Comments and discussion.

Acemoglu, Daron ; Johnson, Simon ; Querubin, Pablo 等

COMMENT BY ALBERTO ALESINA This paper by Daron Acemoglu and his coauthors addresses an important topic. In general terms, the question is when and how policy reforms are successful or unsuccessful. More specifically, the paper focuses on central bank independence (CBI), and its goal is to shed some light on the inconclusive results in the literature regarding the benefits of CBI on inflation. I will focus relatively briefly on the authors' model and more extensively on the empirical evidence. I conclude that the model is not especially appropriate for studying this issue and that the empirical evidence is not robust.

I-HE MODEL. What models have been used in the literature to study the pros and cons of CBI? The first was the model of time inconsistency and inflation bias developed by Finn Kydland and Edward Prescott and later extended by Robert Barro and David Gordon. (1) The source of inflation in all these models is an interplay between wage setting behavior and the incentive of the central bank to "surprise" the economy with an inflation shock to reduce unemployment. Kenneth Rogoff pointed out how an inflation-averse and independent central banker can improve on the tradeoff between rules and discretion. (2)

A second approach emphasizes the political influences on monetary policy, in particular, political business cycles, both "opportunistic" and "partisan." CBI can limit the inefficient fluctuations of inflation created by political cycles. In this model the source of inflation is monetary policy that is excessively loose because of political influences. Alesina, Gerald Cohen, and Nouriel Roubini discuss this approach in detail. (3)

The third type of model focuses on biases toward fiscal deficit and the pressure on central banks to accommodate such deficits? In this context CBI is part of a package that ensures fiscal responsibility and avoids all excessive use of the inflation tax. Fiscal deficits are the source of inflation in this model.

The model of this paper is quite different. There is a lobby pushing for inflation, but the general public is against inflation. Not much is said in the paper about which lobby that would be in reality, nor does the paper explain why the lobby benefits from inflation. This is a good model for other issues, such as trade protection or regulation of certain sectors. And in fact, similar models have been widely used to discuss how policies that are inefficient for society as a whole are adopted to favor a specific, narrowly defined constituency. But this is not a good model for explaining why countries find themselves in a high-inflation equilibrium. Inflation is almost universally the result of some form of macroeconomic imbalance, not the effect of a small organized lobby pushing for it. The implications in terms of which governments are more likely to be successful and which will undertake reforms are different. The authors mention in passing that the model is meant to be illustrative, but l am not at all sure how their empirical implications would generalize to a more aggregate model of inflation along the lines sketched above.

An additional issue concerns the endogeneity of central bank reform. Many authors share the same problem that this paper has, (5) namely, the assumption that central bank laws are considered an exogenous variable, an issue first raised by Adam Posen. (6) However, I feel that this shortcoming is especially serious for this paper. In fact, it discusses interactions between political regimes and central bank reforms and the role of lobbies in pushing for more or less inflationary policies. The same political variables that make it more or less likely that a central bank reform will be successful are at the root of the question of whether central bank reform is adopted in the first place.

THE EMPIRICAL EVIDENCE. Here I will focus on some sensitivity analysis of the paper's two key tables, table 2 and table 4. (7) First of all, the results as presented in the paper are not especially strong. The theory has implications for the differences between coefficients for three groups of countries defined in terms of the strength of executive constraints. In many cases, as the authors acknowledge, these coefficients, although different from zero, are not statistically different from each other.

My first sensitivity test examined to what extent the results were driven by the Latin American countries, since this is a region that has experienced much inflation and several central bank reforms. I found that, in table 2, dropping Argentina makes all the relevant coefficients insignificant in the full sample; similar patterns hold for table 4. The results are thus very sensitive to dropping one country.

Then I examined the groupings. The authors define their country groups as follows: the intermediate group includes all the countries within one standard deviation of the mean of their constraints index, and the strong and weak groups include those more than one standard deviation above and below the mean, respectively. The problem is that the distribution of countries is very asymmetric: 36.5 percent of the countries are bunched at the top of the distribution, with an index value of seven. Thus, the authors' distribution is as follows:

--strong constraints (index of 7): 36.5 percent of the countries

--intermediate constraints (index between 3.92 and 6.99): 48 percent

--weak constraints (index between 0 and 3.2): 15.5 percent (that is, a handful of countries).

The results hold only if the weak-constraints group is restricted to a very small set of countries. I tried two alternative groupings that seem more reasonable: (8)

--strong constraints: 36.5 percent

--weak constraints: 36.5 percent

--intermediate constraints: the rest,

and

--strong constraints: 36.5 percent

--intermediate and weak constraints: split 50-50 between the remaining countries.

With either grouping, the results of the paper disappear: the weak-constraints group has a negative and significant coefficient, and the middle group has an insignificant and often positive coefficient. Thus even moving a handful of countries from the middle group to the weak group changes the results. Therefore the strategy of looking at the interaction of a quadratic function of the executive constraint variable with the CBI variable does not guarantee robustness to the grouping classification, contrary to the authors' claims. (9)

In conclusion, investigating the success of policy reforms in alternative institutional settings is a very good idea, but the paper does not use the most appropriate model to address the issue at hand, and thus it derives implications that may not be plausible. Moreover, the empirical evidence is not robust.

(1.) Finn E. Kydland and Edward C. Prescott, "Rules Rather Than Discretion: The Inconsistency of Optimal Plans," Journal of Political Economy 85, no. 3 (1977): 473-91; Robert Barro and David Gordon, "Rules, Discretion, and Reputation in a Model of Monetary Policy," Journal of Monetary Economics 12, no. 1 (1983): 101-21.

(2.) Kenneth Rogoff, "The Optimal Degree of Commitment to an Intermediate Monetary Target," Quarterly Journal of Economics 100, no. 4 (1985): 1169-89.

(3.) Alberto Alesina, Gerald Cohen, and Nouriel Roubini, Political Cycles and the Macroeconomy (MIT Press, November 1997).

(4.) For an extended discussion, see Torsten Persson and Guido Tabellini, Political Economics: Explaining Economic Policy (MIT Press, 2002).

(5.) For example, Alberto Alesina and Lawrence Summers, "Central Bank Independence and Macroeconomic Performance: Some Comparative Evidence," Journal of Money, Credit and Banking 25 (May 1993): 151-62.

(6.) Adam Posen, "Declarations Are Not Enough: Financial Sector Sources of Central Bank Independence," NBER Macroeconomic Annual 10 (1995): 253-74.

(7.) I thank the authors for providing their data and codes, which made it very easy to replicate and examine their results.

(8.) Note that, because of the bunching at the top, one cannot quite split the sample evenly into thirds.

(9.) The results of all these sensitivity tests are available from the author upon request.

COMMENT BY DAVID ROMER The thesis of this paper by Daron Acemoglu, Simon Johnson, Pablo Querubin, and James Robinson is simple but important. There has been a vast amount of work on the political economy of poor policies. Yet when it comes to thinking about the effects of policy reform, economists and policymakers too often still adopt the perspective of a benevolent social planner and assume that reforms will be implemented as intended. The basic insight of the paper is that in thinking about the effects of reform, it is a first-order mistake to overlook the forces that gave rise to the need for reform in the first place. Considering those forces can critically change what one should expect. Often the change is particularly stark: if a country is following poor policies because the rule of law is not respected, reforms that consist of passing new laws are unlikely to change anything at all. These core ideas more than repay the effort of reading the paper.

THE THEORY. The authors develop their thesis both theoretically and empirically, focusing on how central bank independence affects inflation. The main result of their theoretical model is that central bank reform is unlikely to reduce inflation much either in countries with unconstrained leaders (because the reform will be ineffective) or in countries with highly constrained leaders (because inflation will already be low). The largest effect is likely to occur in countries with moderate levels of constraints on their leaders.

I have three main comments about the theory. First, it appears that nothing critical hinges on the paper's assumption that excessive inflation is the result of special-interest lobbying. For example, suppose inflation arises instead from dynamic inconsistency or the political business cycle. Central bank reform would be unlikely to reduce inflation if power in fact remained in the hands of politicians, but would be likely to do so if the reform genuinely put control of monetary policy in the hands of an independent central bank. The same is true if--to mention my favorite theory of high inflation--inflation occurs because politicians understand the short-run benefits of demand stimulus but not the costs of inflation or the difficulty of bringing it down. The fact that the theoretical results do not rest on a specific view of the source of excessive inflation has the advantage that the theory is not tied to the debatable view that inflation is the result of redistributive efforts. But it has the disadvantage of reducing the ability of tests based on the theory to discriminate among competing views of the reasons for poor policies.

Second, the prediction of an inverted U-shaped relationship between constraints on politicians and the effects of central bank reform on inflation seems unlikely to be robust. There are two competing effects. On the one hand, when constraints on politicians are weaker, the prereform situation is worse, so the room for improvement is greater. On the other, when constraints are stronger, the extent to which nominal reforms are likely to have force is greater. This suggests that the overall effect can go either way, but it does not suggest a strong reason for expecting an inverted U-shaped relationship in particular. It is not difficult to find changes in functional forms or other assumptions of the model that alter the prediction of an inverted U-shaped relationship. To give a simple example, just dropping the assumption in the paper's model that inflation cannot be negative changes the relationship from an inverted U to monotonically increasing.

Third, a prediction that is likely to be more robust is that if one controls for initial inflation, the amount that central bank reform reduces inflation will be monotonically increasing in the strength of political constraints. Controlling for initial inflation eliminates one of the two competing effects, and so leaves only the fact that nominal reforms are likely to have more force when politicians are more constrained.

The paper's model is not well suited to analyzing this issue, for two reasons. First, the prereform level of inflation is completely determined by the extent of constraints, so once one controls for initial inflation there is no remaining variation in political constraints. Second, the model's assumption that inflation cannot be negative means that inflation is sometimes at a corner solution. This is important to the model's implications, but it seems quite unrealistic and complicates the analysis.

I therefore consider a modest variation on the paper's model. First, I replace the paper's social welfare function with a conventional quadratic one,

(1') u([pi]) = - b/2 [[pi].sup.2], b > 0,

where I have normalized the socially optimal level of inflation to zero. Second, I assume that the lobbyists' preferred inflation rate (or, more generally, the inflation rate that would prevail in the absence of constraints on politicians) varies across countries. Specifically, I assume that in the lobbyists' utility function, w([pi],t) = [alpha][pi] - [beta]/2 [[pi].sup.2] - t (the paper's equation 3), [alpha] may be heterogeneous. And third, I model central bank reform as an increase in the social cost of departures of inflation from its optimal level--that is, as an increase in b. Thus, I interpret central bank reform as an increased emphasis in the conventional (noncorrupted) policymaking arena on reaching the optimal level of inflation. The rest of the model is the same as the paper's. The politician maximizes [lambda]u([pi]) + (1 - [lambda])t, where [lambda] measures the strength of constraints on politicians and t is the transfer from the lobby, and the lobby makes a take-it-or-leave-it offer to the politician.

Analyzing the model along the lines of the paper yields an expression for equilibrium inflation:

(7') [pi] = [alpha]/[beta] + [lambda]/1 - [lambda] b.

This expression has two implications. First, it shows that my variant of the authors' model captures their idea that reform does not affect inflation if political constraints are either very low or very high. When [lambda] is zero (politicians are unconstrained), inflation equals lobbyists' preferred level of [alpha]/[beta] regardless of how much weight is given to achieving low inflation. in the conventional policymaking process (b). And when [lambda] is one (politicians are completely constrained), lobbyists have no influence, and so inflation is zero for any level of b. More generally, equation 7' implies the inverted U-shaped relationship that the authors' emphasize.

The second implication of equation 7' is that

[partial derivative][pi]/[partial derivative]b = -([lambda]/1 - [lambda]) (1/[beta] + [lambda]/1 - [lambda] b) [pi].

That is, at a given level of inflation, the amount by which inflation falls in response to central bank reform is monotonically increasing in the strength of political constraints. The intuition is exactly that described above: trying to correct a bad outcome through some kind of nominal reform is more effective when the rule of law is stronger.

THE EMPIRICAL WORK. Let me now turn to the empirical findings. One of the most interesting aspects of the results is the absence of a strong correlation between increases in central bank independence and decreases in inflation. This is most easily seen in the figures in appendix B, which are a marvelous feature of the paper. Over the paper's sample period, forty countries increased the independence of their central bank. But in thirty-four of those countries, inflation relative to the world average either showed no clear change from its behavior before the reform or, in a number of cases, rose modestly but clearly. In only six countries was inflation following the reform clearly lower than one would have expected based on its prior behavior. This is hardly a stunning endorsement of the anti-inflationary power of central bank independence.

One can also see from these figures why the authors obtain their result about the relationship between political constraints and the behavior of inflation following central bank reform. The six countries where inflation clearly fell following reform are Argentina, Mongolia, Nicaragua, Peru, Turkey, and Uruguay; all six are in the middle group in terms of political constraints. Since only slightly more than half of the forty countries that implemented reform are in this group, this pattern would be very unlikely to occur by chance.

Looking at these correlations is interesting, and the finding that increases in central bank independence are associated with falls in inflation only in countries with intermediate levels of political constraints is intriguing. At the same time, I do not think one should get overly excited about this result, for three reasons.

First, as the authors emphasize, in most specifications the effect is only moderately statistically significant.

Second, there is considerable room for improvement on the authors' measure of political constraints. The authors employ the average of the Polity IV measure of constraints on the executive over their full sample period, 1972-2004. This means that if a country reformed its central bank in, say, 1992, the authors' measure puts as much weight on political constraints twenty years before the reform, and ten years after the reform, as at the time of the reform. If another country had its reform in 2002, the paper's measure for this country goes back thirty years rather than twenty.

I think it is clear that, conceptually, one would like to know how constrained politicians were at the time of the reform. The extent of constraints at some point decades earlier or a decade later should not matter. What should matter for whether reform puts genuine control over monetary policy in the hands of a legally independent central bank is whether the rule of law is respected when the reform takes place. Because the Polity IV measure is not perfect, its lagged values surely convey some information about current constraints given its current value. But this hardly makes a case for the paper's approach of treating each year identically.

Putting much more weight on the measure of constraints near the time of the central bank reform changes one's views of many of the reforms. Consider the countries that the authors classify as having an intermediate level of constraints. Four of these countries (Bolivia, Spain, Turkey, and Uruguay) achieved the highest possible level of constraints a full decade before their reforms and kept them at that level through the end of the sample, and four more (Ecuador, Greece, Mongolia, and Portugal) attained the highest possible rating at least three years before their reforms and maintained it well afterward. In addition, Chile had an essentially unconstrained executive for fifteen years before its reform (although the constraints jumped to the highest possible level in the year of the reform), and Peru had a very low level of constraints from the year before its reform through six years afterward. Thus, a more appropriate measure of the constraints relevant to the effects of central bank reform would look quite different from the authors' measure. (1)

Third, the paper looks only at the model's predictions about changes in inflation in response to changes in central bank independence, while the model makes a rich set of predictions about the level of inflation and how it responds to changes in both central bank independence and political constraints. For example, as I described, the model makes predictions about the strength of political constraints and the fall in inflation following central bank reform controlling for initial inflation. And it appears that at least one of the model's major predictions about those additional dimensions fails. The model predicts that with weak political constraints, inflation will be high both before and after central bank reform. But as the authors mention, this is not what the data show: in the eight countries in the weak-constraints group, inflation has been generally below the world average in seven.

Thus, as with many other valuable papers, this paper's contribution lies more in raising issues than in resolving them. The paper does not come close to settling the issue of how the political economy forces that give rise to poor policies affect the consequences of policy reform. But it makes a compelling case that those effects are likely to be large and important.

(1.) The results based on an alternative measure of political constraints reported in table 3 of the paper are not reassuring in this regard, for two reasons. First, this measure also does not systematically attempt to measure constraints around the times of central bank reforms. Second, the results with this measure suggest that central bank reform reduces inflation by at least as much under weak political constraints as under moderate constraints, which contradicts one of the paper's main hypotheses.

GENERAL DISCUSSION Gregory Mankiw wondered how best to take political economy constraints into consideration when doing policy analysis. If all outcomes are assumed to be the result of political constraints, analyzing the welfare implications of alternative policies becomes a purely academic exercise. This paper seemed to propose a compromise: sometimes policy can be altered to improve overall welfare, but other times it is determined by politics. Mankiw was not sure how helpful this perspective would be in practice.

Justin Wolfers suggested that some of the data used in the paper may be unreliable. For example, the data from Zimbabwe indicate that the central bank is independent, which, practically speaking, is probably not the case. The lesson to draw from this paper may be not that central bank reform is futile when political institutions are weak, but rather that central bank reform is futile when institutions are weak and the reforms are not actually being implemented. Conversely, some countries, such as Australia, have de facto central bank independence before they formalize it in law, so the formal policy change appears to have little effect. This should not lead one to conclude that central bank independence is ineffectual in such countries, however.

Michael Woodford questioned the relationship between constraints on the executive and the effect of policy reform on inflation. In countries with strong constraints, changes in central bank law might simply be less correlated with other important monetary policy changes. For example, the paper cites as pertinent policy changes the 1999 Riksbank Law in Sweden and the 1998 Bank of England Act in the United Kingdom. In fact, the most important anti-inflationary policy changes in both countries probably occurred in 1992, following the crisis in the European fixed exchange rate regime. Likewise, the 1998 Bank Act in the Netherlands is cited as an important policy change, but probably the most important anti-inflationary actions were those taken following the Maastricht Treaty at the beginning of the 1990s. The paper does not list any major reform in the United States during the sample period because there were no changes to the Federal Reserve Act; however, there was an important change in the character of U.S. monetary policy with the accession of Paul Volcker as Federal Reserve chairman at the beginning of the 1980s.

Benjamin Friedman raised another issue that appears in the literature on central bank independence, namely, the correlation between independence and the country's sacrifice ratio, a measure of the output cost of lowering inflation. If central bank independence delivered lower inflation by anchoring inflation expectations, this would imply that if some supply shock were to cause inflation to increase unexpectedly, a country that has an independent central bank should be able to return to low inflation at lower cost than one that does not. However, it turns out that the correlation goes in the other direction: countries with more independent central banks tend to have worse sacrifice ratios. This casts doubt on the notion that the relationship between central bank independence and low inflation is causal, and it suggests that these phenomena may be jointly caused by some third variable. Therefore Friedman proposed that the authors look at the relationship between the variables they examine and the countries' sacrifice ratios to see what this implies about their model.

Edward Glaeser also recommended that the authors examine the interaction between central bank independence and other country-level variables. The observed correlation between central bank independence and constraints on the executive may be caused by a third variable that jointly determines both, such as GDP.

Gary Becker observed that shifts in political power often facilitate policy changes. Examining such power shifts, in addition to the legal reforms studied in the paper, might be useful. Olivier Blanchard noted that the reasons for central bank reform are important in influencing outcomes and are missing from the paper's argument. Some countries adopt central bank independence because of pressure from international organizations or from other countries. Christopher Sims added that, in some instances, central bank independence is simply a bad idea that economists have forced onto a complicated and unfavorable political environment. Without the institutions and fiscal policies necessary for an independent central bank to be effective, central bank reform can become little more than an excuse for the legislature to abdicate responsibility for inflation.

ACKNOWLEDGMENTS We are grateful to Ioannis Tokatlidis for help in constructing the data, to Marco Arnone for sending us his data, and to Alberto Alesina and David Romer for detailed comments. We are also grateful to seminar participants at the Canadian Institute for Advanced Research, the macro lunch at the Massachusetts Institute of Technology, and the Brookings Papers conference, and to Marios Angeletos, Timothy Besley, Olivier Blanchard, Leopoldo Fergusson, Torsten Persson, and Guido Tabellini for comments and suggestions. Finally, we thank Jeffry Frieden for several very helpful conversations and Lawrence Broz for guidance to the literature.

References

Acemoglu, Daron. 2006. "A Simple Model of Inefficient Institutions." Scandinavian Journal of Economics 108, no. 4: 515-46.

Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2005. "The Rise of Europe: Atlantic Trade, Institutional Change, and Economic Growth." American Economic Review 95, no. 3: 546-79.

Acemoglu, Daron, James A. Robinson, and Thierry Verdier. 2004. "Kleptocracy and Divide-and-Rule: A Model of Personal Rule." Journal of the European Economic Association 2, no. 2-3: 162-92.

Acemoglu, Daron, Simon Johnson, James A. Robinson, and Yunyong Thaicharoen. 2003. "Institutional Causes, Macroeconomic Symptoms: Volatility, Crises and Growth." Journal of Monetary Economics 50, no. 1: 49-123.

Alesina, Alberto. 1988. "Macroeconomics and Politics." NBER Macroeconomics Annual 3: 13-52.

Alesina, Alberto, and Allan Drazen. 1991. "Why Are Stabilizations Delayed?" American Economic Review 81, no. 5:1170-88.

Alesina, Alberto, and Lawrence H. Summers. 1993. "Central Bank Independence and Macroeconomic Performance: Some Comparative Evidence." Journal of Money, Credit and Banking 25, no. 2:151-62.

Arellano, Manuel, and Stephen R. Bond. 1991. "Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations." Review of Economic Studies 58, no. 2: 277-97.

Arnone, Marco, Bernard J. Laurens, and Jean-Francois Segalotto. 2006. "The Measurement of Central Bank Autonomy: Survey of Models, Indicators, and Empirical Evidence." IMF Working Paper WP/06/227. Washington: International Monetary Fund.

Arnone, Marco, Bernard J. Laurens, Jean-Francois Segalotto, and Martin Sommer. 2007. "Central Bank Autonomy: Lessons from Global Trends." IMF Working Paper WP/07/88. Washington: International Monetary Fund.

Aufricht, Hans. 1967. Central Banking Legislation: A Collection of Central Bank, Monetary, and Banking Laws. International Monetary Fund Monograph Series. Washington.

Banaian, King, Richard C. K. Burdekin, and Thomas D. Willett. 1998. "Reconsidering the Principal Components of Central Bank Independence: The More the Merrier?" Public Choice 97, no. 1-2: 1-12.

Barro, Robert J., and David B. Gordon. 1983. "A Positive Theory of Monetary Policy in a Natural Rate Model." Journal of Political Economy 91, no. 4: 589-610.

Bates, Robert H., and Anne O. Krueger, eds. 1993. Political and Economic Interactions in Economic Policy Reform: Evidence from Eight Countries. Oxford, U.K.: Blackwell.

Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer, and Patrick Walsh. 2001. "New Tools in Comparative Political Economy: The Database of Political Institutions." World Bank Economic Review 15, no. 1: 165-76.

Bekaert, Geert, Campbell R. Harvey, and Christian Lundblad. 2005. "Does Financial Liberalization Spur Growth?" Journal of Financial Economics 77, no. 1: 3-55.

Boycko, Maxim, Andrei Shleifer, and Robert W. Vishny. 1996. "A Theory of Privatisation." Economic Journal 106, no. 435: 309-19.

Burnside, Craig, and David Dollar. 2000. "Aid, Policies, and Growth." American Economic Review 90, no. 4: 847-68.

Cabrera Galvis, Mauricio, and Jose Antonio Ocampo. 1980. "Precios Internacionales, Tipo de Cambio e Inflacion." In Inflacion y Politica Economica, edited by Mauricio Cabrera Galvis. Bogotaq Asociacion Bancaria de Colombia.

Campillo, Marta, and Jeffrey A. Miron. 1997. "Why Does Inflation Differ across Countries?" In Reducing Inflation: Motivation and Strategy, edited by Christina D. Romer and David H. Romer. University of Chicago Press.

Carrasquilla, Alberto. 1999. Estabilidad y Gradualismo : Ensayos sobre Economia Colombiana. Santafe de Bogota: TM Editores.

Coate, Stephen T., and Stephen E. Morris. 2006. "Policy Conditionality." In Globalization and the Nation State: The Impact of the IMF and World Bank, edited by Gustav Ranis, James Vreeland, and Stephen Kosack. Routledge.

Collier, David, and Ruth Berins Collier. 1991. Shaping the Political Arena: Critical Junctures, the Labor Movement, and Regime Dynamics in Latin America. Princeton University Press.

Crowe, Christopher, and Ellen E. Meade. 2007. "The Evolution of Central Bank Governance around the World." Journal of Economic Perspectives 21, no. 4: 69-90.

Cukierman, Alex. 1992. Central Bank Strategy, Credibility and Independence: Theory and Evidence. MIT Press.

Cukierman, Alex, Geoffrey Miller, and Bilin Neyapti. 2002. "Central Bank Reform, Liberalization and Inflation in Transition Economies: An International Perspective." Journal of Monetary Economics 49, no. 2: 237-64.

Cukierman, Alex, Steven B. Webb, and Bilin Neyapti. 1992. "Measuring the Independence of Central Banks and Its Effect on Policy Outcomes." World Bank Economic Review 6, no. 3: 353-98.

Daunfeldt, Sven-Olov, and Xavier de Luna. 2003. "Central Bank Independence and Price Stability: Evidence from 23 OECD-Countries." Working paper. Umea University.

Davila Ladron de Guevara, Andres, Rodolfo Escobedo, Adriana Gavina, and Mauricio Vargas. 2000. "El Ejercito Colombiano durante el Periodo Samper: Paradojas de un Proceso Tendencialmente Critico." Colombia Internacional 49/50: 148-77.

Dewatripont, Mathias, and Gerard Roland. 1997. "Transition as a Process of Large Scale Institutional Change." In Advances in Economics and Econometrics: Theories and Applications: The Seventh World Congress, edited by David M. Kreps and Kenneth Wallis. Cambridge University Press.

Dollar, David, and Jakob Svensson. 2000. "What Explains the Success or Failure of Structural Adjustment Programmes?" Economic Journal 110, no. 466: 894-917.

Drazen, Allan. 2000. Political Economy in Macroeconomics. Princeton University Press.

Easterly, William. 2005. "National Policies and Economic Growth: A Reappraisal." In The Handbook of Economic Growth, edited by Philippe Aghion and Steve Durlauf. Amsterdam: North Holland.

Easterly, William, Ross Levine, and David Roodman. 2004. "Aid, Policies, and Growth: Comment." American Economic Review 94, no. 3: 774-80.

Echeverry, Juan Carlos. 2002. Las Claves del Futuro: Economia y Conflicto en Colombia. Bogota: Editorial Oveja Negra.

Eijffinger, Sylvester C. W., and Jakob de Haan. 1996. "The Political Economy of Central-Bank Independence." Princeton University Department of Economics International Finance Section.

Fernandez, Raquel, and Dani Rodrik. 1991. "Resistance to Reform: Status Quo Bias in the Presence of Individual Specific Uncertainty." American Economic Review 81, no. 5:1146-55.

Forder, James. 1998. "The Case for an Independent European Central Bank: A Reassessment of Evidence and Sources." European Journal of Political Economy 14, no. 1: 53-71.

Galiani, Sebastian, Paul Gertler, and Ernesto Schargrodsky. 2005. "Water for Life: The Impact of the Privatization of Water Service on Child Mortality." Journal of Political Economy 113, no. 1: 83-120.

Gibson, Edward L. 1997. "The Populist Road to Market Reform: Policy and Electoral Coalitions in Mexico and Argentina." World Politics 49, no. 3: 339-70.

Grilli, Vittorio, Donato Masciandaro, and Guido Tabellini. 1991. "Political and Monetary Institutions and Public Financial Policies in the Industrial Countries." Economic Policy 13: 341-92.

Grossman, Gene M., and Elhanan Helpman. 1994. "Protection for Sale." American Economic Review 84, no. 4: 833-50.

Gutirrrez, Eva. 2003. "Inflation Performance and Constitutional Central Bank Independence: Evidence from Latin America and the Caribbean." IMF Working Paper WP/03/53. Washington: International Monetary Fund.

Henisz, Witold J. 2000. "The Institutional Environment for Economic Growth." Economics and Politics 12, no. 1: 1-31.

Jacome, Luis I. 2001. "Legal Central Bank Independence and Inflation in Latin America during the 1990s." IMF Working Paper WP/01/212. Washington: International Monetary Fund.

Jacome, Luis I., and Francisco Vazquez. 2005. "Any Link between Legal Central Bank Independence and Inflation? Evidence from Latin America and the Caribbean." IMF Working Paper WP/05/75. Washington: International Monetary Fund.

Jaramillo, Juan Carlos, Roberto Steiner, and Natalia Salazar. 1999. "La Economia Politica de la Politica Cambiaria en Colombia." Coyuntura Economica 29, no. 2.

Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2007. "Governance Matters VI: Aggregate and Individual Governance Indicators 1996-2006." World Bank Policy Research Working Paper 4280. Washington: World Bank.

Keefer, Philip, and David Stasavage. 2002. "Checks and Balances, Private Information, and the Credibility of Monetary Commitments." International Organization 56, no. 4: 751-74.

--. 2003. "The Limits of Delegation: Veto Players, Central Bank Independence, and the Credibility of Monetary Policy." American Political Science Review 97, no. 3: 407-23.

Killick, Tony. 1978. Development Economics in Action: A Study of Economic Policies in Ghana. London: Heinemann.

King, Michael. 2001a. "The Bank of Canada's Pursuit of Price Stability: Reputation as an Alternative to Independence." Central Banking 12, no. 1: 68-78.

--. 2001b. "Contrasting Approaches to Central Bank Independence: Australia and New Zealand." Central Banking 12, no. 2: 58-68.

La Porta, Rafael, and Florencio Lopez-de-Silanes. 1999. "The Benefits of Privatization: Evidence from Mexico." Quarterly Journal of Economics 114, no. 4: 1193-1242.

Levitsky, Steven. 2003. Transforming Labor-Based Parties in Latin America: Argentine Peronism in Comparative Perspective. Cambridge University Press.

Mangano, Gabriel. 1998. "Measuring Central Bank Independence: A Tale of Subjectivity and of Its Consequences." Oxford Economic Papers 50, no. 3: 468-92.

Maxfield, Sylvia. 1997. Gatekeepers of Growth: The International Political Economy of Central Banking in Developing Countries. Princeton University Press.

McNamara, Kathleen. 2002. "Rational Fictions: Central Bank Independence and the Social Logic of Delegation." West European Politics 25, no. 1 : 47-76.

Mehlum, Halvar, Karl-Ove Moene, and Ragnar Torvik. 2006. "Institutions and the Resource Curse." Economic Journal 116, no. 508: 1-20.

Mukand, Sharun, and Dani Rodrik. 2005. "In Search of the Holy Grail: Policy Convergence, Experimentation, and Economic Performance." American Economic Review 95, no. 1 : 374-83.

Mumtaz, Haroon, and Paolo Surico. 2006. "Evolving International Inflation Dynamics: World and Country Specific Factors." Working paper. London: Bank of England.

North, Douglass C., and Barry R. Weingast. 1989. "Constitutions and Commitment: The Evolution of Institutions Governing Public Choice in Seventeenth Century England." Journal of Economic History 49, no. 4: 803-32.

Oatley, Thomas. 1999. "Central Bank Independence and Inflation: Corporatism, Partisanship, and Alternative Indices of Central Bank Independence." Public Choice 98, no. 3/4: 399-413.

Panagiotidis, Theodore, and Afrioditi Triampella. 2006. "Central Bank Independence and Inflation: The Case of Greece." Revista de Economia del Rosario 9, no. 1: 95-109.

Persson, Torsten, Gerard Roland, and Guido Tabellini. 1997. "Separation of Powers and Political Accountability." Quarterly Journal of Economics 112, no. 4: 1163-1202.

Polillo, Simone, and Mauro Guillen. 2005. "Globalization Pressures and the State: The Worldwide Spread of Central Bank Independence." American Journal of Sociology 110, no. 6: 1764-1802

Reinhart, Carmen, and Kenneth S. Rogoff. 2004. "The Modern History of Exchange Rate Arrangements: A Reinterpretation." Quarterly Journal of Economics 119, no. 1: 1-48.

Reyes, Alvaro, Stefano Farne, Luis Angel Rodriguez, Angela Forero, and Emilio A. Carrasco. 1998. "Empleo, Productividad e Ingresos: Colombia 1990-1996." Documento de Trabajo 66. Lima: Organizacion Internacional del Trabajo, ETM Parses Andinos.

Roberts, Kenneth M. 1995. "Neoliberalism and the Transformation of Populism in Latin America: The Peruvian Case." World Politics 48, no. 1 : 82-116. --. 2008. Changing Course: Parties, Populism and Political Representation in Latin America's Neoliberal Era. Cambridge University Press.

Rodrik, Dani. 1996. "Understanding Economic Policy Reform." Journal of Economic Literature 34, no. 1: 9-41. --. 2005. "Growth Strategies." In The Handbook of Economic Growth, edited by Philippe Aghion and Steve Durlauf. Amsterdam: Elsevier.

Rogoff, Kenneth S. 1985. "The Optimal Degree of Commitment to an Intermediate Monetary Target." Quarterly Journal of Economics 100, no. 4:1169-89.

Shleifer, Andrei, and Robert Vishny. 1994. "Politicians and Firms." Quarterly Journal of Economics 109, no. 4: 995-1025.

Sierra Montoya, Jorge Emilio. 2004. 50 Protagonistas de la Economia Colombiana. Bogota: Universidad Jorge Tadeo Lozano.

Slok, Torsten. 2002. "Money Demand in Mongolia: A Panel Data Analysis." 1MF Staff Papers 49, no. l: 128-36.

Stigler, George. 1971. "The Theory of Economic Regulation." Bell Journal of Economics and Management Science 2, no. 1 : 3-21. --. 1972. "Economic Competition and Political Competition." Public Choice 13, no. 1: 91-106.

Stiglitz, Joseph E. 2002. Globalization and its Discontents. W. W. Norton.

Stock, James, and Mark Watson. 2007. "Why Has U.S. Inflation Become Harder to Forecast?" Journal of Money, Credit and Banking (Supplement) 39: 3-33.

van de Walle, Nicolas. 1993. "The Politics of Nonreform in the Cameroon." In Hemmed In: Responses to Africa's Economic Decline, edited by Thomas M. Callaghy and John Ravenhill. Columbia University Press. 2001. African Economies and the Politics of Permanent Crisis, 1979-1999. Cambridge University Press.

Velasco, Andres. 2005. "Why Doesn't Latin America Grow More, and What Can We Do about It?" Harvard University, Kennedy School of Government.

Williamson, John. 1990. "The Progress of Policy Reform in Latin America." Policy Analyses in International Economics 28. Washington: Institute for International Economics.

Wooldridge, Jeffery M. 2002. Econometric Analysis of Cross Section and Panel Data. MIT Press.

DARON ACEMOGLU

Massachusetts Institute of Technology

SIMON JOHNSON

International Monetary Fund

PABLO Q.UERUBIN

Massachusetts Institute of Technology

JAMESA. ROBINSON

Harvard University

(1.) Williamson (1990).

(2.) Van de Walle (2001, pp. 3-4).

(3.) Velasco (2005, p. 2).

(4.) See Rodrik (2005).

(5.) Stiglitz (2002, p. 16).

(6.) An alternative and complementary perspective is that reformers, aware of the political constraints underlying policy decisions, may push for reforms and emphasize their economic benefits as part of a bargaining process among political groups and policymakers. This perspective, although plausible, has not been developed in the literature and must also ultimately model the political economy of reform.

(7.) See, for example, Acemoglu (2006).

(8.) See, for example, North and Weingast (1989), Persson, Roland, and Tabellini (1997), Henisz (2000), and Acemoglu, Johnson, and Robinson (2005).

(9.) Van de Walle (1993, 2001).

(10.) Van de Walle (2001, p. 13).

(11.) Van de Walle (2001, p. 76).

(12.) The 1995 Reserve Bank of Zimbabwe Act legislated a greater degree of autonomy for the central bank. After 1995 the bank had its own budget and could decide on its own finances. The act also established the control of inflation as the unique objective of monetary policy.

(13.) This example and our results below raise the question of why potentially ineffective reforms are implemented in the first place. One obvious answer is that they are partly a response to external pressures. For example, Jacome (2001) documents how an International Monetary Fund technical assistance mission to Zimbabwe exerted pressure for the reform of central bank laws in 1995. This perspective suggests that externally imposed policy reforms might be less successful and effective than those generated by internal dynamics.

(14.) Inflation increased in the United Kingdom in the 1970s and early 1980s but never approached hyperinflationary levels. The desire to avoid a return of such episodes, as well as to realize the other benefits of CBI, such as greater credibility and transparency of monetary policy, might be among the reasons why countries with relatively good policies still prefer to implement central bank reform.

(15.) Acemoglu and others (2003).

(16.) A clear example is Argentina, where, shortly before the introduction of CBI in 1992, a currency board was established to peg the exchange rate to the dollar. Figure B2 in appendix B also shows that, in a number of countries, inflation begins to decline a few years before the introduction of CBI.

(17.) See, for example, Sierra Montoya (2004).

(18.) Jaramillo, Steiner, and Salazar (1997).

(19.) Cabrera Galvis and Ocampo (1980, p. 136).

(20.) Echeverry (2002); Davila Ladron de Guevara and others (2000).

(21.) Reyes and others (1998).

(22.) See Rodrik (1996) and Drazen (2000) for overviews.

(23.) See, for example, Alesina and Drazen (1991) and Fernandez and Rodrik (1991).

(24.) Mukand and Rodrik (2005).

(25.) For example, Dewatripont and Roland (1997).

(26.) Shleifer and Vishny (1994); Boycko, Shleifer, and Vishny (1996).

(27.) Coate and Morris (2006); Stigler (1971, 1972).

(28.) See Easterly (2005) for a review.

(29.) There is also convincing microeconomic evidence that some specific types of reforms, such as privatization, can have large beneficial effects. See, for example, La Porta and Lopez-de-Silanes (1999) and Galiani, Gertler, and Schargrodsky (2005).

(30.) For example, Bates and Krueger (1993).

(31.) Dollar and Svensson (2000).

(32.) Burnside and Dollar (2000).

(33.) Easterly, Levine, and Roodman (2004).

(34.) Bekaert, Harvey, and Lundblad (2005); Mehlum, Moene, and Torvik (2006).

(35.) Notably, Roberts (1995, 2008), Gibson (1997), and Levitsky (2003).

(36.) Most notable in this literature is Barro and Gordon (1983).

(37.) Rogoff (1985).

(38.) See Eijffinger and de Haan (1996) for an overview.

(39.) Alesina (1988); Grilli, Masciandaro, and Tabellini (1991); Alesina and Summers (1993).

(40.) Alesina and Summers (1993, p. 154)

(41.) Cukierman, Webb, and Neyapti (1992).

(42.) Cukierman (1992).

(43.) Various other papers, such as Gutierrez (2003) and Arnone and others (2007), report similar results.

(44.) Crowe and Meade (2007).

(45.) Campillo and Miron (1997); Oatley (1999).

(46.) Oatley (1999); Mangano (1998); Forder (1998); Banaian, Burdekin, and Willen (1998).

(47.) Keefer and Stasavage (2002, 2003).

(48.) Beck and others (2001).

(49.) This does not imply that CBI has no benefit in relatively developed countries. Given the lower inflation observed in OECD economies, the effect of CBI will be harder to detect, particularly if it is small. Moreover, CBI might create other benefits by introducing transparency and creating insurance against possible future relaxations of monetary policy.

(50.) Grossman and Helpman (1994).

(51.) Barro and Gordon (1983); Rogoff (1985).

(52.) Alesina and Drazen (1991).

(53.) In a richer model, strong constraints might make lobbies more powerful, because a well-meaning politician might be unable to act decisively to reduce inflation. However, we believe that this consideration is second-order because well-meaning politicians are relatively rare.

(54.) Yet another alternative, which in fact gives even more stark results, is to assume that these reforms act as costly commitment devices, and thus that they make distortionary policies more costly for the citizens. In this case policy reform would discourage distortionary policies by increasing the costs that these policies impose on the society, and indirectly on the politician. In the context of CBI, for example, high inflation becomes both more costly to society and potentially more costly to implement for the government, both because it will destroy the beneficial reputation that monetary policy may have established and because workers' and firms' behavior would have been shaped by expectations of low inflation. The assumption adopted in the text may have wider applicability, motivating our choice here.

(55.) To simplify the notation, this expression already imposes the choice that will prevail in equilibrium when the politician is indifferent.

(56.) This will also be true when inflation or other distortionary policies also become more costly for the society as a whole after reform. For example, inflation might be disastrous for the future of the economy and ruin the potential benefits that might have resulted from credibly establishing CBI (this can be incorporated by including [rho][pi] in the utility function of the citizens), but when [lambda] is low, this still will not deter politicians from using distortionary policies.

(57.) Grossman and Helpman (1994).

(58.) See Acemoglu, Robinson, and Verdier (2004) for a discussion.

(59.) Killick (1978, p. 35).

(60.) See, for example, Collier and Collier (1991).

(61.) See Arnone, Laurens, and Segalotto (2006) for a comprehensive overview.

(62.) Grilli, Masciandaro, and Tabellini (1991); Aufricht (1967).

(63.) Cukierman (1992); Cukierman, Webb, and Neyapti (1992).

(64.) Eight characteristics assessed the extent of limitations on central bank lending to the government, and these collectively received half of the weight in the index. Four legal characteristics that concern the way governors are appointed or dismissed were given one-fifth of the weight, three other characteristics that determined the degree of independence of the bank's policymaking process were given 15 percent of the total weight, and a measure related to the objectives of the central bank was given the remaining 15 percent.

(65.) Mangano (1998). In particular, from a comparison of the Cukierman and Grilli-Masciandaro-Tabellini (GMT) indices, Mangano (1998, pp. 476-77) concludes that "[there is] a significant degree of inconsistency between the two indices' valuation of their common criteria. In only one country out of seventeen and in the case of one criterion out of nine have Cukierman and GMT translated the legislation in exactly the same way: their interpretations of the laws governing the Italian central bank, and of the regulations concerning the central bank governors' terms of office in the countries sampled, exhibit no divergence. On the other hand, the average spread between their interpretations when examining Danish, French, Greek, and Japanese legislation is close to 50 percent, and they disagree in nearly 60 percent of countries when deciding whether the central bank is legally allowed to purchase government debt in the primary market. Overall, it appears that in the seventeen countries included in both [Cukierman's] and GMT's samples, virtually a third of the values attributed to their nine common criteria are subject to nonnegligible interpretation problems."

(66.) Exceptions that construct and use time-varying indices are Polillo and Guillen (2005), who provide values of the Cukierman index for the period 1989-2000 for a sample of ninety countries; Jacome and Vasquez (2005) for Latin America; Cukierman, Miller, and Neyapti (2002) for a sample of former Soviet countries; Arnone and others (2007) for a sample of emerging market countries; and Crowe and Meade (2007) for a sample of 102 countries worldwide. Both Arnone and others (2007) and Crowe and Meade (2007) explore time-series variation in these indices by taking values of them at two points in time (the late 1980s and the early twenty-first century) and assume that the change from one value to another occurs in the year in which central bank reform takes place.

(67.) Jacome and Vasquez (2005); Polillo and Guillen (2005).

(68.) For example, the Cukierman index of central bank independence reported by Polillo and Guillen (2005) is very low for most African countries, and reforms to the central bank charters during the 1990s caused only very small increases in these indices.

(69.) See the online appendix at econ-www.mit.edu/faculty/acemoglu.

(70.) Jacome and Vasquez (2005).

(71.) Polillo and Guillen (2005).

(72.) Arnone and others (2007).

(73.) Crowe and Meade (2007).

(74.) These data are maintained by the Center for Systemic Peace and the Center for Global Policy at George Mason University and are available at www.systemicpeace.org/polity/polity4.htm.

(75.) We use the average of constraints on the executive over 1972-2004 rather than its value at the beginning of the period or its year-to-year variation, because the average value of this variable appears to provide a better and less noisy measure of how constrained politicians are in a given country. The changes in this variable from year to year are subject to potential miscodings, which are averaged out when we consider the average constraints on the executive over a reasonable period of time.

(76.) Kaufmann, Kraay, and Mastruzzi (2007). One advantage of these indices is that they measure institutional quality around the time of CBI for most countries.

(77.) The rule-of-law question asks respondents the extent to which they have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, the police, and the courts, as well as the likelihood of crime and violence; the control-of-corruption question asks the extent to which respondents perceive public power as exercised for private gain, including both petty and grand forms of corruption as well as capture of the state by elites and private interests (Kaufmann, Kraay, and Mastruzzi, 2007, p. 4).

(78.) The inflation data reported for Qatar by IFS show an unusual spike in 1998 and 1999. Rather than verify and, where necessary, modify the IMF data for every country, we use the data exactly as reported. However, all our results are robust to excluding Qatar from the sample.

(79.) Normalized inflation is defined as inflation/(l + inflation); see equation 11 in the next section.

(80.) Government expenditure corresponds to "Total Expenditure and Net Lending by the General Government" (variable GGTENL in World Economic Outlook).

(81.) Reinhart and Rogoff (2004).

(82.) See Wooldridge (2002).

(83.) To calculate the economic magnitude of the coefficient estimates, we use average annual inflation over our sample period. If instead we use median annual inflation (8 percent instead of 58 percent), [[phi].sub.0] is even smaller and suggests that CBI reduces inflation by 4 percentage points. Throughout, the economic magnitudes implied by median inflation are about 60 percent smaller than those implied by mean inflation.

(84.) If we use median inflation to calculate the magnitude of the coefficient, the coefficient implies that CBI reduces inflation by 11 percentage points.

(85.) Kaufmann, Kraay, and Mastruzzi (2007).

(86.) See, for example, Wooldridge (2002, ch. 11).

(87.) Arellano and Bond (1991).

(88.) Reinhart and Rogoff (2004).

(89.) The coefficient in column 7-9 in the bottom panel, -0.224, implies that an increase of one standard deviation in the index is associated with a decrease of 11 percentage points in the inflation rate in countries with intermediate constraints. The coefficient in column 7-10, on the other hand, suggests that an increase of one standard deviation in the CBI index decreases inflation by 18 percentage points in the long run.

(90.) Mumtaz and Surico (2006); Stock and Watson (2007).

Table 1. OLS Fixed-Effects Regressions of
Inflation on Central Bank Independence (a)

 Full sample (52 countries)

Variable 1-1 1-2 1-3

CBI dummy, (b) short-run -0.036 -0.019 -0.031
 effect (0.034) (0.012) (0.022)
CBI dummy, long-run -0.079 -0.087
 effect
p-value, long-run effect [0.129] [0.126]
p-value, five lags of [0.000] [0.000]
 inflation = 0
p-value, current and [0.826]
 five lags of CBI
 dummy = 0
No. of observations 1,670 1,500 1,500
Adjusted [R.sup.2] 0.50 0.83 0.83

 Countries with change in
 CBI only (40 countries)

Variable 1-4 1-5 1-6

CBI dummy, (b) short-run -0.063 -0.028 -0.040
 effect (0.030) (0.013) (0.023)
CBI dummy, long-run -0.119 -0.164
 effect
p-value, long-run effect [0.051] [0.021]
p-value, five lags of [0.000] [0.000]
 inflation = 0
p-value, current and [0.072]
 five lags of CBI
 dummy = 0
No. of observations 1,300 1,172 1,172
Adjusted [R.sup.2] 0.47 0.84 0.84

Source: Authors' regressions.

(a.) The dependent variable is inflation/(1 + inflation), using
inflation data from International Monetary Fund, International
Financial Statistics. The sample period is 1972-2005. Each column
reports a single ordinary least squares regression using unbalanced
panel data with one observation per year, per country and including
country and year fixed effects. Robust standard errors, adjusted
for clustering by country, are in parentheses.

(b.) Takes a value of one in every year after a substantial reform
to the country's central bank laws leading to more independence is
introduced.

Table 2. OLS Fixed-Effects Regressions of Inflation on Central
Bank Independence Interacted with Executive Constraints

 Full sample

Variable 2-1 2-2 2-3

CBI dummy x weak-constraints dummy 0.023 -0.016 -0.051
 (0.025) (0.013) (0.032)
CBI dummy x medium-constraints dummy -0.071 -0.029 -0.048
 (0.044) (0.016) (0.033)
CBI dummy x strong-constraints dummy 0.023 0.003 0.011
 (0.027) (0.008) (0.011)
CBI x weak constraints, long-run -0.068 -0.023
 effect p-value, long-run effect, [0.203] [0.700]
 weak constraints = 0
CBI x medium constraints, long-run -0.119 -0.125
 effect p-value, long-run effect, [0.070] [0.056]
 medium constraints = 0
CBI x strong constraints, long-run 0.012 0.013
 effect p-value, long-run effect, [0.711] [0.751]
 strong constraints = 0
p-value, five lags of inflation = 0 [0.000] 10.000]
p-value, current and five lags of CBI [0.081]
 dummy, weak constraints = 0
p-value, current and five lags of CBI [0.4411
 dummy, medium constraints = 0
p-value, current and five lags of CBI [0.5001
 dummy, strong constraints = 0
p-value, medium effect = weak [0.017] [0.391] [0.087]
 effect (b)
p-value, medium effect = strong 10.016] [0.010] [0.004]
 effect (b)
No. of observations 1,670 1,500 1,500
Adjusted [R.sup.2] 0.51 0.83 0.83

 Countries with change
 in CBI only

Variable 2-4 2-5 2-6

CBI dummy x weak-constraints dummy 0.000 -0.026 -0.070
 (0.017) (0.014) (0.034)
CBI dummy x medium-constraints dummy -0.097 -0.039 -0.050
 (0.041) (0.017) (0.032)
CBI dummy x strong-constraints dummy -0.004 -0.007 -0.004
 (0.021) (0.008) (0.011)
CBI x weak constraints, long-run -0.104 -0.097
 effect p-value, long-run effect, [0.087] [0.133]
 weak constraints = 0
CBI x medium constraints, long-run -0.158 -0.196
 effect p-value, long-run effect, [0.033] [0.010]
 medium constraints = 0
CBI x strong constraints, long-run -0.028 -0.060
 effect p-value, long-run effect, [0.428] [0.171]
 strong constraints = 0
p-value, five lags of inflation = 0 [0.000] [0.000]
p-value, current and five lags of CBI [0.006]
 dummy, weak constraints = 0
p-value, current and five lags of CBI [0.035]
 dummy, medium constraints = 0
p-value, current and five lags of CBI [0.469]
 dummy, strong constraints = 0
p-value, medium effect = weak [0.014] [0.347] [0.098]
 effect (b)
p-value, medium effect = strong [0.017] [0.011] [0.005]
 effect (b)
No. of observations 1,300 1,172 1,172
Adjusted [R.sup.2] 0.49 0.84 0.84

Source: Authors' regressions.

(a.) The dependent variable is inflation/(1 + inflation), using
inflation data from International Monetary Fund, International
Financial Statistics. Each column reports a single ordinary least
squares regression using unbalanced panel data with one observation
per year per country and including country and year fixed effects.
Robust standard errors, adjusted for clustering by country, are in
parentheses. Assignment of countries to weak-, medium-, and
strong-constraints categories is based on the average of the
constraints on the executive (xcons) variable in the Polity IV data
for 1972-2004. All countries within one standard deviation of the
sample mean were assigned to the medium-constraints category.

(b.) p-values for are for the short-run effects in columns 2-1 and
2-4 and for the long run effects in the other columns.

Table 3. OLS Fixed-Effects Regressions of Inflation on Central Bank
Independence Interacted with Institutional Governance Measures (a)

 Full sample

 Coding based
 Coding based on control of
 oil rule of law corruption

Variable 3-1 3-2 3-3 3-4

CBI dummy x weak- -0.009 -0.064 0.003 -0.072
 constraints dummy (0.054) (0.055) (0.059) (0.074)
CBI dummy x medium- -0.069 -0.025 -0.062 -0.025
 constraints dummy (0.042) (0.022) (0.040) (0.019)
CBI dummy x strong- 0.022 0.004 0.020 0.003
 constraints dummy (0.030) (0.009) (0.030) (0.008)
CBI x weak constraints, -0.125 -0.096
 long-run effect p-value, [0.123] [0.244]
 long-run effect, weak
 constraints = 0
CBI x medium constraints, -0.084 -0.097
 long-run effect (b) p-value, [0.189] [0.123]
 long-run effect, medium
 constraints = 0
CBI x strong constraints, 0.001 0.001
 long-run effect (b) p-value, [0.986] [0.987]
 long-run effect, strong
 constraints = 0
p-value, five lags of [0.000] [0.000]
 inflation = 0
p-value, current and five [0.521] [0.0291
 lags of CBI dummy, weak
 constraints = 0
p-value, current and five [0.730] [0.592]
 lags of CBI dummy, medium
 constraints = 0
p-value, current and five [0.269] [0.3541
 lags of CBI dummy, strong
 constraints = 0
p-value, medium effect = [0.322] [0.586] [0.301] [0.985]
 weak effect (b)
p-value, medium effect = [0.018] [0.073] [0.024] [0.034]
 strong effect (b)
No. of observations 1,670 1,500 1,670 1,500
Adjusted [R.sup.2] 0.50 0.83 0.50 0.83

 Countries with change in CBI only

 Coding based
 Coding based oil control of
 on rule of law corruption

Variable 3-5 3-6 3-7 3-8

CBI dummy x weak- -0.035 -0.069 -0.023 -0.081
 constraints dummy (0.052) (0.055) (0.057) (0.073)
CBI dummy x medium- -0.095 -0.032 -0.088 -0.031
 constraints dummy (0.038) (0.021) (0.036) (0.018)
CBI dummy x strong- -0.004 -0.010 -0.006 -0.009
 constraints dummy (0.024) (0.011) (0.025) (0.011)
CBI x weak constraints, -0.193 -0.167
 long-run effect p-value, [0.044] [0.087]
 long-run effect, weak
 constraints = 0
CBI x medium constraints, -0.157 -0.170
 long-run effect (b) p-value, [0.032] [0.022]
 long-run effect, medium
 constraints = 0
CBI x strong constraints, -0.068 -0.070
 long-run effect (b) p-value, [0.136] [0.140]
 long-run effect, strong
 constraints = 0
p-value, five lags of [0.000] [0.000]
 inflation = 0
p-value, current and five [0.137] [0.006]
 lags of CBI dummy, weak
 constraints = 0
p-value, current and five [0.144] [0.098]
 lags of CBI dummy, medium
 constraints = 0
p-value, current and five [0.120] [0.292]
 lags of CBI dummy, strong
 constraints = 0
p-value, medium effect = [0.325] [0.637] [0.311] [0.960]
 weak effect (b)
p-value, medium effect = [0.019] [0.065] [0.025] [0.034]
 strong effect (b)
No. of observations 1,300 1,172 1,300 1,172
Adjusted [R.sup.2] 0.48 0.84 0.48 0.84

Source: Authors' regressions.

(a.) The dependent variable is inflation/(1 + inflation), using
inflation data from International Monetary Fund, International
Financial Statistic. Each column reports a single ordinary least
squares regression using unbalanced panel data with one observation
per year per country and including country and year fixed effects.
Robust standard errors, adjusted for clustering by country, are in
parentheses. Assignment of countries to weak-, medium-, and
strong-institutions categories is based on the average of the
rule-of-law index (columns 3-1, 3-2, 3-5, and 3-6) or the
control-of-corruption index (columns 3-3, 3-4, 3-7, and 3-8) from
Kaufmann, Kraay, and Mastruzzi (2007) for 1996-2005 (see appendix
table A1). All countries within one standard deviation of the
sample mean were assigned to the medium-institutions category.

(b.) p-values are for the short-run effects in columns 3-1, 3-3,
3-5, and 3-7 and for the long-run effects in the remaining columns.

Table 4. Arellano-Bond GMM Regressions of Inflation on Central
Bank Independence Interacted with Executive Constraints (a)

 Countries with
 change in
 Full sample CBI only

Variable 4-1 4-2 4-3 4-4

CBI dummy x weak- -0.010 -0.059 -0.050 -0.091
 constraints dummy (0.040) (0.048) (0.024) (0.042)
CBI dummy x medium- -0.046 -0.060 -0.053 -0.063
 constraints dummy (0.027) (0.037) (0.021) (0.032)
CBI dummy x strong- 0.032 0.034 0.010 0.012
 constraints dummy (0.018) (0.015) (0.015) (0.013)
CBI x weak constraints, -0.039 -0.049 -0.203 -0.203
 long-run effect
 p-value, long-run effect, [0.804] [0.725] [0.043] [0.035]
 weak constraints = 0
CBI x medium constraints, -0.181 -0.191 -0.217 -0.273
 long-run effect
 p-value, long-run effect, [0.068] [0.063] [0.034] [0.011]
 medium constraints = 0
CBI x strong constraints, 0.129 0.111 0.040 -0.014
 long-run effect
 p-value, long-run effect, [0.072] [0.186] [0.480] [0.830]
 strong constraints = 0
p-value, five lags of [0.000] [0.000] [0.000] [0.000]
 inflation = 0
p-value, current and five lags [0.042] [0.005]
 of CBI dummy,
 weak constraints = 0
p-value, current and five lags [0.541] [0.074]
 of CBI dummy,
 medium constraints = 0
p-value, current and five lags [0.210] [0.326]
 of CBI dummy,
 strong constraints = 0
p-value, medium effect = [0.387] [0.236] [0.890] [0.445]
 weak effect (b)
p-value, medium effect= [0.001] [0.000] [0.003] [0.001]
 strong effect (b)
Sargan test [0.00] [0.00] [0.00] [0.00]
Second-order serial correlation [0.59] [0.45] [0.99] [0.85]
No. of observations 1,358 1,358 1,060 1,060

Source: Authors' regressions.

(a.) The dependent variable is inflation/(1 + inflation), using
inflation data from International Monetary Fund, International
Financial Statistics. Each column reports a single regression using
unbalanced panel data with one observation per year per country and
including year dummies. Robust standard errors are in parentheses.
The GMM (Arellano-Bond) estimation method uses all available lags
of inflation as instruments.

(b.) p-values in all columns are for the long-run effects.

Table S. OLS Fixed-Effects Regressions of Inflation on Central Bank
Independence Interacted with a Quadratic for Executive Constraints (a)

 Full sample

Variable 5-1 5-2 5-3

CBI short-run effect at constraints = 1 0.217 0.000 -0.197
 p-value [0.264] [0.998] [0.152]
CBI short-run effect at constraints = 4 -0.075 -0.035 -0.073
 p-value [0.093] [0.056] [0.066]
CBI short-run effect at constraints = 7 0.038 0.005 0.017
 p-value [0.202] [0.544] [0.185]
CBI long-run effect at constraints = 1 0.001 0.189
 p-value [0.998] [0.290]
CBI long-run effect at constraints = 4 -0.145 -0.144
 p-value [0.046] [0.050]
CBI long-run effect at constraints = 7 0.019 0.020
 p-value [0.558] [0.633]
p-value, five lags of inflation = 0 [0.000] [0.000]
p-value, effect at constraints = 4 = [0.172] [0.5771 [0.221]
 effect at constraints = 1
p-value, effect at constraints = 4 = [0.017] [0.020] [0.014]
 effect at constraints = 7
Five lags of CBI dummy and linear No No Yes
 and quadratic interactions with
 constraints?
No. of observations 1,670 1,500 1,500
Adjusted [R.sup.2] 0.51 0.83 0.83

 Countries with
 change in CBI only

Variable 5-4 5-5 5-6

CBI short-run effect at constraints = 1 0.198 -0.006 -0.266
 p-value [0.300] [0.916] [0.048]
CBI short-run effect at constraints = 4 -0.099 -0.045 -0.079
 p-value [0.016] [0.018] [0.047]
CBI short-run effect at constraints = 7 0.012 -0.005 0.002
 p-value [0.617] [0.541] [0.861]
CBI long-run effect at constraints = 1 -0.026 0.094
 p-value [0.915] [0.515]
CBI long-run effect at constraints = 4 -0.183 -0.221
 p-value [0.0261 [0.010]
CBI long-run effect at constraints = 7 -0.020 -0.057
 p-value [0.540] [0.180]
p-value, five lags of inflation = 0 [0.000] [0.000]
p-value, effect at constraints = 4 = [0.164] [0.544] [0.255]
 effect at constraints = 1
p-value, effect at constraints = 4 = [0.019] [0.024] [0.015]
 effect at constraints = 7
Five lags of CBI dummy and linear No No Yes
 and quadratic interactions with
 constraints?
No. of observations 1,300 1,172 1,172
Adjusted [R.sup.2] 0.49 0.84 0.84

Source: Authors' regressions.

(a.) The dependent variable is inflation/(1 + inflation), using
inflation data from International Monetary Fund, International
Financial Statistics. Each column reports a single ordinary least
squares regression using unbalanced panel data with one observation
per year per country and including country and year fixed effects.
Robust standard errors, adjusted for clustering by country, are in
parentheses. All regressions include linear and quadratic
interactions of the CBI dummy with the average of the executive
constraints variable for 1972-2004.

Table 6. Regressions of Inflation on Five-Year Lead of Central
Bank Independence Interacted with Executive Constraints (a)

 Countries with
 Full sample change in CBI only

Variable 6-1 6-2 6-3 6-4

CBI dummy x weak-constraints -0.007 -0.044 -0.013 -0.048
 dummy (0.017) (0.020) (0.017) (0.019)
CBI dummy x medium-constraints -0.119 -0.035 -0.132 -0.042
 dummy (0.056) (0.020) (0.050) (0.020)
CBI dummy x strong-constraints 0.032 0.006 0.023 0.002
 dummy (0.020) (0.005) (0.019) (0.006)
CBI dummy, t + 5 x weak- 0.052 0.030 0.032 0.022
 constraints dummy (0.024) (0.015) (0.029) (0.014)
CBI dummy, t+ 5 x medium- 0.083 0.016 0.070 0.012
 constraints dummy (0.042) (0.010) (0.046) (0.010)
CBI dummy, t + 5 x strong- -0.001 0.002 -0.018 -0.006
 constraints dummy (0.019) (0.007) (0.024) (0.008)
p-value, five lags of [0.000] [0.000]
 inflation = 0
No. of observations 1,514 1,344 1,180 1,052
Adjusted [R.sup.2] 0.54 0.83 0.51 0.84

Source: Authors' regressions.

(a.) The dependent variable is inflation/(1 + inflation), using
inflation data from International Monetary Fund, International
Financial Statistics. Each column reports a single ordinary least
squares regression using unbalanced panel data with one observation
per year per country and including country and year fixed effects.
Robust standard errors, adjusted for clustering by country, are in
parentheses.

Table 7. Robustness Checks (a)

 Adding Adding
 time-varying region-specific
 covariates (b) quadratic trend (c)

Variable 7-1 7-2 7-3 7-4

 Full sample

CBI x weak-constraints -0.025 -0.054 0.025 -0.049
 dummy (0.028) (0.043) (0.031) (0.035)
CBI x medium-constraints -0.018 -0.034 -0.065 -0.045
 dummy (0.013) (0.025) (0.044) (0.033)
CBI x strong-constraints 0.001 0.007 0.023 0.012
 dummy (0.008) (0.010) (0.025) (0.011)
CBI x weak constraints, -0.021 -0.004
 long-run effect
 p-value, long-run effect, [0.856] [0.961]
 weak constraints = 0
CBI x medium constraints, -0.055 -0.090
 long-run effect
 p-value, long-run effect, [0.306] [0.165]
 medium constraints = 0
CBI x strong constraints, 0.016 0.029
 long-run effect
 p-value, long-run effect, [0.757] [0.426]
 strong constraints = 0
p-value, five lags [0.000] [0.000]
 of inflation = 0
p-value, current and five [0.449] [0.098]
 lags of CBI, weak
 constraints = 0
p-value, current and [0.603] [0.596]
 five lags of CBI,
 medium constraints = 0
p-value, current and [0.900] [0.449]
 five lags of CBI,
 strong constraints = 0
p-value, five lags of [0.011] [0.008]
 exchange rate
 flexibility index = 0
p-value, five lags [0.052] [0.079]
 of log(GDP) = 0
p-value, European
 Central Bank dummy
 and five lags = 0
p-value, medium effect = [0.794] [0.770] [0.014] [0.207]
 weak effect (g)
p-value, medium effect = [0.051] [0.144] [0.016] [0.006]
 strong effect (g)
No. of observations 1,323 1,323 1,670 1,500
Adjusted [R.sup.2] 0.871 0.871 0.509 0.833

 Countries with change in CBI only

CBI x weak-constraints -0.039 -0.069 -0.008 -0.073
 dummy (0.028) (0.042) (0.015) (0.037)
CBI x medium-constraints -0.033 -0.039 -0.098 -0.050
 dummy (0.013) (0.025) (0.037) (0.032)
CBI x strong-constraints -0.012 -0.006 0.006 0.000
 dummy (0.009) (0.012) (0.028) (0.011)
CBI x weak constraints, -0.096 -0.108
 long-run effect
 p-value, long-run effect, [0.385] [0.180]
 weak constraints = 0
CBI x medium constraints, -0.130 -0.179
 long-run effect
 p-value, long-run effect, [0.013] [0.0081
 medium constraints = 0
CBI x strong constraints, -0.061 -0.032
 long-run effect
 p-value, long-run effect, [0.228] [0.469]
 strong constraints = 0
p-value, five lags [0.000] [0.000]
 of inflation = 0
p-value, current and [0.226] [0.009]
 five lags of CBI,
 weak constraints = 0
p-value, current and [0.111] [0.067]
 five lags of CBI,
 medium constraints = 0
p-value, current and [0.663] [0.6521
 five lags of CBI,
 strong constraints = 0
p-value, five lags of [0.010] [0.009]
 exchange rate
 flexibility index = 0
p-value, five lags [0.167] [0.229]
 of log(GDP) = 0
p-value, European Central
 Bank dummy
 and five lags = 0
p-value, medium effect = [0.833] [0.742] [0.015] [0.303]
 weak effect (g)
p-value, medium effect = [0.041] [0.153] [0.006] [0.000]
 strong effect (g)
No. of observations 1,049 1.049 1,300 1,172
Adjusted [R.sup.2] 0.859 0.859 0.490 0.843

 Excluding Adding European
 Western Central Bank dummy
 Europe (d) and five lags (e)

Variable 7-5 7-6 7-7 7-8

 Full sample

CBI x weak-constraints 0.035 -0.052 0.030 -0.051
 dummy (0.034) (0.034) (0.028) (0.032)
CBI x medium-constraints -0.078 -0.050 -0.078 -0.048
 dummy (0.057) (0.039) (0.046) (0.033)
CBI x strong-constraints 0.008 0.022 0.009 0.010
 dummy (0.044) (0.026) (0.027) (0.011)
CBI x weak constraints, -0.013 -0.016
 long-run effect
 p-value, long-run effect, [0.853] [0.778]
 weak constraints = 0
CBI x medium constraints, -0.138 -0.131
 long-run effect
 p-value, long-run effect, [0.105] [0.045]
 medium constraints = 0
CBI x strong constraints, 0.016 -0.015
 long-run effect
 p-value, long-run effect, [0.790] [0.724]
 strong constraints = 0
p-value, five lags [0.000] [0.000]
 of inflation = 0
p-value, current and five [0.150] [0.087]
 lags of CBI, weak
 constraints = 0
p-value, current and [0.583] [0.397]
 five lags of CBI,
 medium constraints = 0
p-value, current and [0.082] [0.195]
 five lags of CBI,
 strong constraints = 0
p-value, five lags of
 exchange rate
 flexibility index = 0
p-value, five lags
 of log(GDP) = 0
p-value, European [0.009] [0.012]
 Central Bank dummy
 and five lags = 0
p-value, medium effect = [0.018] [0.067] [0.013] [0.059]
 weak effect (g)
p-value, medium effect = [0.109] [0.014] [0.021] [0.006]
 strong effect (g)
No. of observations 1,147 1,027 1,533 1,500
Adjusted [R.sup.2] 0.482 0.822 0.529 0.833

 Countries with change in CBI only

CBI x weak-constraints -0.002 -0.076 0.013 -0.069
 dummy (0.022) (0.036) (0.018) (0.033)
CBI x medium-constraints -0.121 -0.054 -0.099 -0.050
 dummy (0.052) (0.038) (0.042) (0.032)
CBI x strong-constraints -0.036 0.003 -0.012 -0.004
 dummy (0.038) (0.020) (0.020) (0.011)
CBI x weak constraints, -0.112 -0.087
 long-run effect
 p-value, long-run effect, [0.153] [0.1761
 weak constraints = 0
CBI x medium constraints, -0.234 -0.198
 long-run effect
 p-value, long-run effect, 10.0181 10.009]
 medium constraints = 0
CBI x strong constraints, -0.085 -0.084
 long-run effect
 p-value, long-run effect, [0.123] [0.067]
 strong constraints = 0
p-value, five lags [0.000] [0.000]
 of inflation = 0
p-value, current and [0.021] [0.010]
 five lags of CBI,
 weak constraints = 0
p-value, current and [0.050] [0.028]
 five lags of CBI,
 medium constraints = 0
p-value, current and [0.100] [0.256]
 five lags of CBI,
 strong constraints = 0
p-value, five lags of
 exchange rate
 flexibility index = 0
p-value, five lags
 of log(GDP) = 0
p-value, European Central [0.013] [0.010]
 Bank dummy
 and five lags = 0
p-value, medium effect = [0.016] [0.074] [0.012] [0.069]
 weak effect (g)
p-value, medium effect = [0.110] [0.019] [0.021] [0.008]
 strong effect (g)
No. of observations 892 800 1,190 1,172
Adjusted [R.sup.2] 0.460 0.830 0.514 0.842

 Using
 Cukierman
 CBI index (f)

Variable 7-9 7-10

 Full sample

CBI x weak-constraints 0.075 -0.127
 dummy (0.052) (0.080)
CBI x medium-constraints -0.182 -0.124
 dummy (0.115) (0.089)
CBI x strong-constraints 0.064 0.020
 dummy (0.052) (0.021)
CBI x weak constraints, 0.045
 long-run effect
 p-value, long-run effect, [0.712]
 weak constraints = 0
CBI x medium constraints, -0.294
 long-run effect
 p-value, long-run effect, [0.071]
 medium constraints = 0
CBI x strong constraints, 0.061
 long-run effect
 p-value, long-run effect, [0.406]
 strong constraints = 0
p-value, five lags [0.000]
 of inflation = 0
p-value, current and five [0.035]
 lags of CBI, weak
 constraints = 0
p-value, current and [0.664]
 five lags of CBI,
 medium constraints = 0
p-value, current and [0.403]
 five lags of CBI,
 strong constraints = 0
p-value, five lags of
 exchange rate
 flexibility index = 0
p-value, five lags
 of log(GDP) = 0
p-value, European
 Central Bank dummy
 and five lags = 0
p-value, medium effect = [0.017] [0.034]
 weak effect (g)
p-value, medium effect = [0.023] [0.009]
 strong effect (g)
No. of observations 1,670 1,500
Adjusted [R.sup.2] 0.508 0.833

 Countries with
 change in CBI only

CBI x weak-constraints 0.040 -0.154
 dummy (0.042) (0.082)
CBI x medium-constraints -0.224 -0.120
 dummy (0.114) (0.086)
CBI x strong-constraints 0.027 0.002
 dummy (0.043) (0.022)
CBI x weak constraints, -0.028
 long-run effect
 p-value, long-run effect, [0.825]
 weak constraints = 0
CBI x medium constraints, -0.362
 long-run effect
 p-value, long-run effect, [0.044]
 medium constraints = 0
CBI x strong constraints, 0.002
 long-run effect
 p-value, long-run effect, [0.982]
 strong constraints = 0
p-value, five lags [0.000]
 of inflation = 0
p-value, current and [0.072]
 five lags of CBI,
 weak constraints = 0
p-value, current and [0.345]
 five lags of CBI,
 medium constraints = 0
p-value, current and [0.632]
 five lags of CBI,
 strong constraints = 0
p-value, five lags of
 exchange rate
 flexibility index = 0
p-value, five lags
 of log(GDP) = 0
p-value, European Central
 Bank dummy
 and five lags = 0
p-value, medium effect = [0.017] [0.050]
 weak effect (g)
p-value, medium effect = [0.024] [0.011]
 strong effect (g)
No. of observations 1,242 1,127
Adjusted [R.sup.2] 0.494 0.849

Source: Authors' regressions.

(a.) The dependent variable is inflation/(1 + inflation), using
inflation data from International Monetary Fund, International
Financial Statistics. Each column reports a single ordinary least
squares regression using unbalanced panel data with one observation
per year per country and including country and year fixed effects.
The CBI variable is the CBI dummy as in previous tables except in
columns 7-9 and 7-10 (see note f). Robust standard errors. adjusted
for clustering by country. are in parentheses.

(b.) Regressions include five lags of log(GDP per capita), obtained
from the Penn World Tables 6.2, and five lags of the exchange rate
flexibility index constructed by Reinhart and Rogoff (2004),
available for the period 1972-2001.

(c.) Regressions include a quadratic time trend interacted with
OECD and Latin America dummies. d. Regressions exclude Austria,
Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy,
Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the
United Kingdom.

(e.) The European Central Bank dummy takes a value of one from 1999
onward for Austria. Belgium. Finland. France, Germany. Ireland.
Italy. Netherlands, Portugal, and Spain and from 2001 onward for
Greece.

(f.) Central bank independence is measured by the Cukierman index
provided by Crowe and Meade (2007) for 1989 and 2003. We assume
that the 1989 value holds for the period before central bank reform
and the 2003 value for the period after reform. We assign the
average value for Latin America to Ecuador, El Salvador, Guatemala,
and Paraguay for the prereform period (1989), because Crowe and
Meade (2007) provide a value for these countries for 2003 but not
for 1989. No values of the Cukierman index were provided for the
Dominican Republic, Guyana, and Mongolia for 1989 or 2003.

(g.) p-values are for the short-run effects in columns 7-1, 7-3.
7-5. 7-7, and 7-9 and for the long-run effects in the remaining
columns.

Table 8. OLS Fixed-Effects Regressions of Government
Expenditure on Central Bank Independence (a)

 Full sample

Variable 8-1 8-2 8-3

CBI dummy x weak- -0.047 0.003 -0.022
 constraints dummy (0.065) (0.012) (0.033)
CBI dummy x medium- 0.024 0.011 0.015
 constraints dummy (0.023) (0.004) (0.008)
CBI dummy x strong- -0.003 0.001 0.000
 constraints dummy (0.025) (0.005) (0.006)
CBI x weak constraints, 0.012 0.029
 long-run effect
p-value, long-run effect, [0.830] [0.426]
 weak constraints = 0
CBI x medium constraints, 0.050 0.047
 long-run effect
p-value, long-run effect, [0.013] [0.099]
 medium constraints = 0
CBI x strong constraints, 0.007 0.015
 long-run effect
p-value, long-run effect, [0.753] [0.646]
 strong constraints = 0
p-value, five lags of [0.000] [0.000]
 government expenditure = 0
p-value, current and five [0.000]
 lags of CBI dummy, weak
 constraints = 0
p-value, current and five [0.025]
 lags of CBI dummy, medium
 constraints = 0
p-value, current and five [0.523]
 lags of CBI dummy, strong
 constraints = 0
p-value, medium effect = [0.346] [0.510] [0.617]
 weak effect
p-value, medium effect = [0.300] [0.070] [0.277]
 strong effect
No. of observations 1,431 1,227 1,227
Adjusted [R.sup.2] 0.78 0.94 0.94

 Countries with
 change in CBI only

Variable 8-4 8-5 8-6

CBI dummy x weak- -0.033 0.005 -0.021
 constraints dummy (0.056) (0.011) (0.033)
CBI dummy x medium- 0.039 0.014 0.017
 constraints dummy (0.034) (0.005) (0.008)
CBI dummy x strong- 0.010 0.005 0.004
 constraints dummy (0.035) (0.005) (0.007)
CBI x weak constraints, 0.023 0.041
 long-run effect
p-value, long-run effect, [0.646] [0.328]
 weak constraints = 0
CBI x medium constraints, 0.066 0.062
 long-run effect
p-value, long-run effect, [0.008] [0.140]
 medium constraints = 0
CBI x strong constraints, 0.025 0.035
 long-run effect
p-value, long-run effect, [0.337] [0.451]
 strong constraints = 0
p-value, five lags of [0.000] [0.000]
 government expenditure = 0
p-value, current and five [0.009]
 lags of CBI dummy, weak
 constraints = 0
p-value, current and five [0.026]
 lags of CBI dummy, medium
 constraints = 0
p-value, current and five [0.534]
 lags of CBI dummy, strong
 constraints = 0
p-value, medium effect = [0.343] [0.467] [0.560]
 weak effect
p-value, medium effect = [0.282] [0.088] [0.349]
 strong effect
No. of observations 1,070 913 913
Adjusted [R.sup.2] 0.78 0.94 0.94

Source: Authors' regressions.

(a.) The dependent variable is government expenditure, from
International Monetary Fund, World Economic Outlook, as a percent
of GDP. Each column reports single ordinary least squares
regression using unbalanced panel data with one observation per
year per country and including country and year fixed effects.
Robust standard errors, adjusted for clustering by country, are
in parentheses.

Table 9. Arellano-Bond GMM Regressions of Government
Expenditure on Central Bank Independence Interacted
with Executive Constraints (a)

 Countries with
 Full sample change in CBI only

Variable 9-1 9-2 9-3 9-4

CBI dummy x weak-constraints 0.017 -0.008 0.008 -0.018
 dummy (0.010) (0.027) (0.007) (0.026)
CBI dummy x medium-constraints 0.022 0.020 0.017 0.019
 dummy (0.008) (0.008) (0.006) (0.008)
CBI dummy x strong-constraints 0.003 -0.001 -0.002 -0.004
 dummy (0.005) (0.007) (0.005) (0.008)
CBI x weak constraints, 0.079 0.119 0.035 0.050
 long-run effect
 p-value, long-run effect, [0.069] [0.008] [0.242] [0.111]
 weak constraints = 0
CBI x medium constraints, 0.105 0.111 0.075 0.063
 long-run effect
 p-value, long-run effect, [0.002] [0.016] [0.003] [0.160]
 medium constraints = 0
CBI x strong constraints, 0.013 0.027 -0.010 -0.006
 long-run effect
 p-value, long-run effect, [0.577] [0.385] [0.633] [0.871]
 strong constraints = 0
p-value, five lags of [0.000] [0.000] [0.000] [0.000]
 government expenditure = 0
p-value, current and five lags [0.000] [0.036]
 of CBI dummy,
 weak constraints = 0
p-value, current and five lags [0.024] [0.003]
 of CBI dummy, medium
 constraints = 0
p-value, current and five lags [0.487] [0.206]
 of CBI dummy, strong
 constraints = 0
p-value, medium [0.629] [0.855] [0.381] [0.707]
 effect = weak effect'
p-value, medium [0.000] [0.004] [0.000] [0.004]
 effect = strong effect'
Sargan test [0.00] [0.00] [0.001 [0.00]
Second-order serial correlation [0.41] [0.40] [0.25] [0.25]
No. of observations 1,137 1,137 848 848

Source: Authors' regressions.

(a.) The dependent variable is general government expenditure, from
International Monetary Fund, World Economic Outlook. Each column
reports a single regression using unbalanced panel data with one
observation per year per country and including year dummies. Robust
standard errors are in parentheses. The GMM (Arellano-Bond)
estimation method uses all available lags of government expenditure
as instruments.

(b.) p-values in all columns are for the long-run effects.

Table 10. Regressions of Government Expenditure on Five-Year Lead
of Central Bank Independence Interacted with Executive Constraints (a)

 Countries with
 Full sample change in CBI only

Variable 10-1 10-2 10-3 10-4

CBI dummy x weak- 0.061 0.018 0.069 0.018
 constraints dummy (0.035) (0.007) (0.039) (0.008)
CBI dummy x medium- 0.008 0.013 0.028 0.017
 constraints dummy (0.016) (0.004) (0.023) (0.005)
CBI dummy x strong- -0.021 -0.004 -0.011 -0.001
 constraints dummy (0.018) (0.004) (0.025) (0.004)
CBI dummy x weak- -0.140 -0.027 -0.111 -0.019
 constraints dummy, t + 5 (0.094) (0.021) (0.079) (0.018)
CBI dummy x medium- 0.015 -0.003 0.033 0.001
 constraints dummy, t + 5 (0.018) (0.009) (0.028) (0.012)
CBI dummy x strong- 0.025 0.005 0.052 0.011
 constraints dummy, t + 5 (0.021) (0.007) (0.032) (0.010)
p-value, five lags of [0.000] [0.000]
 government expenditure = 0
No. of observations 1,284 1,080 959 802
Adjusted [R.sup.2] 0.80 0.93 0.79 0.94

Source: Authors' regressions.

(a.) The dependent variable is general government expenditure, from
International Monetary Fund, World Economic Outlook, as a percent
of GDP. Each column reports a single ordinary least squares
regression using unbalanced panel data with one observation per
year per country and including country and year fixed effects.
Robust standard errors, adjusted for clustering by country, are in
parentheses.