گزینه های آموزش و سیاست اقتصادی

This paper investigates whether governments embark on market-oriented reforms as a result of learning. I assume that governments are rational learners who update initial beliefs about the effectiveness of different policies and choose policies on the basis of the updated beliefs. The model of learning is applied to four policy choices: the decision to grant independence to central banks, the decision to liberalize trade, the decision to privatize, and the decision to enter into agreements with the IMF. I also explore whether convergence toward neo-liberal economic policies results from external imposition or simple imitation. I find that learning, in isolation from or in combination with the other mechanisms, explains the decision to liberalize trade, to privatize, and to enter into agreements with the IMF, while none of the mechanisms of convergence explains why governments grant independence to central banks.

In much of the developing world, the 1980s and 1990s were decades of radical economic change. Whereas in the 1960s and 1970s, the prevailing model of development was based on state intervention and inward-looking policies, the 1980s and 1990s were more characterized by the advocacy of market-oriented reforms. These reforms, packaged under the so-called Washington Consensus, aimed to open up national economies and to reduce the role of the state.1 The extent of the consensus became so broad that some referred to ‘universal convergence’ (Williamson, 1990, Williamson and Haggard, 1994, Biersteker, 1995 and Rodrick, 1996).2 A widespread explanation for economic reforms is that governments learned from the experiences of alternative models of development. Learning changes the mapping from policies to economic outcomes and changes beliefs about the consequences of actions and the strategies in a changing economic environment (Kahler, 1990, Kahler, 1992, Hall, 1993, Biersteker, 1995, Tommasi and Velasco, 1995, Haggard and Webb, 1994, Maravall, 1997 and Krueger, 1993). Yet the learning hypothesis remains untested and the question remains unanswered whether governments switched to market-oriented policies as a result of learning. The story of ‘universal convergence’ can be described in the following way: the model of inward-oriented industrialization, epitomized by the experience of many Latin American countries in the 1960s and 1970s, resulted in resounding failure. The bias against exports caused balance of payments crises; devaluations, inflation and fiscal indiscipline became common; and governments borrowed massively from abroad to close the external and fiscal gaps. At the beginning of the 1980s, Mexico's debt moratorium alarmed foreign creditors, who cut off lending. Without credit to finance the pervasive fiscal deficits, governments resorted to the printing press, which resulted in hyperinflations and economic stagnation. Moreover, the proliferation of controls and trade protection was an invitation to tax evasion, rent seeking, and corruption (Tommasi and Velasco, 1995. p. 1–3; Krueger, 1993 and Krueger, 1997). In clear contrast, Chile and the East Asian tigers (Korea, Singapore, Hong Kong and Taiwan) achieved high rates of growth by relying on market mechanisms and greater integration in the world economy. The hallmark of this strategy was export promotion policy, taken to be the quintessential illustration of virtuous policy for a small state.3 The collapse of communist rule in Eastern Europe and the former Soviet Union provided the final blow to the idea that state control and pervasive government intervention were requisites for development. The changes in the South and the East took place amidst a neo-liberal revolution in the North. At the beginning of the 1980s, Conservatives in Great Britain and Republicans in the U. S. undertook campaigns against ‘big government’. The neo-liberal revolution put an end to the Keynesian Consensus, which had dominated public affairs since World War II. Governments could observe the contrasting experiences and change beliefs about the economic consequences of alternative models. Even short-sighted politicians could not avoid the conclusion that the old policies had failed and that the new orthodoxy had produced economic success (Kahler, 1990). Bad experiences thus discredited a particular course of action and successful experiences validated an alternative. The way in which the contrasting experiences were interpreted was crucial: the diagnosis of the cases of success (less state intervention and outward-oriented policies) was precisely the opposite of the diagnosis of the cases of failure (excessive state involvement and inward-oriented development). Lessons were drawn somewhat selectively on the basis of geographic propinquity or linguistic and cultural similarities (Robinson, 1998, Hacking, 1997 and Ramamurti, 1999). As a result of this learning process, switches to market-oriented policies occurred. A test of this story requires an operational concept of learning. Although the discussion about learning has been prolific, it has focused more on definitional questions than on empirical issues or theory building (March and Olsen, 1989, Rose, 1991, Bennett and Howlett, 1992, May, 1992, Hall, 1993, Sabatier and Jenkins-Smith, 1993, Levy, 1994, Adler and Haas, 1997 and Stone, 1999). To fill the gap on empirics, I assume that governments act as rational learners: governments update initial beliefs about the expected results of alternative policies with all available information about policy outcomes in the past and elsewhere. After updating beliefs, governments choose the policy that is expected to yield the best growth results. Hence, the model I test is one of benevolent government in which politicians first learn in the light of experience and then make rational decisions on the basis of what they have learned. Having been exposed to the same information, governments converge in beliefs, and hence in their choices. This is a model in which governments' preferences for market reforms are endogenous and dynamic. It is the experience under alternative policies that determines the evolution of preferences overtime. Learning is one possible mechanism that can cause policies to converge towards the neo-liberal model. Governments could also engage in economic policy reform as a result of outside influence, in particular, International Financial Institutions (IFIs) conditionality (exchange of policies for loans), in which case policies are imposed. Alternatively, governments might have adopted market-oriented policies through imitation (Weyland, 2004). In distinction to learning, imitation does not entail an enhanced understanding of the mapping from economic policies to economic outcomes. Rather, imitation entails following international trends in policy making. In this study, coercion and imitation are the alternative hypotheses. The paper proceeds as follows. In Section 2, I present the Bayesian model of learning and the policy choice problem. In Section 3, the model is tested on a set of developed and developing countries over the period 1951 through 1990 and for four market-oriented policies—central bank independence (CBI), privatization, trade liberalization, and agreements with the International Monetary Fund (IMF). In Section 4, I present an extended model of learning that includes imitation and coercion as alternative mechanisms of policy convergence. I conclude in Section 5 with a reassessment of the lessons about learning provided by this study.

This paper has had two objectives. First, I tested the contention of the literature on economic reforms that learning caused a wave of neo-liberal reforms in developing countries in the 1980s and 1990s. The second objective has been to investigate empirically the concept of learning. Bennett and Howlett conclude: “there has not been a shortage of theorization. Our review suggests that, if anything, the concept has been overtheorized and underapplied” (1992, p. 280). Bennett (in Stone, 1999, p. 52) similarly points out the “paucity of systematic research that can convincingly make the case that cross-national policy learning has had a determining influence on policy choice”. This paper makes an important, even if preliminary, methodological contribution to the empirical literature on learning and public policy. To address these issues, I have modeled government behavior as rational and assumed updating of initial beliefs about the outcomes of policies using all available information. Learning in isolation or in combination with the alternative mechanisms of emulation and imposition explains the decision to liberalize trade, to privatize, and to enter into agreements with the IMF. However, neither learning nor emulation or coercion explains the decision to grant independence to central banks, which is the only policy in which choices did not converge overtime. There is no evidence that own experience is more relevant than the experience of others in the decision to switch to market-oriented policies. And rational learning cannot explain why governments remain under market-oriented reforms after they adopt them. As a first approach to an elusive topic, I should emphasize that there is considerable room for improvement both in method and substance. First and most obvious, I have assumed that governments are benevolent and evaluate policies only in economic terms but it is quite clear that political and self-interested calculations must also enter the picture. Governments are very likely to care not only about the economic consequences of policies but also about the political consequences of economic outcomes. Thinking about how learning from the political outcomes of others may influence policy switches and the empirical test of this hypothesis is a pending task. Second, the results concerning own experience should be taken with caution. The amount of available experience was crucial in order for the influence of prior beliefs to vanish overtime. However, for own experience, the influence of prior beliefs is likely to persist given that only a piece of information is available at a time. And, finally, the analysis should be extended to include the 1990s and new policy illustrations. Given the collapse of communist rule in the late 1980s and early 1990s, it may not be by chance that the learning variables appear to be the most robust precisely in the illustration that includes the 1990s in the analysis.