The phenomenon of suicide bombers inspires some of this interest—being a suicide
bomber hardly appears to improve genetic fitness. Further interest in vengeance is
inspired by experiments demonstrating that the standard income-maximizing model
does not work well in certain situations. For example, consider the ultimatum game
in which person A is given a certain amount of money (say ten dollars); A then offers
a share of this money to B; B then either rejects or accepts the offer. If B rejects the
offer, neither gets any of the money and the game is over. A theory that is based on
humans being purely self-regarding predicts that A should offer B a trivial amount,
say one cent. Because one cent is better than nothing, B is better off accepting the
offer than rejecting it. Experiments consistently reveal that B subjects often reject low
offers even though this hurts them financially. Further, experiments also reveal that
Asubjectsoftenoffer significant amounts to B to forestall such a rejection.
20
This vengeful behavior by B is contrary to income maximization. Thus, the key
intellectual puzzle to resolve is how vengeful behavior can be evolutionarily stable.
Scholars provide two types of answers. One is that a reputation for vengeful be-
havior may enhance fitness because others may avoid provoking revenge by avoiding
doing harm to the vengeful person in the first place. Following this intuition, some
evolutionary models show that, under certain circumstances, two types of people,
vengeful and non-vengeful, can survive in equilibrium (Friedman and Singh 2005).
21
Here we will concentrate on the second approach—the co-evolution of memes
(social constructs) and genes—because that is more relevant to our understanding of
collective choice and social cooperation. Humans are more social than their ancestors,
and many argue that this sociability evolved along with the social institutions that
made such sociability result in greater reproductive success. Consider the following
thought experiment. If chimpanzees had language (which in itself enhances sociabil-
ity) and could do calculus, would chimpanzee society look like human society if they
were able to observe our customs? The co-evolution argument says no. Shame, guilt,
the ability to be empathetic or vengeful, and certain conceptual possibilities that make
us human would all be much more circumscribed in chimpanzees, which themselves
²⁰ The explanation for the rejection is that the person dividing the money has not been “fair.” For a
further discussion of fairness in experiments and the implications for political behavior see Palfrey (this
volume).
²¹ To get this result, Friedman and Singh develop a new equilibrium concept—evolutionary perfect
Bayesian equilibrium. Their paper, as is the case for much of the research on the evolutionary stability of
vengeance, employs sophisticated mathematical modeling. This illustrates another theme of our
chapter—that political economy, unlike other intellectual approaches to political science, emphasizes
logical rigor, which often requires considerable mathematics.
barry r. weingast & donald a. wittman 19
show more of these qualities than marsupials. Without pro-social emotions, all hu-
mans (rather than just a few) might be sociopaths, and human society as we know
it might not exist despite the institutions of contract, government law enforcement,
and reputation.
22
Groups that overcome prisoner’s dilemmas (and other social dilemmas) are likely
to be more productive in gathering food and more successful in warfare against other
groups. In turn, this leads to greater reproductive success. The central question for
evolutionary models is how, if at all, evolutionary pressure keeps individual shirking
in check. It seems, for example, that a person who is slightly less brave in battle is
more likely to survive and have children than his braver compatriots. Bravery at once
increases the risk for the brave while making it more likely that the less brave survive.
If bravery/cowardice is genetic, how is a downward spiral of cowardice prevented?
The answer proceeds along the following lines: if the individuals are punished for
shirking (in this case, being cowardly), this will keep them in line. But, because en-
gaging in punishment is costly (possibly resulting in the would-be punisher’s death),
who will do the punishing? The evolutionary approach suggests that punishment, a
kind of vengeance, will be a successful strategy for the punisher if he gains even a mild
fitness advantage (status, more females, etc.). This is because, in equilibrium, the cost
to the punisher is relatively small since punishment does not have to be meted out
very often. Punishment need not be carried out frequently to be effective. It is the
threat that is important. To the degree that shirkers by being punished (possibly by
being banished from the tribe) become less fit, the need to engage in punishment
decreases even more as there are fewer shirkers. And given that those who punish are
more aligned with the interests of the society and therefore may be more likely to
survive, there may be enough potential punishers so that the need for any individual
to bear the costs of punishment is reduced still further (which of course means that
the benefits received will also be reduced). If altruism and vengeance are gene based
rather than meme based, there may have been a co-evolution of memes and genes.
Over the eons, human society may have encouraged pro-social genetically based
emotions.
23
The force of this argument is that pro-social emotions bypass the cognitive opti-
mizing process that is at the core of rational economic man. This cognitive difference
implies that at times we should observe profound differences between the evolu-
tionary model and the economic model. Under certain circumstances, seemingly
irrational behavior, such as vengeance or shame, may be evolutionarily stable even if
it runs counter to utility maximization. Moreover the relatively slow genetic evolution
in comparison with meme evolution (especially in the last 100 years) yields a further
²² For further discussion along these lines see Bowles and Gintis, this volume; Friedman and Singh
2000;Boydetal.2003;Gintisetal.2005.
²³ This just gives the flavor of the argument. Once again, it is worthwhile to emphasize that the
research summarized here employs very carefully specified models. The challenge for researchers in the
field is to characterize a situation where vengeance survives, but does not become so intense that it
undermines social relations. At the same time, the researcher must account for the possibility that
non-vengeful types may want to mimic vengeful types. Finally, the researcher must mix the memes and
genes so that they are in a stable equilibrium.
20 introduction: the reach of political economy
conclusion: it is quite possible that some of the pro-social emotions whose genetic
basis evolved over the last 100,000 or more years are maladapted to the modern
world.
At present the evolutionary study of genes and memes has produced very tentative
results. Human behavior is part mammalian (possibly even reptilian), part primate,
and part hominid. Although some have argued that much of human psychology
developed in the savannah, it is not clear what part of human psychology developed
then or earlier or, to a lesser degree, later. Also, we have only a rudimentary picture of
human life in the savannah so evolutionary models of this period are very speculative.
Furthermore, it is not clear whether the transmission of behavior is through memes
or genes. On the other side of the coin, there appears to be much room for further
research, and we believe that in the coming decade there will be many advances.
24
4 Pushing the Envelope
of Investigation
.............................................................................
As political economy has matured, it has begun to tackle a wider range of topics.
This work includes a series of larger questions, such as the origins of dictatorship
and democracy. In this section, we consider one of these frontier topics—the size of
nations.
Much of history reflects the expansion and contraction of nations. The conquests
of Alexander the Great, the rise and decline of the Roman Empire, the aggressive
expansions of Napoleon and Hitler, and the dissolution of the USSR are just a few
examples. At the other end of the spectrum, many tiny countries, such as Singapore
and Andorra, have survived a considerable length of time.
For over two millennia, historians and philosophers have asked why some nations
have expanded, why others have contracted, and what is the optimal size of a polity
(Plato, for example, said that the optimal size was 5,040 families). In this section, we
discuss the political economy contribution to this area. In the process, we show how
research in political economy builds upon earlier foundations.
The political economy approach to the size of nations starts with the basic Down-
sian characterization of voter preferences. In this case, voter or citizen preferences can
be placed along a line (or a circle). This line or circle is then divided into n parts (not
necessarily equal), each part representing a country. Each country chooses a policy
position, X
j
, which might be the median or mean of the citizens’ preferences. A
citizen’s utility for her nation’s policy is assumed to be decreasing in distance from
her own preferred position, x (e.g. −|x − X
j
| or − (x − X
j
)
2
). Individuals at the
²⁴ Another group of social scientists employ a different strategy for generalizing about human
behavior from the standard model of rationality, by drawing on cognitive science and psychology (e.g.
North 2005). Space constraints prevent an adequate treatment.
barry r. weingast & donald a. wittman 21
boundary of two countries can choose in which country to reside (see Spoloare, this
volume; Alesina and Spoloare 2003, for a more complete coverage.).
25
Each citizen would like to have his or her country’s policy as closely aligned as
possible to her preferred policy. If policy were the only factor, all countries would
be composed of only one citizen. But other factors run counter to such extreme
decentralization. The most important are economies of scale in production and
military power. When barriers to free trade exist between countries, a more populous
country achieves greater economies of scale through its larger domestic market. A
larger population also allows for greater military power, which may make war against
smaller and weaker states more profitable because of the higher probability of success.
At the same time greater military power makes predation by other states less profitable
to these other states and therefore less likely (see Skaperdas, this volume).
These insights can readily be converted into a comparative statics analysis. When
barriers to free trade are reduced (so that economies of scale can be achieved within a
small country as long as it has sufficient international sales) and the returns to warfare
are decreased, the number of countries will increase and the average country size will
decrease. The returns to warfare depend greatly on the nature of the victim country’s
wealth. If the wealth is in oil, the predating county can expropriate most of the wealth;
when the wealth is in human capital, the predating country can expropriate very
little. In the latter case, the benefits to predation are reduced, the threat of war is
less credible, and the benefits of being a large country are diminished.
In a simple model where wealth is distributed evenly among the citizens, those
citizens at the periphery of the country will be most dissatisfied with their country’s
policy. They are therefore the most likely citizens to exit and join the adjacent country.
Because of economies of scale in production and military power, this “migration” is
costly to those citizens left behind. In order to forestall such migration, countries
might institute a non-linear transfer scheme that grants citizens on the periphery
greater resources. Le Breton and Weber (2001) make this argument and point to a
number of cases, such as Quebec in Canada and some of the border states of India,
where the center grants special rights to the peripheral states. This extension of the
basic model affords a nice illustration of how political economy often grows. Instead
of two competing models, the basic model is expanded so that we have a more general
theory.
In the basic model, all of the countries have the same characteristics; and, when
population is uniformly distributed on the line or circle, all countries are of the same
size. Extensions of the basic model allow countries to have different characteristics.
Nations are characterized as a nexus of public goods. A wise public policy choice
may significantly increase the overall wealth of the citizenry. Successful countries
create conditions for high productivity in the economic sphere by enforcing property
rights and providing social overhead capital, while at the same time minimizing
political costs by creating a system of rules that reduce influence costs and allow for
²⁵ Of course, individuals might not be free to migrate. See Friedman 1977 for an explanation for the
Iron Curtain and the promotion of linguistic boundaries.