Hartmut Kliemt: Papers on Buchanan and Related Subjects. Munich: Accedo
Verlagsgesellschaft, 1990, 167 pp.

George Tsebelis: Nested Games. Rational Choice in Comparative Politics. Berkeley,
California: University of California Press, 1990, 274 pp.

I

The books under review here have two things in common. First, they are concerned with social institutions. Second, the approach used by the authors is that of rational choice models. The space here does not allow a full presentation of each book. Therefore, I will only sketch the underlying research programme in each case, and give one or two illustrations from the more detailed analysis. In addition, I will comment briefly on the general fruitfulness of the ventures undertaken by the authors, especially by Dunleavy and Tsebelis.

II

Tsebelis's ambition is to account for political phenomena which, at first sight, seem to involve irrational behaviour. For example, British Labour activists have on several occasions rejected popular candidates who were insufficiently left-wing. A dramatic illustration is provided by the case of Bob Mellish, who in March 1983

. . . resigned his seat in Bermondsey in response to pressure from local activists. He was replaced by the leader of these activists, Peter Tatchel, who received the Labour Party nomination. In the by-election, Conservatives and former Labour supporters voted for the Liberal candidate, who won by some nine thousand votes. One of the safest Labour seats in the London area was lost (Tsebelis, p. 120).

When a rational choice analyst is being faced with apparently irrational behaviour, there are essentially two research strategies available. First, he may simply abandon the assumption that the relevant decision-makers act rationally. In that case, there will no longer exist any puzzle to be explained. The second possibility is to revise the basic model, while the assumption of rationality is being updated. Tsebelis follows the latter strategy. For the case of the British Labour Party, he starts by presenting a simple model, which describes the nomination process as a one-shot sequential game between the MP and party activists. He then shows that rational behaviour in this model is inconsistent with observations of the type referred to above. In order to resolve the puzzle, he revises the model in several directions: first, he introduces a third player (the National Executive Committee of the Labour

Party - NEC). Second, he considers repeated play of the game, thereby allowing long-term considerations to influence the players' behaviour. Third, he modifies the original assumption of complete and perfect information, meaning that reputation effects can play a part in the game. Finally, he argues that the reselection game is 'nested' inside a game of party competition, which modifies the actors' payoffs in the former game. With these modifications of the original model, the behaviour of the activists can be accounted for as rational. Thus, the main conclusion is that the activists

. . . were choosing what would have been a suicidal strategy in a one-shot game because they were involved in an iterated and nested game. (...) Each time activists rejected a standing MP, they were sending an additional signal and creating a reputation for 'toughness' ... In the future, they could cash in on this reputation because MPs would not dare to be moderate or because the NEC would think twice before engaging in a battle with the G(eneral) M(anagement) C(ouncil) (Tsebelis, p. 154).

I find Tsebelis's work very attractive. There are several reasons for this, of which two will be discussed here. First, it illustrates clearly the fruitfulness of the general methodological principle that one ought to start with the simplest case, and then proceed stepwise in the direction of more complicated (and allegedly more realistic) models. Indeed, the entire research programme underlying Tsebelis's book may be said to mirror this principle.

Second, it introduces the use of explicit models to a field where this has hitherto been a somewhat rare phenomenon, namely comparative politics. In a way, the lack of explicit models in this field is something of a puzzle, since it is only through the use of models that comparisons become meaningful. Social phenomena cannot be compared directly, meaning that comparisons presuppose models. Thus, models can be said to be the very essence of this discipline, and at least implicitly present in any comparing venture. What is 'new' is therefore not the use of models as such, but that the models are being stated explicitly and precisely. This said, however, it should be added that Tsebelis's book is really more concerned with isolated case studies than with comparisons between different countries. Nevertheless, the book contains several ideas with clear relevance for comparative applications. An example is his combined rationalist and institutionalist approach. If political outcomes vary across countries, one obvious explanation may be that political institutions differ. One way to explore this possibility is to ask how decision-makers respond to different institutional arrangements. But in order to do this in a precise way, an explicit theory of human behaviour is needed. Of course, this is exactly what the theory of rational choice offers.

III

Any reader of Patrick Dunleavy's Democracy, Bureaucracy and Public Choice is
immediately stunned by the following passage, opening the preface:

When I first began working on some of the ideas in this book, my attitudes towards public choice theory were wholly critical. My idea was to develop an ad hominem critique of some key rational choice models, to show that they were internally inconsistent and ideologically slanted, and to do what I could to demonstrate that this was only a speciously attractive research programme. My own work was to be merely a ladder used to reach this limited goal, later to be kicked down when it was no longer needed. Instead, ten years later, much of my research has been restructured around a particular kind of institutional public choice method. I now recognize the value of instrumental models as a mode of thinking clearly

about the manifold complexities of political life, and could not pursue research without
using them. Yet my scepticism about the scientific status of much liberal political economy
remains, especially about its close links with the new right in practical politics (p. xi).

I have chosen to cite this long quotation for two reasons. First, it leaves the impression of an open-mindedness all too rare in the apparently endless quarrels between adherents of rational choice theory and their critics. More important in our context, however, is that it also mirrors nicely the kind of research reported by Dunleavy in the present book. Thus, on the one hand, he draws extensively from instrumental models found in standard works on public choice, such as Olson's theory of collective action, Downs's models of voter behaviour and party competition, Niskanen's work on budget maximizing bureaucrats. On the other hand, Dunleavy remains highly critical to these models, and seeks to revise them, drawing on contributions from other traditions in political science. Thus, throughout the book chapters or chapter sections which set out existing public choice approaches alternate with a rival or reconstructed view of the same institutions or processes. For example, a revised model of group joining is presented in chapter 3, after discussing Olson's work in Chapter 2. Furthermore, this revised model is later also applied to explain why people bother to vote.

While Dunleavy remains basically critical to classical public choice models, his revisions proceed along the same basic lines as do the ones he criticizes. Notably, he preserves the view of human action as basically instrumental. These revisions no doubt yield insights about the social processes which are being considered, over and above those which can be reached through the more limited focus offered by the classical models. Nevertheless, I remain somewhat undecided about the extent to which any research programme based on a purely instrumental picture of human action can possibly have more than limited success in this field. One way of modelling human behaviour is to say that it is basically influenced by two sorts of determinants. On the one hand there are incentives, which are the focus of the public choice literature. On the other hand, there are all sorts of 'noice', including moral principles, idealism, altruism, social norms, and so on. Where incentives are strong, public choice models tend to do well. For example, even if it is true that many collective action problems are in fact being solved through voluntary contributions, no one would seriously suggest that taxes be collected through any sort of voluntary arrangement. The reason is, I believe, that the problem involves very strong incentives, since there may be large sums of money to be saved by acting as a free-rider. The practically universal resort to force for collecting taxes can be considered as strong support for the relevance of Olson's hypothesized failure of large groups in providing collective goods through voluntary arrangements.

Where incentives are weak, on the other hand, all sorts of noice may be expected to take over. One notable example is the decision to vote. According to the standard Downsian model, a rational voter's calculus consists of three components. These are the party differential, the probability that his vote will be pivotal for the outcome, and the cost of voting. Here, the cost of voting is minimal, comparable to the price of a bus ticket or a few drops of gas. The likelihood that a single vote will be decisive is of course negligible in any national election. Finally, according to Down's own theory of party competition, the party differential, as derived from official party statements, is likely to be small. Furthermore, any promises made by a political party before an election must be discounted by the chance that it will not pay to fulfill them, once in power. On these grounds, voting decisions are not likely to be determined by rational calculations, and it is therefore not surprising that the theory of rational abstention does not carry much predictive power.

I believe that this viewpoint may also be carried over to the decision to join voluntary associations. Again, where this decision involves only minor costs, like a limited membership fee, it is not surprising that Olson's theory does not apply, since incentives are easily bogged down by various sorts of noice.

IV

Hartmut Kliemt's book is called Papers on Buchanan and Related Subjects. Unusual as it may seem, the title is surprisingly accurate. For one thing, this is not a monograph, but a collection of (previously unpublished) papers. Furthermore, at least five of the six essays are directly connected to Buchanan's work, although the closeness of the tie varies somewhat. It peaks in the closing paper, which is more or less a detailed review of Brennan & Buchanan's The Reason of Rules. In other chapters, Kliemt elaborates on, or further develops, thoughts and ideas stemming from the Nobel prize winner.

Two interrelated themes seem to recur throughout the book. First, as a moral philosopher well trained in rational choice theory, Kliemt is generally concerned with the relation between moral philosophy on the one hand and economics/ political science on the other. Second, he discusses in detail the possibility of investing in activities that will impose certain constraints on other people's behaviour if the imposition of the constraints furthers the aims, ends, or values of the investors. This, of course, implies that new institutions, or changes in existing ones, are seen as the result of conscious actions made by rational decision-makers. It deserves mention that this idea also plays a central role in Tsebelis's book (especially chapter 4). Kliemt, however, is specifically concerned with the role that morals can play in this context:

The investors use morals as an instrument of the pursuit of their own ends. They exert a moral externality on other individuals because this is helpful for reaching investors' aims, goals, etc. - regardless of whether the imposed moral code is in the sense of any enlightened moral system 'good' or 'bad', 'right' or 'wrong' (p. 37).

An interesting special case discussed by the author (essay III) is the possibility that such 'moral externalities' may further the rise of a work ethic. In a market economy, it is in my interest that your preference for leisure, as compared to work, is reduced. The reason is that if you work more, you will earn more, and hence be able to buy more goods (from me). This, of course, is also true the other way around, so that the result may be that the overall size of the market is expanded beyond its original limit. Assuming that the original state is an equilibrium, this leads to a paradox:

If we look at this result from the point of view of the original equilibrium the work ethic must be regarded as a public bad. . . It seems to push everybody beyond the limits of what he originally found to be his maximum utility choice of work (cum consumption goods) vs. leisure. It is the outcome of moral war. From a contractarian point of view one seems to be forced to draw the conclusion that a market club with a work ethic should be transformed by mutual agreement into one without such an ethic. . . (Kliemt, p. 65).

As Kliemt points out, however, a work ethic can also be conceptualized as a public
good.

An intriguing aspect of this kind of reasoning is which set of preferences should

be decisive when evaluating the final outcome - the initial ones, or those emerging
as a result of mutual moral influence? Kliemt's answer concords with Buchanan's,
namely that the original ones are special:

If we want to change the evaluations of others by means of argument and not by imposing
our values on them we necessarily have to appeal to their values (p. 69).

I find this collection of papers stimulating reading. However, I think the book would have benefited from an introductory chapter, giving the reader some general background, and drawing up the main problems to be discussed in later chapters. Similarly, a short conclusion might have summed up some main differences between Kliemt's own position on the one hand, and Buchanan's on the other.

V

For many, otherwise potential readers, technical difficulties remain a serious barrier to parts of the rational choice literature. A few words on the present authors' use of mathematics therefore seem appropriate. Dunleavy uses the least technical style of the three. Thus, he has invariably chosen verbal over mathematical reasoning. he does make some use of diagrammatic exposition, and on rare occasions of mathematical symbols, no reader should have any difficulties whatsoever of technicalities.

In Kliemt's book, two out of six essays involve the use of mathematics. However, only one of them could possibly cause any difficulties, even for readers who are unfamiliar with this kind of literature. In fact, for the average political scientist, I expect that more difficulties may be caused by the relatively extensive use of specialized terms from rational choice theory in general, and from game theory in particular.

Tsebelis's book is the most technical one. Even here, however, there are long passages without any mathematics, and the reader will have no problem in getting the main messages even if the mathematical parts are skipped altogether. Furthermore, writes extremely tidily, making the book readable despite the technicalities.

This said, however, my personal opinion is that certain parts of the book would have benefited from some more technicalities. It is something of a puzzle when Tsebelis drops the mathematics when proceeding from one-shot games with complete to repeated games with incomplete information (ch. 5). Certainly, as models grow more complex, there is generally more - not less - need for a mathematical language. I am not saying that mathematical deduction would have altered the main conclusions reached by the author. However, it might certainly have 'sharpened' them. For example, Tsebelis might have been able to specify the exact conditions under which a given combination of strategies constitutes an equilibrium (e.g. perfect Bayesian or sequential) in the repeated version of the nomination game. Although more technicalities might also have alienated some potential readers, there are ways of satisfying students of both categories - e.g. by saving the technical material for the appendices. Indeed, this is a technique that Tsebilis himself uses successfully in other parts of the book.