One Economics

The main point of Rodrik’s book is that economics leaves a lot of slack for policy prescriptions. As I see it, this may be true for two distinct reasons. One is that economic knowledge is not sufficiently robust in general to indicate appropriate policies. Another is that it is inherently incomplete, specifically, that it cannot and does not consider all the factors that may matter in the particular situations to which policies are applied. Rodrik emphasizes the second reason but first I want to comment on the title.

To put it in a nutshell, is it “One Economics, Many Recipes” or “Many Economics, …”? In what sense do we have “one economics”? Economics is a science that derives conclusions about states of collectivities (“the economy” but also “the polity,” since the same methods are now applied to politics) from assumptions about preferences of individuals and the constraints they face. What unifies economics are the methods for making such inferences. The assumptions vary: they are at best disciplined by “stylized facts” and often at variance with more direct, psychological, evidence. Moreover, these assumptions often reflect ideological priors. One example that jumps to my mind is an article, published in a leading journal of political economy, that went like this: assume that tax revenues do not finance inputs into production, assume that they do not subsidize consumption, write a growth model, and — surprise — taxes are bad for growth! But almost the same year, another article, published in an equally reputable journal, assumed that public inputs are complementary to private ones and that tax revenues are used to finance public productive services, only to find that growth is maximized at a positive, indeed sizeable, tax rate.

One may think that even if the assumptions are to some extent arbitrary, they are at least disciplined by empirical tests of the conclusions. Assumptions are “as if,” but conclusions must defend themselves against facts. But econometrics are a tolerant disciplinarian. As a frequent practitioner I am not cynical: some results withstand all specifications and all estimators, some results will not appear however one massages the data. But many are simply not robust. Econometric models are also based on assumptions and, indeed, these assumptions are so demanding that at least some of them are, perhaps always, violated. In the presence of endogeneity, errors of measurement, and omitted variables, all estimators are biased. And since different estimators correct for different biases, they often generate disparate results. Moreover, I suspect again that ideology has a way of sneaking even into econometrics. When in 1993 Limongi and I “reviewed studies”:http://www.jstor.org/view/08953309/di980576/98p01105/0 of the effect of political regimes on economic growth, we found that the results perfectly fitted the ideological climate of the period when they were published: before 1982 none found that democracies promoted growth, after 1982 none found that dictatorships did.

All this leads me to think that economics is one only with regard to the criteria of inference and evidence that are required to publish in the journals of the discipline. But with different assumptions and divergent statistical results, the resulting body of knowledge would not be unequivocal even if it were generally applicable to all possible contexts. The neo-liberal tenet that dismantling the state is sufficient for “the market” to usher a country on the road to development had little justification in the body of economic knowledge: after all, the theory says that economies stagnate if markets are perfectly, as distinctly from monopolistically, competitive. Yet this was the advice offered around the world by eminent economists.

Rodrik does not get entangled in these issues of philosophy of science because his point of departure is appropriately practical. His emphasis is that even if economics were one, the concrete situation of particular countries always includes factors that are not considered in general theories or that coexist in interactive combinations not entertained by such theories. And ignoring them may cause havoc. What we need, according to Rodrik, are theoretically and statistically informed analyses of particular situations. Such analyses must be consistent with what we believe theoretically and what we think to be true empirically, but they also must pay attention to facts specific to each context. And since different countries face different problems, these call for different remedies: his axiom is that “appropriate growth policies are almost always context
specific” (page 4).

This is a salutary approach, appropriately cautious about general theories, appropriately wary about treating them as ready-made recipes, and importantly calling for considering the particular facts at hand. As I was reading the book, I remembered a public debate in 1990 about economic reforms in Poland with a prominent economist who had been an advisor to a Latin American country. At one time during the discussion I asked him what proportion of the Polish labor force was employed in agriculture. After some squirming, he gave the number of the country from which he had just returned, which was 40 percent of the true Polish number. No wonder Poland experienced unemployment rates nearing 20 percent during the following fifteen years. Hence, I am taken by Rodrik’s approach to policy design: it is practical, responsible, and modest. Moreover, I found some of his analyses, particularly of industrial policy, exceptionally illuminating.

Yet studying specific situations is a craft, not a science. I am convinced by the spirit of “growth diagnostics” that seeks to identify the constraints to development that are most binding in a particular economic and political context. I also see the theoretical justification of the Hausmann-Rodrik-Velasco method for identifying such most binding constraints. Yet I am not sure that this approach leads to unique conclusions. Indeed, all the applications of the method I have seen rely on intuitive understandings. I can easily imagine someone applying the same method to El Salvador, Brazil and the Dominican Republic and arriving at conclusions different from those of Rodrik. By construction, when we treat situations as unique, at one point we have to stop relying on general knowledge, and then what else can guide us than intuition and perhaps caution?

The chapter on institutions diverges in tone from the pragmatism that permeates the rest of the book. We now get a general recipe: “participatory politics.” Rodrik offers several regression analyses intended to support this recipe, but within the confines of this book he obviously cannot consider the endogeneity of institutions and other econometric issues. The impact of institutions on development is the new rage and even a narrower topic of the impact of democracy on growth has recently been subject to meta-analysis based on over 400 statistical studies. And is the observation that “Mauritius found its own way to economic development because it created social and political institutions that encouraged participation, negotiation, and compromise” (page 168) based on growth diagnostics?

In spite of my reaction against this chapter, it is still an advance over the view in which the political and economic future of a country was sealed by the first mosquito that bit a European settler when he stepped on its shore. I wish, however, that Rodrik had clarified the relation between two points he makes, namely, that a successful market economy can be sustained by a variety of institutional arrangements and that “institutions matter,” even more narrowly, that “participatory institutions” are the engine of growth. If “there is no single mapping between the market and the set of nonmarket institutions required to sustain it” (page 162), then identifying the institutions that are compatible with sustained growth calls for a different methodology than Rodrik, and everyone else, uses. Yet the first point — that development can be sustained under different institutional frameworks as long as they manage conflicts, coordinate, and regulate — is tantalizing and calls for systematic attention. I am less taken by secure property rights, because property rights can be excessively as well as insufficiently protected, but this is a digression.

Altogether, this is a book that convinces by its intelligence and evokes trust by its modesty and its sense of responsibility. In spite of my specific doubts and disagreements, I believe that Rodrik has opened a new vista for policy analysis. After reading it, one can no longer believe in mantras, whether they are agreed upon in Washington or elsewhere. We have to pay attention to real problems that people face in particular contexts and seek to relieve their particular deprivations in ways that are suited to these contexts.