Printer Friendly

Mobile Capital, Domestic Institutions, and Electorally Induced Monetary and Fiscal Policy.

The literature on global integration and national policy autonomy often ignores a central result from open economy macroeconomics: Capital mobility constrains monetary policy when the exchange rate is fixed and fiscal policy when the exchange rate is flexible. Similarly, examinations of the electoral determinants of monetary and fiscal policy typically ignore international pressures altogether. We develop a formal model to analyze the interaction between fiscal and monetary policymakers under various exchange rate regimes and the degrees of central bank independence. We test the model using data from OECD countries. We find evidence that preelectoral monetaty expansions occur only when the exchange rate is flexible and central bank independence is low; preelectoral fiscal expansions occur when the exchange rate is fixed. We then explore the implications of our model for arguments that emphasize the partisan sources of macroeconomic policy and for the conduct of fiscal policy after economic and monetary union in Europe.

Two literatures in political economy investigate the constraints imposed upon those who determine macroeconomic policy. One school argues that domestic institutions influence both the character of the policy process and the outcomes. Rules that require more than a simple majority for a measure to pass, such as the unanimity requirement on many issues within the European Union's Council of Ministers, or an increase in the number of institutional veto players make change in policy difficult (Garrett and Tsebelis 1996; Tsebelis 1995). Particular combinations of labor market institutions and the partisan composition of government have been linked to macroeconomic performance (Alvarez, Garrett, and Lange 1991). Voters are less likely to punish incumbents for poor macroeconomic performance when responsibility for outcomes is blurred by such factors as minority or coalition government (Powell and Whitten 1993). For similar reasons, governments in majoritarian systems find it difficult to maintain fixed exchange rate s (Bernhard and Leblang 1999). Central bank independence has been associated with price stability, higher real interest rates, lower variability in those rates, higher growth, greater real output per worker (Alesina and Summers 1993; Cukierman, Kalaitzidakis, Summers, and Webb 1993; Cukierman and Webb 1994; Cukierman, Webb, and Neyapti 1992; De Long and Summers 1992), and a propensity to put external adjustment before internal adjustment (Simmons 1996b). More recently, studies by a number of political scientists suggest that the effect of central bank independence on macroeconomic outcomes depends upon the structure of labor market institutions (Al Marhubi and Willett 1995; Franzese and Hall 1998; Garrett and Way 1995; Hall 1994; Iverson 1998). The common theme in this diverse literature is that institutions, in whatever form, alter the dynamics of the policymaking process and influence the outcomes it produces.

A second school explores whether the international environment limits states' freedom of action (Andrews 1994; Cerny 1994; Kurzer 1993; Moses 1994; Notermans 1993; Pauley 1995; Wallerstein and. Przeworski 1995; Webb 1995). Some analysts maintain that increased globalization requires governments to be more sensitive to the demands of fickle markets. Capital will flee states that levy high taxes on business and provide generous social benefits to labor in favor of states with lower tax burdens. States compete for mobile capital by lowering the tax burden and making deep cuts in welfare spending. Over time this competition will lead to the demise of the European social democratic model in favor of the Anglo-American liberal model (Cerny 1994; Sinn 1992; Strange 1996). Others recognize the growing integration of world trade and capital markets but disagree with the globalists about the consequences of these changes. They argue that states retain considerable room for maneuver in more open economies and, indeed, may respond to the risks of international openness by expanding the use of compensatory measures (Garrett 1995, 1998; Rodrik 1997).

These two literatures are not necessarily mutually exclusive. A third body of work examines the interaction between the two schools and considers how institutions refract changes in the international system.

Domestic political institutions can insulate policymakers from systemic changes or, conversely, can leave policymakers more exposed (Garrett and Lange 1995). A large number of domestic veto players also may inhibit states from adjusting policies to changes in their competitors' positions (Hallerberg and Basinger 1998). The general conclusion is that internationalization does not translate frictionlessly into policy change.

Following this last group of authors, we examine how the interaction of the international environment with domestic political institutions constrains a government's ability to engineer preelectoral macroeconomic expansion. It has long been argued that incumbents who want to win the next election may manipulate economic tools at their disposal in an attempt to satisfy the electorate enough to ensure their reelection (Keech 1995, chap. 3; MacRae 1977; Nordhaus 1975; Tufte 1978). Clark and Reichert (1998) find that political business cycles (PBCs) are almost entirely absent in states with independent central banks. They also find that opportunistic cycles are not likely to occur when capital is mobile and the exchange rate is fixed.

One aspect of this finding is curious, however. There are good reasons to expect the identified constraints to make it difficult to use monetary policy for electoral purposes, but there is no a priori reason to expect fiscal policy to be constrained under such circumstances. [1] Even when a central bank has complete control over monetary policy, the government can still attempt to affect macroeconomic outcomes through cuts in taxes or increases in expenditures. Similarly, although monetary policy may be constrained when exchange rates are fixed and capital is mobile, incumbents are free to use fiscal policy to create preelectoral expansions. The temptation to use these instruments before elections in order to sway undecided voters may be especially high; in particular, discussions of and attempts to reduce taxes often seem to dominate political campaigns.

Open economy models often privilege monetary policy over fiscal policy, but the literature on the domestic determinants of fiscal policy tends to ignore the international environment in which policy is made (an exception is Oatley [1999]). This may be one reason these studies have produced inconclusive results. For example, Burdekin and Laney (1988) and Franzese (1996) find that central bank independence is related to lower deficits, but several others have not confirmed the relationship (De Haan and Sturm 1994a; Grilli, Masciandaro, and Tabellini 1991; Pollard 1993 [the first and last of these as quoted in Eijffinger and De Haan 1996]). The Mundell-Fleming model indicates that governments can, at best, pursue only two of the following three: capital mobility, fixed exchange rates, and independent monetary policy. [2] The absence or presence of each of these conditions affects the nature of the game between a government, which we assume wants to use fiscal and monetary instruments to stimulate the economy sh ortly before elections, and a central bank, which, so long as it is independent from the government, is immune from electoral pressures. In particular, an independent central bank may be able to prevent opportunistic business cycles when it has control over monetary policy. Such control is possible when capital is immobile, as was generally the case in the industrialized world in the 1960s, or when exchange rates are flexible, as is the case in several countries since the collapse of the Bretton Woods system. When capital is mobile and exchange rates are fixed, however, monetary policy is ineffective, and a central bank, independent or not, cannot deter fiscal expansions. These different expectations based on the Mundell-Fleming conditions may explain why the empirical evidence so far has been mixed.

The article is organized as follows. To sort out the effects of elections, central bank independence, capital mobility, and exchange rate regime on the strategic interaction between fiscal and monetary policymakers, we develop a game theoretic model of the interaction between the government and a potentially independent central bank. We begin the analysis in a closed economy, in which capital is not mobile and exchange rates are fixed, and we find that central bank independence is important in preventing opportunistic monetary cycles but may exacerbate the size of fiscal cycles. We then consider how different Mundell-Fleming conditions affect the game. When exchange rates are fixed and capital is mobile, the international environment restrains governments from pursuing monetary opportunism, but states are expected to engineer preelectoral fiscal expansions. Once again, the independence of the central bank has no causal effect on the occurrence of fiscal cycles. We next examine the predictions of the model wi th quarterly data from the OECD from 1973 to 1989 for monetary policy and with annual data from the OECD and European Union sources from 1981 to 1992 for fiscal policy. The results largely confirm the theoretical model: There is evidence of preelectoral monetary expansions only when the exchange rate is flexible and central bank independence is low; preelectoral fiscal expansions occur when the exchange rate is fixed. We then consider whether there is evidence for partisan political business cycles that is consistent with our model. Based on the Mundell-Fleming model, Oatley (1999) contends that clear partisan differences existed before the 1990s for fiscal and monetary policy depending upon the underlying economic conditions. Differences between the Left and the Right were present for fiscal policy in the absence of full capital mobility and flexible exchange rates and were present for monetary policy in the absence of capital mobility and fixed exchange rates. Unlike Oatley, we find no evidence either for m onetary or for fiscal policy. Finally, we argue that political business cycles are likely to become more frequent and worrisome under the fixed exchange rate regime established by the European Economic and Monetary Union.

THE MODEL

Assumptions

Domestic Institutions. "Government" broadly construed can use some combination of monetary and fiscal policy to affect the economy. Monetary policy instruments (the discount rate, purchase and sale of government securities and foreign currencies) can be aimed at policy targets (such as interest rates, the money supply, and exchange rates) in an attempt to produce desired macroeconomic outcomes (growth, price and exchange rate stability, and/or stable balance of payments). The central bank (CB), not elected politicians, ultimately sets the policy instruments. This is not to say that elected officials are irrelevant to the conduct of monetary policy. The extent of their influence varies from country to country and across time. In Germany and the United States, for example, central banks have considerable latitude both in selecting instruments and in choosing targets and goals, whereas in Britain and Norway the central banks traditionally have had much less freedom of action. Although statute and tradition may s upport a high degree of autonomy, even the Federal Reserve and the Bundesbank are only relatively autonomous. Central bankers are appointed by elected officials, and any latitude they enjoy is granted (and potentially taken away) by elected officials. [3]

Fiscal policy manipulates taxation and expenditure in an attempt to produce desired macroeconomic outcomes. [4] The taxing and spending powers of the central government are controlled by elected officials, but they rely upon the finance ministry for crucial information, analysis, and day-to-day implementation. We adopt the simplifying assumption that the minister of finance is the dutiful agent of the government. The central banker and the minister of finance not only have separate responsibilities and information but also differ in how they evaluate alternative macroeconomic outcomes. We adopt the standard political business cycle assumption that elected officials derive sufficient benefits from electorally timed increases in growth and employment that they heavily discount the negative effect of such behavior on inflation rates. In contrast, as much of the literature argues, decision makers in central banks tend to be more inflation averse than their principals and, therefore, are likely to discount the va lue of electorally induced expansions. [5]

We represent actor preferences over macroeconomic outcomes with the following quadratic loss function, which describes how actors assess deviations from their ideal values for unemployment and inflation:

[L.sub.i] = [(y - [y.sup.*i]).sup.2] + [[alpha].sup.i][([pi] - [[pi].sup.*i]).sup.2], (1)

where [pi] and y are inflation and output, [[pi].sup.*] and [y.sup.*] are the actor's ideal points, [alpha] is the weight the actor attaches to inflation stabilization relative to achieving his or her output goals, and i indexes the actor. Deviations from ideal points are squared to capture the intuition that small deviations result in a proportionately smaller utility loss than large ones.

Differences between and among actors in their macroeconomic preferences can be captured by varying assumptions about particular parameters in the model. We emphasize differences between the government and the central bank concerning their ideal rate of growth, but we assume that actors share an ideal point of zero inflation. [6] In addition, both central bankers and governments would like to stabilize the economy at the natural rate of growth, that is, the rate consistent with price stability. The opportunistic political business cycle argument, however, suggests that incumbents are tempted to try to push output above the natural rate just before elections in order to increase electoral support before inflationary consequences are felt. Thus, elections induce a change in preferences, such that

[y.sup.*gov] = [k.sup.gov][y.sup.n], (2)

where [y.sup.n] is the natural rate of growth, and k is a coefficient that equals 1 during nonelectoral periods and is greater than 1 in electoral periods. [7] In the absence of central bank independence, these inflationary pressures are transmitted to the central bank through rewards and punishments the government has at its disposal. The central bank then acts "as if" it were minimizing the government's electorally induced loss function. [8] In the absence of independence, the central bank's growth target is as defined in equation 2. If the bank is independent, we assume that it is insulated from government pressure. [9] Consequently, the bank's growth target will be the same in electoral periods as in nonelectoral periods, such that

[y.sup.*cb] = [k.sup.cb][y.sup.n], K = 1. (3)

The government and central bank both effect changes in macroeconomic outcomes through the manipulation of a short-term, expectations-enhanced Phillips curve:

y = [y.sup.n] + [mu]([pi] - [[pi].sup.e]) + [phi]g, (4)

where [[pi].sup.e] is the rate of inflation expected to prevail at time t, embodied in sticky wage contracts signed at time t - 1; g is net government spending; and [mu] and [phi] are coefficients that characterize the effectiveness of fiscal and monetary policy transmission. The central bank is assumed to control inflation directly, [10] whereas the government controls spending.

The International Environment. Based on the standard Mundell-Fleming model, the effectiveness with which a government or central bank can manipulate these instruments depends upon the level of capital mobility as well as the exchange rate regime. The Mundell-Fleming model indicates that both instruments are somewhat effective when there is no capital mobility. When capital becomes mobile, however, fiscal policy is effective only when exchange rates are fixed, and monetary policy is effective only when exchange rates are flexible. [11]

Our model captures the salient aspects of the international environment in the following ways. When capital is immobile, fiscal and monetary policy are only partially effective (0 [less than] [mu] [less than] 1, 0 [less than] [phi] [less than] 1). When capital is fully mobile and exchange rates are fixed, monetary policy is assumed to be ineffective, and fiscal policy is assumed to be hypereffective (standardized here so that [mu] = 0 and [phi] = 1). When capital is fully mobile and exchange rates are not fixed, fiscal policy is ineffective, and monetary policy is hypereffective ([mu] = 1 and [phi] = 0). [12]

Order of Play. At the start of the game, the players learn about the structure of the game, that is, they learn whether it occurs in an electoral period, the central bank is independent, capital is mobile, and the exchange rate is fixed. The government then chooses the level of net spending by passing a budget. The central bank observes the budget and then chooses the rate of inflation by deciding whether to raise or lower interest rates or to expand or contract the money supply.

Equilibria

We will solve the game for electoral and nonelectoral periods under alternative assumptions about the independence of the central bank. We first address the closed economy case, then consider the conditional effects of capital flows.

PROPOSITION 1. In general, the government adopts a budget (g), such that net spending is an increasing function of the propensity to push growth above the natural rate and a decreasing function of unanticipated inflation:

g = [frac{1}{[phi]}][[y.sup.n](k - 1) - [mu]([pi] - [[pi].sup.e])]. (5)

The central bank responds by setting inflation ([pi]), such that it increases with inflationary expectations and the propensity to push output above the natural rate and decreases with the net budget deficit:

[mu] = [frac{1}{[mu] + [frac{[alpha]}{[mu]}]}] [[mu][[pi].sup.e] + [y.sup.n](k - 1) - [phi]g]. (6)

Derivation: See Appendix A.

Proposition 1 can be used to derive predictions about central bank and government behavior under different sets of (ideal typical) structural conditions. Specifically, we consider the effect of elections on government and bank behavior and examine whether this effect is modified in important ways by central bank independence and the satisfaction of Mundell-Fleming conditions for national policy autonomy. To establish a baseline, we first discuss the strategic interaction between the government and the central bank in a hypothetical closed economy. We will then consider the effects of different exchange rate regimes in the context of fully mobile capital.

Strategic Interaction under Imperfectly Mobile Capital. In a closed economy, both fiscal and monetary policy are assumed to have some effect on output (0 [less than] [mu] [less than] 1, 0 [less than] [phi] [less than] 1). In the absence of central bank independence, both fiscal and monetary actors feel pressure to push output above the natural rate (k [greater than] 1) during electoral periods. Under these circumstances, both actors' reaction functions are as described in proposition 1. The government anticipates the central bank's response and incorporates it into the budget, setting net government spending so that it is a function of the propensity to push the economy above the natural rate of growth and inflationary expectations (g = (1/[phi]) [[y.sup.n](k - 1) + [mu][[pi].sup.e]]). The central bank observes the level of spending chosen by the government and responds by setting inflation equal to zero ([pi] = 0). Since k = 1 for both actors during nonelectoral periods, the closed economy nonelectoral equil ibrium is a function of expected inflation (g = (1/[phi])[mu][[pi].sup.e]) while inflation is zero ([pi] = 0). The effect of elections on monetary and fiscal policy is simply the difference between the equilibrium policies in the electoral and nonelectoral periods, as shown in column F of Table 1. Hypothesis 1 summarizes the model's primary predictions regarding the effect of elections on fiscal and monetary policy during electoral periods when capital is immobile and the central bank is dependent.

HYPOTHESIS 1. When capital is immobile and the central bank is dependent, the government initiates a fiscal expansion during electoral periods, and the central bank maintains its nonelection period policies. Fiscal, but not monetary, expansions are expected during electoral periods.

When the central bank is independent, equation 6 is reduced to [pi] = (l([mu]+[alpha]/[mu]))[[mu][[pi].sup.e] - [phi]g] during electoral periods. The government anticipates this and adopts a budget that increases with both inflationary expectations and the propensity to push growth above the natural rate (g = (1/[phi])[mu][(([mu]+[alpha])/[alpha])[y.sup.n](k - 1) + [[pi].sup.e]]). The central bank responds in an offsetting manner by choosing a rate of inflation that is a decreasing function of the government's propensity to push growth above the natural rate [[pi].sub.t] = (-1/(([mu]+[alpha])/[mu]))[[mu](([mu]+[alpha])/[alpha])[y.sup.n](k - 1)]. Once again, the absence of elections implies that k = 1, so in the nonelectoral equilibrium with an independent central bank, the government sets the net budget deficit in response to inflationary expectations (g = (1/[phi])[mu][[pi].sup.e]), and the central bank sets inflation equal to zero ([[pi].sub.t] = 0), which is the same as the nonelectoral period with a depen dent central bank. As before, the effect of elections on monetary and fiscal policies in the presence of central bank independence can be seen in column F of Table 1 and is summarized in the following hypothesis.

HYPOTHESIS 2. When capital is immobile and the central bank is independent, the government initiates a fiscal expansion during electoral periods, and the central bank responds with a monetary contraction. Fiscal expansions and monetary contractions are expected in electoral periods.

A number of further implications can be drawn from the closed economy equilibrium. First, when the central bank is under the influence of the government, it sets inflation equal to zero in both electoral and nonelectoral periods and allows the government to pursue its electoral goals unimpeded. This particular solution to the "assignment" problem is fairly arbitrary and is the result of the way our model is constructed. It is possible to construct a model in which the government "gets out of the way" and lets a subservient central bank achieve the government's electoral goals through interventionist monetary policy or a model in which the government and bank coordinate their actions around whatever policy mix is optimal, given the relative effectiveness of fiscal and monetary instruments. What is significant, however, is that when the bank is subservient, the government's goals are pursued as if one actor were making policy with two goals and two instruments.

The policy dynamic is starkly different when the central bank is independent. The bank initiates a monetary contraction during electoral periods in order to offset the government's electorally induced fiscal expansion. The size of the expansion increases in accord with the effectiveness of monetary policy and decreases with the weight policymakers assign to price stabilization compared to growth targets. To understand why, it is useful to think of the effectiveness of monetary policy ([mu]) as the "price" (in terms of inflation) policymakers pay for an additional increment of output (refer to equation 5). Consequently, an independent central bank is more willing to use contractionary measures to discipline the government when monetary policy is effective than when it is not. Since the government anticipates a more aggressive stance by the central bank when monetary policy is effective, it counters with a more expansionary policy than it would otherwise adopt. [13] When monetary policy is relatively ineffecti ve, the government cannot count on a large offsetting monetary contraction, so it restrains its expansionary impulse. Since fiscal expansions decrease in accord with the weight policymakers place on price stabilization ([alpha]), the net effect of central bank independence on an electorally induced fiscal expansion depends on the relationship between these two parameters. Specifically, if ([[mu].sup.2]/([alpha] + [mu])) is greater than 1, then preelectoral fiscal expansions are larger when the central bank is independent than when it is not.

Equilibrium behavior under imperfectly mobile capital also suggests that net spending decreases as a result of increases in fiscal policy effectiveness ([phi]). As fiscal policy becomes less effective, the levers of policy have to be moved farther in order to hit any given set of targets. This result has profound implications for empirical attempts to gauge the effect of increased capital mobility on the use of fiscal instruments. Garrett (1995), for example, argues that global financial integration has not reduced the effectiveness of fiscal interventions aimed at shielding domestic populations from the vagaries of the international economy (Katzenstein 1985). In fact, Garrett believes there is evidence of increased activism, presumably as the result of governments responding to the increased risks posed by globalization (see also Rodrik 1997). Our model suggests that the apparent evidence of increased fiscal activism may be as much a sign of growing policy ineffectiveness as of continued support for the "c ompensatory hypothesis."

The results of the model under imperfectly mobile capital are sufficiently provocative to warrant future research, but our concern here is with how the international environment, and more specifically capital mobility, constrains policymakers. We now turn to the sharper conclusions of the model under the assumption that capital is fully mobile.

Fixed Exchange Rates and Fully Mobile Capital. According to the Mundell-Fleming approach to the balance of payments, national monetary policy is ineffective when exchange rates are fixed and capital is fully mobile, but fiscal policy increases in effectiveness under these conditions. [14] We have standardized the parameters of monetary and fiscal policy effectiveness so that, when these conditions are met, [mu] = 0 and [phi] = 1. To deduce the behavior of the government and the central bank, we need only substitute these values into the reaction functions given in proposition 1 and solve for Stackelberg equilibria, [15] as in the previous case of imperfectly mobile capital.

When exchange rates are fixed, capital is fully mobile, and the central bank is dependent, the government's reaction function reduces to its propensity to push growth above the natural rate (g = [y.sup.n](k - 1)), and the central bank sets inflation equal to zero. During nonelectoral periods the government's reaction function reduces to g = 0, and the central bank continues to adopt a policy of zero inflation. As before, the effect of elections on monetary and fiscal policy under these conditions can be seen in column F of Table 1 and is summarized in the following hypothesis.

HYPOTHESIS 3. When capital is mobile and the exchange rate is fixed, the government initiates a fiscal expansion during electoral periods, but the central bank maintains its nonelection period policies (whether or not it is independent). Fiscal, but not monetary, expansions are expected during electoral periods.

This result has an important implication for the interaction between the international environment and the central bank. Since the bank does not have effective control over domestic monetary policy when the exchange rate is fixed, it no longer conditions its behavior on that of the government. Thus, it is not surprising that a change from a dependent to an independent central bank is not expected to affect the behavior of either fiscal or monetary policymakers.

Flexible Rates and Fully Mobile Capital. The situation is markedly different under flexible exchange rates. In that case, the Mundell-Fleming model tells us that fiscal policy will be ineffective, and monetary policy will be more effective ([mu] = 1 and [phi] = 0). If we insert these values into the government's loss function, the first-order condition becomes

[frac{[partial][L.sup.gov]}{[partial]g}] = 0. (7)

Thus, the ineffectiveness of fiscal policy leads the government to abandon the manipulation of budgets for electoral purposes.

The central bank pursues its best policy in view of zero net spending by government. During nonelectoral periods, the bank sets inflation in response to inflationary expectations ([[pi].sub.t] = (1/(1+[alpha])) [[pi].sup.e] (refer to Table 1). During electoral periods, inflation is an increasing function of the government's propensity to push growth above the natural rate, if the central bank is dependent ([[pi].sub.t] = (1/(1 + [alpha]))([[pi].sup.e] + [y.sup.n](k - 1))), but not otherwise ([pi] = (1/(1+[alpha])) [[pi].sup.e]) for both electoral and nonelectoral periods if the bank is independent.

HYPOTHESIS 4. When capital is mobile and the exchange rate is allowed to fluctuate, the government maintains its nonelection period policies throughout the electoral period. If it is independent, then the central bank also maintains its nonelection period policies; if it is dependent, the central bank initiates a monetaty expansion during electoral periods. Fiscal expansions during electoral periods are not expected. Monetary expansions are expected during electoral periods only if the central bank is dependent.

Summary and Comparison of Results under Alternative Structural Conditions

The model's predictions regarding the effects of elections on monetary and fiscal policies under various structural conditions were reported in column F of Table 1. Comparisons across the rows in that table produce comparative statics related to the effects of structural change on the policy consequences of elections. According to hypotheses 1 and 2, net government spending will be higher in electoral periods than in nonelectoral periods if there are

substantial barriers to capital mobility. This is true regardless of central bank independence. Similarly, hypothesis 3 indicates that net government spending will increase in electoral periods when capital is mobile and exchange rates are fixed (the increase is likely to be greater than in the closed economy case). In contrast, hypothesis 4 suggests that no fiscal expansion is expected when capital is mobile and exchange rates are allowed to float. Note that central bank independence is not important in determining the absence or presence of fiscal cycles when capital is fully mobile.

Unlike fiscal policy, electorally induced monetary cycles are very sensitive to the degree of central bank independence as well as the nature and degree of integration with the international economy. Hypotheses 1 and 2 tell us that electorally induced monetary expansions are not expected when capital mobility is limited. Indeed, monetary contractions are expected when capital is immobile and the central hank is independent. When capital is mobile, both the degree of central bank independence and the nature of the exchange rate regime condition the effect of elections on monetary policy. Hypotheses 3 and 4 indicate that elections are expected to induce monetary expansion only when the exchange rate is flexible and central bank independence is limited. Otherwise, monetary policy is expected to be free from electoral pressures when capital is fully mobile. [16]

A TEST OF THE MODEL'S IMPLICATIONS UNDER FULL CAPITAL MOBILITY

The standard method for testing opportunistic political business cycle arguments is to examine the relationship between various macroeconomic outcomes (such as growth, unemployment, and inflation) and the occurrence of elections. Because our focus is on the strategic interaction between fiscal and monetary policymakers, we examine the instruments that the respective agents are presumed to control. In the preceding theoretical discussion, we maintained the fiction that the central bank controls the rate of inflation directly. We now make the more realistic assumption that the central bank controls the money supply. The government is assumed to control the size of the budget surplus (or deficit).

The model put forth earlier suggests that the relationship between elections and policy instruments depends upon the degree of capital mobility, the exchange rate regime, and the degree of central bank independence. Although we modeled the strategic interaction of monetary and fiscal actors under the hypothetical case of a closed economy to provide an analytical baseline, we will test only hypotheses 3 and 4, which assume fully mobile capital. Following Clark and Reichert (1998), we presume that capital was highly mobile when the Bretton Woods System collapsed. Consequently, our empirical examination concentrates on the period 1973-95. This allows us to treat capital mobility as essentially constant and examine the effects of exchange rate regimes and central bank independence on monetary and fiscal cycles. This coding has limitations, but it follows a tradition that treats capital mobility as a systemwide, rather than country by country, variable (Andrews 1994; Frieden 1991; Hallerberg and Clark 1997; Kurze r 1993; McNamara 1998; Webb 1995). To be sure that this assumption is not what drives our results, we also report regressions that consider the effects of country-specific restrictions on capital movements.

Again following Clark and Reichert (1998), we use a dummy variable interaction model to examine the modifying effects of fixed exchange rates and central banking institutions. We use their codings of participation in fixed exchange rate regimes as well as their dummy variable for central bank independence, updating where necessary. [17] The standard test for political business cycles employs a multivariate regression model aimed at isolating the relationship between elections and macroeconomic variables. We will examine two sets of models to evaluate hypotheses related to budgetary and monetary cycles. Table 2 summarizes the predictions of Table 1 regarding electorally induced monetary and fiscal cycles under full capital mobility.

Monetary Cycles

Our theoretical discussion suggests that the existence of electorally induced monetary cycles is likely to be sensitive to the environment. We will test this using a pooled cross-sectional time-series model that extends the empirical work of Alesina and Roubini (1997). [18] They test for monetary cycles using the following equation:

[m.sub.it] = [[beta].sub.0] + [[beta].sub.1][m.sub.it-1] + [[beta].sub.2][m.sub.it-2] + ... + [[beta].sub.n][m.sub.it-n] + [[beta].sub.n+1][PBCN.sub.it] + [[varepsilon].sub.t], (8)

where [m.sub.it] is the growth rate of money for country i at time t, and PBCN is an electoral dummy variable that equals 1 during electoral quarters and in either (depending on the specific test) the three or five quarters before the election. Adding dummy variables and interaction terms to capture the context-specific effects of our model, equation 8 becomes:

[m.sub.t] = [[beta].sub.0] + [[beta].sub.1][E.sub.t] + [[beta].sub.2]Cbi + [[beta].sub.3][Fixed.sub.t] + [[beta].sub.4]E [cdotp] [Cbi.sub.t] + [[beta].sub.5]E [cdotp] [Fixed.sub.t] + [[beta].sub.6]CBI [cdotp] [Fixed.sub.t] + [[beta].sub.7]E [cdotp] CBI [cdotp] [Fixed.sub.t] + [Sigma]([[beta].sub.j][m.sub.t-j]) + [e.sub.t]. (9)

Table 3 reports the pooled cross-sectional time-series results for four specifications. Following Alesina and Roubini (1997), models A and B use the annual rate of change in M1 as an indicator of change in the money supply. To ensure that the results are not driven by the way we operationalize the dependent variable, models C and D use the natural log of a monetary aggregate that includes M1 and quasimoney as the dependent variable. [19] Columns C and D also differ from Alesina and Roubini's specification in that we control for macroeconomic conditions that are expected to influence monetary policy. Specifically, it is expected to respond in a countercyclical manner to increased unemployment and inflation with a one-period lag. [20] Columns A and C use qualitative indicators of the hypothesized constraints on monetary policy. These specifications implicitly assume that capital was uniformly mobile during the period examined and that central banks were either independent or not. The models in columns B and D r elax these assumptions by using continuous measures of capital mobility and central bank independence. [21] All models include country dummies to control for country-specific fixed effects as well as a single lagged dependent variable to control for autocorrelation. Since the coefficients on the lagged dependent variables are several standard errors away from one, we are confident that unit root and cointegration problems do not exist. Standard errors were calculated using Beck and Katz (1995) panel-corrected standard errors.

Note that the results of the models are qualitatively similar. Our model suggested that electorally induced monetary expansions should occur if and only if the central bank is not independent and the exchange rate is allowed to fluctuate. Because the electoral coefficient in Table 3 describes the relationship between elections and the money supply when both modifying variables equal zero, it describes the situation when central banks are dependent and exchange rates are flexible. The statistically significant positive coefficient for Election in all four specifications in Table 3 is evidence that monetary expansions occur, as expected, when these conditions prevail.

Furthermore, specifications of equation 9 that use qualitative modifying variables make it easy to test all four implications of hypotheses 3 and 4 related to monetary policy by calculating the conditional coefficients for each relevant institutional combination (refer to Table 2). [22] The conditional coefficient for the effect of elections on the money supply when exchange rates are fixed and the central bank is independent can be determined by substituting the appropriate values of the institutional variables into equation 9.

[m.sub.i] = [[beta].sub.0] + [[beta].sub.1][E.sub.t] + [[beta].sub.2](1) + [[beta].sub.3](1) + [[beta].sub.4](E [cdotp] (1)) + [[beta].sub.5](E [cdotp] (1)) + [[beta].sub.6](1) + [[beta].sub.7](E [cdotp] (1) [cdotp] (1)) + [Sigma]([[beta].sub.j][m.sub.t-j]) + [e.sub.t], (10)

which simplifies to

[m.sub.t] = [[beta].sub.0] + [[beta].sub.1][E.sub.t] + [[beta].sub.2] + [[beta].sub.3] + [[beta].sub.4]E + [[beta].sub.5]E + [[beta].sub.6] + [[beta].sub.7]E + [Sigma]([[beta].sub.j][m.sub.t-j]) + [e.sub.t], (11)

and then comparing the electoral (E = 1) and non-electoral (E = 0) periods in the presence of both constraints. Were a monetary cycle to occur under such conditions, the change in the money supply should be greater during electoral (left side of equation 12) than nonelectoral (right side of equation 12) periods:

[[beta].sub.1](1) + [[beta].sub.2] + [[beta].sub.3] + [[beta].sub.4](1) + [[beta].sub.5](1) + [[beta].sub.6] + [[beta].sub.7](1) [greater than] [[beta].sub.1](0) + [[beta].sub.2] + [[beta].sub.3] + [[beta].sub.4](0) + [[beta].sub.5](0) + [[beta].sub.6] + [[beta].sub.7](0), (12)

which simplifies to [[beta].sub.1] + [[beta].sub.4] + [[beta].sub.5] + [[beta].sub.7] [greater than] 0. By analogous reasoning, a finding that [[beta].sub.1] + [[beta].sub.4] [greater than] 0 or [[beta].sub.1] + [[beta].sub.5] [greater than] 0 would be evidence of electorally induced monetary expansion under dependent central banks with fixed exchange rates and independent central banks with floating exchange rates, respectively. Hypotheses 3 and 4, however, predict that monetary cycles will not occur under such circumstances, so we expect that [[beta].sub.1] + [[beta].sub.4] + [[beta].sub.5] + [[beta].sub.7], [[beta].sub.1] + [[beta].sub.4], and [[beta].sub.1] + [[beta].sub.5] will each be indistinguishable from zero. In contrast, hypothesis 4 predicts that electorally induced monetary expansions will occur when the exchange rate is flexible and the central bank is dependent. As noted above, the test for this last proposition is simply [[beta].sub.1] [greater than] 0.

The conditional coefficients, and their associated standard errors, based on the results in column C of Table 3, are reported in the upper portion of Table 4. Note that the coefficients in Table 4 are estimates of the relationship between elections and monetary policy under the various open economy conditions presented in Table 1. As hypothesis 4 predicts, when capital is mobile, there is evidence of electorally induced monetary expansions if the exchange rate is allowed to fluctuate, but only when central bank independence is limited. As hypothesis 3 predicts, there is no evidence of such expansions when the exchange rate is fixed, regardless of central bank independence.

The lower portion of Table 4, which reports the conditional coefficients and standard errors calculated from column A, suggests that the relationship between elections and monetary policy may be more complicated than our model suggests. As in the upper portion of the table, the results are consistent with hypothesis 4: There is evidence of electorally induced monetary cycles when the exchange rate is flexible, but only when the central bank is dependent. Yet, because they suggest that electorally induced monetary cycles occur when central bank independence is limited and the exchange rate is fixed (lower right cell), the results in the lower part of Table 4 are not consistent with the monetary implications of hypothesis 3.

One explanation for this anomalous finding is that the Mundell-Fleming model rules out monetary policy as an independent instrument when the exchange rate is fixed, but it also explains why the money supply may nonetheless be tied to the electoral calendar. [23] If, as our model predicts, preelectoral fiscal expansions occur under a fixed exchange rate regime, then the resulting increase in interest rates will put upward pressure on the domestic currency. The central bank is obligated by the exchange rate commitment to counteract this pressure by selling domestic currency, thereby expanding the money supply. [24] If this link between monetary and fiscal policy is sufficiently strong, the electorally induced fiscal expansions that, according to hypothesis 3, are expected to occur under fixed exchange may also be reflected in monetary aggregates. The central bank could sterilize the effects of the foreign exchange intervention by engaging in offsetting sales of government securities. Since this would reduce th e expansionary consequences of the government's electorally induced fiscal policy, we would expect dependent central banks to be less apt to sterilize than independent ones. The insignificant conditional coefficient for the case of fixed exchange rates and an independent central bank is consistent with this conjecture.

In sum, the evidence supports the main implications of our model for monetary policy. The existence of opportunistic monetary cycles is conditioned by central bank independence and the exchange rate regime. Furthermore, there is evidence of monetary expansion when the government retains national monetary policy and influence over the central bank. There is no evidence of such cycles when the central bank is independent or the exchange rate is allowed to float. [25] Contrary to our expectations, there is some evidence of electorally induced cycles in monetary policy when the exchange rate is fixed. But, in contrast with our other findings, this evidence is not robust with respect to the choice of the aggregate used to measure the money supply, and it could be interpreted as indirect support for the model's predictions about fiscal policy. [26] We now turn to a direct examination of the fiscal policy results.

Fiscal Cycles

In contrast to monetary policy, hypothesis 3 indicates that electorally induced fiscal cycles are likely when exchange rates are fixed and that the effectiveness of fiscal policy on output increases under fixed exchange rates when capital is mobile. Hypothesis 4 predicts that fiscal policy is ineffective when exchange rates are flexible and capital is mobile, and under these conditions there should not be fiscal cycles.

Previous research generally asks whether deficits are likely to increase in electoral years. Unlike the empirical work on monetary cycles, the literature codes data on a yearly rather than quarterly basis. [27] The evidence is decidedly mixed. Some authors indicate support for opportunistic fiscal cycles in some subset of industrialized countries (Alesina and Roubini 1997; Hallerberg and von Hagen 1999; Franzese 1996 finds such cycles when the replacement risk to a sitting government is high), and others find no support for such cycles (De Haan and Sturm 1994b). One reason may be that the studies do not differentiate between the effects of fixed and flexible exchange rates or mobile and immobile capital.

We rely on data sets from two recent articles, De Haan and Sturm (1997) and Hallerberg and von Hagen (1999). Our rationale is as follows. Each covers a somewhat different group--De Haan and Sturm (1997) examine 19 OECD countries, and Hallerberg and von Hagen (1998, 1999) use the 15 current members of the European Union. [28] Much of the economic data also come from different international organizations. [29] If we find evidence of politically induced fiscal cycles in both data sets, then we will have more confidence in the robustness of the results. We also restrict the period to 1981-92 for the Hallerberg and von Hagen data set and to 1982-92 for the De Haan and Sturm data set. We did this for two reasons, both of which concern policy in Europe: In fall 1992 the Exchange Rate Mechanism suffered a severe crisis in the markets, which put in doubt the credibility of fixed exchange rates; and after 1992 states that wished to join the Economic and Monetary Union faced restrictions on their debt levels, which pre sumably had a direct effect on fiscal policy. The beginning date of the respective data sets is the same as in the published articles, and we want to emphasize that this period is theoretically interesting. Several European Community countries reestablished credible fixed exchange rates with the Exchange Rate Mechanism in 1979, and both data sets allow strong tests of whether a country's exchange rate regime affects the likelihood of fiscal expansion before an election.

The model upon which these articles build, and which has become a standard in the field, is that of Roubini and Sachs (1989). Their regression equation is

d[b.sub.it] = [[beta].sub.0] + [[beta].sub.1]d[b.sub.it-1] + [[beta].sub.2]d[U.sub.it] + [[beta].sub.3]d[y.sub.it] + [[beta].sub.4][b.sub.it-1]d([r.sub.it] - [y.sub.it]) + [[beta].sub.5][pv.sub.it], (13)

where the dependent variable, d[b.sub.it], is the change in the ratio of gross debt to GDP. The equation contains a set of economic variables and a set of political variables (represented as [pv.sub.it]). The economic variables are: d[b.sub.it-1], the lagged debt ratio; d[U.sub.i], the change in the unemployment rate; and [b.sub.it-1]d(r - y), the change in debt servicing costs, which is computed as the change in the real interest rate minus the change in the growth rate times the gross deficit in the previous year. In addition, some authors include the change in real GDP, d[y.sub.i]. To be consistent with previous work, we include that change in our regression using the data of Hallerberg and von Hagen, as they do, and we follow De Haan and Sturm by not including it in our regression with their data. [30] These economic variables are expected to influence the budget in a given year; higher levels of unemployment and debt servicing costs should increase government debt levels, and higher levels of economic growth should decrease debt levels. [31]

The standard political variables generally include codings for institutional differences or the partisanship orientation of the government. Roubini and Sachs (1989) received much attention for their finding that the type of government affects the size of budget deficits. One-party majority governments maintain the tightest fiscal discipline, two- or three-party majority governments less so, and four- or five-party governments even less; minority governments, regardless of the number of parties in the coalition, are the most undisciplined. Many other researchers have examined government type and other political variables. As control variables, we use the political variables in De Haan and Sturm (1997) as well as Hallerberg and von Hagen (1999). De Haan and Sturm found that Roubini and Sachs made several coding errors and, based on the original coding procedures, argue that the type of government does not affect deficit levels.

Hallerberg and von Hagen, following Edin and Ohlsson (1991), break the government type variable into three separate dummies and add a variable--the percentage of cabinet portfolios held by leftist parties--to control for partisan effects. They also add variables for election years and for two fiscal institutions, a strong finance minister and negotiated targets. Among their findings were a connection between government type and a fiscal institution that is meant to reduce the size of deficits, but no effect of government type per se on the size of deficits. In particular, delegation to a strong finance minister who can monitor spending ministers and punish those who "defect" is feasible in states with one-party majority governments. In multiparty and minority governments, the coalition members are not willing to delegate such power to one actor. Commitment to numerical targets negotiated among the coalition partners for each ministry provides an alternative in multiparty governments. [32]

In order to test our contention that political business cycles are most likely to occur when exchange rates are fixed, we structure the regression equations to include the economic and political factors noted above as well as variables to consider the effects of exchange rate regime and of elections. Our equations are:

d[b.sub.it] = [alpha] + [[beta].sub.1]Election + [[beta].sub.2]Flexible + [[beta].sub.3]Election X Flexible + [[beta].sub.4]d[b.sub.it-1] + [[beta].sub.5]d[U.sub.it] + [[beta].sub.6][b.sub.it-1]d([r.sub.it] - [y.sub.it]) + [[beta].sub.7]Government Type, (14)

When employing the De Haan and Strum data, and

d[b.sub.it] = [alpha] + [[beta].sub.1]Election + [[beta].sub.2]Flexible + [[beta].sub.3]Election X Flexible + [[beta].sub.4]d[b.sub.it-1] + [[beta].sub.5]d[U.sub.it] + [[beta].sub.6][b.sub.it-1]d([r.sub.it] - [y.sub.it]) + [[beta].sub.7]d[y.sub.it] + [[beta].sub.8] Two-Three Party Govt. + [[beta].sub.9]Four-Five Party Govt. + [[beta].sub.10]Minority Govt. + [[beta].sub.11]Strong Finance Minister + [[beta].sub.12]Negotiated Targets + [[beta].sub.13]Strong Finance Minister X Election + [[beta].sub.14]Negotiated Targets X Election + [[beta].sub.15]Left, (15)

when the Hallerberg and von Hagen data are used, where [[beta].sub.1] is the coefficient when there are elections under fixed exchange rates, [[beta].sub.2] is the coefficient for flexible exchange rates and no elections, and [[beta].sub.3] captures the effect of a change to a flexible exchange rate on the relationship between the debt and elections. This specification of the variables follows closely the model provided in the section on monetary policy. Our expectations about the effects of these variables differ, however. Based on hypotheses 3 and 4 presented in Table 1, we expect fiscal expansions to occur only when exchange rates are fixed (i.e., when Flexible is equal to zero). When exchange rates are flexible, we expect that fiscal policy is ineffective and a government will not initiate a preelectoral fiscal expansion.

Because of our interest in the effect of elections under different exchange rate regimes, we consider two alternative methods for coding elections. Hallerberg and von Hagen (1999) calculate their election variable as a dummy for years in which an election is held. This follows standard practice in the literature but is at best inexact and at worst inaccurate. Some countries hold elections in early spring, others in late autumn. The same country may even hold elections in different times of the year over different electoral cycles; for example, in the United Kingdom, there were two elections in 1974, in February and October, and just five years later elections were in May. A variable that makes no distinction among these presumably will understate the effect of elections as well as increase the standard error.

We supplement the standard measurement of elections with Franzese's (1996) more exact definition. He calculates Election as the proportion of preelectoral year; for example, a February 1 election would be coded as 1/12 in that year and 11/12 in the previous year. [33] We applied this coding to the two data sets to see whether the coding rules affected the results. Our model predicts that fiscal cycles should occur when exchange rates are fixed, which in our regressions is the case when election equals 1 and flexible equals 0. The sign of election should therefore be positive. The model also predicts that the coefficient for the interaction term election X flexible should be negative; indeed, since governments have no incentive to use fiscal cycles when exchange rates are flexible, one expects that the coefficients for election and election X flexible will sum to zero. To test directly for the effect of flexible exchange rates on fiscal cycles, we also calculated the conditional coefficient for elections when exchange rates are flexible and when they are fixed.

Table 5 reveals strong evidence that countries with fixed exchange rates experience fiscal cycles and that flexible exchange rates eliminate these cycles. In all the equations with the more precise measure for elections, the positive and significant conditional coefficient means that preelectoral fiscal expansions occur when the exchange rate is fixed but not when it is flexible. For example, in column B the coefficient for election indicates an increase in the gross debt level of roughly 1.5 points during electoral periods when the exchange rate is fixed, compared to no increase during electoral periods when the rate is flexible, that is, when the sum of the coefficients for election and election X flexible is close to zero (for convenience, we present the conditional coefficients for each exchange rate regime and their associated standard errors in Table 5). The effect of exchange rate regime on the relationship between budgets and elections is most dramatic with the second equation using the Hallerberg an d von Hagen data set (column E): States with fixed exchange rates will increase their ratio of gross debt level to GDP almost three percentage points in the year before an election.

This difference in the effect of the electoral period in the two data sets may be the result of the institutional variables added by Hallerberg and von Hagen. Most important, because models D, E, and F employ interactions between election and Strong Finance Minister and election and Negotiated Targets, the interpretation of the conditional coefficient changes. The coefficient for election now summarizes the relationship between elections and gross debt when the exchange rate is flexible in the absence of a strong finance minister or the use of negotiated targets.

Table 6 computes conditional coefficients for the effect of fiscal institutions. In both cases the fiscal institution eliminates the increase in debt before elections. It appears, therefore, that these domestic institutions constrain fiscal policy in much the same way that central bank independence constrains monetary policy.

There are additional nuances in the results given in Table 5. As columns A and D indicate, the standard election variable has the correct sign but is not significant when we do not use the Franzese coding for elections. In both data sets, however, the effect almost doubles, and the standard deviation decreases, when Franzese's more exact measure is used (columns B and E). Once again, fiscal electoral cycles are completely absent when exchange rates are flexible. [34]

Models C and F in Table 5 allow us to test the robustness of our results by relaxing the assumption that capital was fully mobile in every country during the entire observation period. Similar to the coding of "no monetary policy autonomy" in the previous section, we created a variable called "no fiscal policy autonomy (nofpa)" and substituted it into the regressions for flexible. This continuous variable is the product of the Simmons (1996a) coding of capital controls and the exchange rate regime dummy variable. It ranges from zero (when flexible = 0 and/or there are no capital controls) to 1 (when flexible = 1 and all capital controls are present). The findings are qualitatively the same as for the regressions that use a rougher measure of capital mobility: Elections are associated with an increase in government debt when fiscal policy autonomy is retained (compare the coefficient on election in columns C and F), but this relationship is diminished when fiscal policy autonomy erodes (compare the coefficien t on the interaction term). This provides indirect support for our earlier contention that capital was more or less mobile during the period examined.

The results fit our expectations. There is evidence of preelectoral fiscal expansion when the exchange rate is fixed, which confirms hypothesis 3. The conditional coefficient for elections is positive and highly significant. In comparison, when we hold relevant domestic institutions constant and move to a flexible exchange rate, in all cases the size of the conditional electoral coefficient is reduced. This result holds for all combinations of domestic fiscal institutions, and it confirms hypothesis 4: All else equal, the conditional coefficient for elections is always greater under a fixed exchange rate than a flexible regime.

EXTENSIONS OF MODEL 1: PARTISAN POLITICAL BUSINESS CYCLES

We have concentrated on the effects of international conditions as well as domestic political institutions on the propensity of governments to engineer preelectoral expansions of either the money supply or the budget deficit. Our model also allows us to consider how these international and domestic factors can be integrated into a second tradition in political economy, namely, whether partisan differences systematically translate into differences in macroeconomic policy. One school contends that there is a clear divide between the policies that left and right governments pursue. Compared to rightists, leftists are expected to prefer higher inflation and the higher employment that accompanies it and to support a consistently higher level of the money supply. This view is largely based upon the assumption of a stable Phillips curve tradeoff between inflation and unemployment. In terms of fiscal policy, leftist governments run larger budget deficits than rightist governments, again to boost employment. The reaso ning behind these arguments is straightforward--leftist governments appeal to their supporters, who are usually laborers, whereas rightist governments appeal to their backers, who are usually capital owners (the classic article is Hibbs 1977).

There is some debate about how internationalization affects the likelihood of partisan cycles. As we stated earlier, several scholars insist that global capital mobility will end the ability of leftist governments to act differently from rightist governments (in addition to earlier citations, for literature surveys see Garrett 1998, chap. 1, and Clark n.d., chaps. 4 and 6). Capital will simply move to the country where it is best treated, and there will be an inevitable harmonization of policy as capital becomes more mobile. Others, such as Garrett (1995, 1998) and Rodrik (1997), insist that partisan differences will remain or even increase in the face of more internationalization. Populations become more uncertain in a global world and demand an expansion of the welfare state rather than a contraction. Clark (n.d.) finds no evidence of partisan cycles before or after an increase in internationalization. In a study with some similarity to ours, Oatley (1999) examines how fixed and flexible exchange rates tog ether with increasing capital mobility affect partisan cycles. Based on a Mundell-Fleming framework, he argues that there are partisan monetary cycles in the form of real money market interest rates that go away when there is open capital and a fixed exchange rate regime, and there are partisan fiscal cycles in the form of larger budget deficits that disappear when there is open capital and a flexible exchange rate regime. [35]

A useful feature of the formal model we developed earlier is the ease with which it can be applied to the partisan debate. Recall that the variable which differentiated between periods before and not before elections was k, which we assumed was greater than 1 in election periods and equal to 1 in nonelection periods. Since k is the parameter that captures the propensity of policymakers to push output above the natural rate, it can be interpreted as a function of the expansionary tendencies of the government. Consequently, an extension of the model is to replace "election" with "leftist government" in the model and to test the resulting hypotheses with data that interact partisanship with the relevant structural conditions.

HYPOTHESIS 5. Our expectations about the effect of partisanship are as follows.

A. When capital is immobile and the central bank is dependent, leftist governments will create fiscal but not monetary expansions.

B. When capital is immobile and the central bank is independent, leftist governments will create fiscal expansions and monetary contractions.

C. When capital is mobile and the exchange rate is fixed, leftist governments will create fiscal but not monetary expansions, regardless of central bank independence.

D. When capital is mobile and the exchange rate is flexible, leftist governments will create monetary but not fiscal expansions, unless the central bank is independent.

If the partisan literature is correct and there is a fundamental difference between the preferences of leftist and rightist governments, then we expect, the following for monetary policy under leftists: monetary contractions when capital is immobile and the central bank is independent; monetary expansions when capital is mobile, the exchange rate is flexible, and the central bank is dependent; and no change in policy under the remaining conditions. For fiscal policy, under leftists there will be fiscal expansions when capital is immobile and central banks are dependent as well as when capital is mobile and exchange rates are fixed, but otherwise no partisan differences should be evident. [36]

To test these hypotheses, we begin with the same empirical models used earlier. This facilitates a ready comparison with the opportunistic political business cycle results as well as with the literature that established these models. In place of the electoral variables we use a variable for partisanship based on Woldendorp, Keman, and Budge (1998). The variable takes a value of 1 for right-wing dominance and 5 for left-wing dominance, and it follows Oatley's (1999) coding.

The results for the test of partisan differences in monetary policy--conditioned on both the exchange rate regime and the degree of central bank independence--are presented in Table 7. Since the coefficient for Partisanship describes the expected causal effect of a one-unit increase in leftist governance when central bank independence is limited and the exchange rate is flexible, hypothesis 5D predicts that this coefficient should be positive and statistically significant. The fact that it is positive but not significant in columns A and B and significant but negative in columns C and D means that, under the conditions in which partisanship is likely to matter the most, there is no evidence for the predicted effect of the ideological orientation of government.

As was the case with our examination of the consequences of elections, it is possible to test the partisan hypothesis under each of the logically possible structural conditions by calculating conditional coefficients and standard errors. These are reported in Table 8. The results in the upper portion of the table, which are calculated from column C in Table 7, are particularly surprising. As noted above, when the government is expected to have the greatest control over monetary policy (i.e., when the exchange rate is flexible and the central bank is dependent), leftist government leads to a monetary contraction rather than a monetary expansion. The upper left-hand cell, however, shows that when monetary policy is controlled by a highly independent central bank, leftist government is associated with monetary expansion. Thus, the results for the case of flexible exchange rates are exactly the opposite of what we might expect; the sort of differences in monetary policy predicted by the partisan model are found only when the central bank is highly independent. Consistent with our argument about the constraining effects of fixed exchange rates on monetary policy, there is no evidence of partisan differences in monetary policy when the exchange rate is fixed--regardless of central bank independence.

The results calculated from column A of Table 7 also are not very supportive of the partisan hypothesis. As predicted, there is no evidence of partisan differences in monetary policy when the central bank is independent, the exchange rate is fixed, or both. Yet, there is no evidence of partisan differences when they should be most readily observed--when the exchange rate is flexible and the central bank is dependent. To summarize, the only indication that leftist governance results in an expansion of the money supply occurs when the central bank is highly independent, and this result is not robust with respect to the choice of indicator for the dependent variable.

The fiscal policy results also have idiosyncrasies that are not consistent with the standard partisan argument. Table 9 presents regressions that parallel those we reported in the opportunistic business cycle case. Note that columns A and C assume full capital mobility during the period, but columns B and D relax this assumption. Because, in the presence of the interactive term, the coefficient on the partisanship variable indicates the effects of partisanship when the exchange rate is fixed, our expectation from hypothesis 5C is that this coefficient should be positive. We also anticipate that this effect should disappear when the exchange rate is flexible (columns A and C) or when there is an erosion of fiscal policy autonomy (columns B and D). In the first two columns the variable is not significant. There appears to be no evidence of partisan cycles in the OECD data set. In the last two columns, which add domestic fiscal institutions and are restricted to current EU members, partisanship does seem to hav e an effect, but the sign is the opposite of what the model predicts. Based on hypothesis 5D, the results indicate that a move to the left reduces the size of growth in the debt burden. Moreover, the conditional coefficients reported in column C indicate that this effect goes away when flexible exchange rates are in place.

To be able to interpret these coefficients, we must consider the relevance of the domestic institutions in the regressions. As in the opportunistic cycle results, the conditional coefficients for partisanship are for the absence of both a strong finance minister and budgetary targets. Table 10 calculates the conditional coefficients for partisanship under different domestic institutional configurations when exchange rates are fixed, which is when we expect partisanship to have an effect. When both fiscal institutions are absent, a move to the left decreases the growth of the debt. When targets are in place, however, a move to the left increases the debt, and there is no partisan effect at all under a strong finance minister.

In general, for both monetary and fiscal policy the predictions of partisan cycles are not borne out. In no case is there evidence of partisan cycles consistent with the predictions about the effectiveness of either monetary or fiscal cycles under given structural conditions. A move to the left leads to an expansion in the money supply only in the most unlikely cases, when the central bank is independent or when spending is constrained by negotiated targets. In the fiscal regressions, there are no partisan effects at all with the OECD data set, and the results are the opposite of those predicted with the EU data set.

EXTENSIONS OF MODEL 2: IMPLICATIONS FOR EUROPEAN ECONOMIC AND MONETARY UNION

Our evidence from Europe in the 1980s and early 1990s indicates that opportunistic fiscal cycles are especially prevalent when capital is mobile and exchange rates are fixed. This result has an important implication for the likely conduct of fiscal policy under the Economic and Monetary Union in Europe. Fixed exchange rates and mobile capital will be the norm in all EU countries that adopt the common currency, either in the first wave on January 1, 1999, or thereafter. [37]

Although preparations for EMU may have reduced fiscal electoral cycles in the 1990s, it is by no means clear that EMU will restrict fiscal cycles. More generally, there is some concern that governments, once accepted into EMU, will have an incentive to renege on their commitment to lower deficits. Excessively high deficits can both increase inflation and depreciate the euro, and a state can hope to gain the political benefits from high deficits under a common currency without suffering the costs as long as the other states maintain their budget discipline. Another concern is that the deficits can become so large that a state cannot hope to pay back its debt without a bailout from the EU. Because the bailout would be paid for by all EU citizens but would reduce directly just the deficit of the country threatening default, states may have an incentive to run larger deficits than they would if the EU did not exist. [38]

The member governments are not ignorant of this situation. The Treaty of Maastricht explicitly bans any EU bailout of members, and the states have established a procedure both to monitor the participants and punish any defectors under the Stability and Growth Pact to which the European Council agreed in June 1997. The European Commission will monitor the fiscal health of member states and will report to the Council of Ministers on whether, in its opinion, a deficit greater than 3% is "excessive." The Council of Ministers will then decide whether to accept the commission's assessment. If a country's deficits are deemed excessive, the transgressor will be sanctioned. The offender must make a noninterest bearing deposit with the European Commission equal to .2% of GDP plus an additional .1% per 1% over the 3% limit. The entire process should take ten months; if the country has not reformed its budget to the European Council's satisfaction within that time, the deposit becomes a fine. [39]

Whether the pact will deter states from running chronic deficits is debatable, but clearly it will not discourage opportunistic political business cycles because of the timing of its different procedures. Consider Germany, which traditionally has elections in the fall. A government could promote a large fiscal expansion in the spring and summer that leads to large deficits. In principle, the European Commission could immediately inform the European Council and begin proceedings, but in practice the commission is likely to wait until the final annual budget figures are published in March of the following year; it would fear losing credibility if the preliminary figures prove incorrect. [40] An additional ten months would pass before fines can be imposed, and only if the government is not taking corrective measures. In other words, any government can push up the deficit shortly before elections so long as it is cut back again after the elections. [41]

An alternative is for states themselves to place domestic institutional constraints on their spending behavior. Consistent with Hallerberg and von Hagen (1999), our empirical results indicate that either commitment to negotiated budget targets or delegation to a strong finance minister reduces preelectoral fiscal cycles. A move to one of these institutions may be in order. More generally, because the temptation to use fiscal policy for short-term electoral gain will be greater under EMU, states may have the incentive to develop such institutions only if they are sure that the EU will not bail them out. Indeed, based on an analysis of federal systems, Eichengreen and Von Hagen (1996a) argue that a formal bailout rule is not even needed as long as states maintain their ability to raise revenues. The EU can simply say much as California did in the mid-1990s when Orange County's creditworthiness plummeted. States may then learn through trial and error to introduce more prudent fiscal policy management. A possibl e precedent again comes from the United States, where fiscal mismanagement eventually led 47 of 50 American states to pass some form of fiscal policy restriction (Eichengreen and von Hagen 1996b). These changes obviously were not driven by concerns about what state budget deficits could do to the value of the dollar, but by "domestic" concerns about the effects on home populations. These institutional rules also seem to have benefits: States with tighter antideficit rules as well as restrictions on the authority of the legislature pay lower interest rates on their bonds (Poterba and Rueben 1998). [42]

CONCLUSION

Our model makes fairly weak assumptions about the preferences of monetary and fiscal agents but very strong assumptions about the institutional constraints under which they operate. The strong assumptions about the Mundell-Fleming conditions are intended in part to make the analysis tractable, but they have the added benefit of stacking the cards in favor of falsification: If the predictions based on such a stark model withstand provisional tests, then there are good reasons to believe further refinement will be fruitful. Indeed, the empirical results support the insights of the model. Preelectoral monetary expansions are likely in states with mobile capital, a flexible exchange rate, and a dependent bank and are absent otherwise; preelectoral fiscal expansions are likely under mobile capital and fixed exchange rates.

These results have several interesting implications for the relationship between institutions and the international environment. In the presence of mobile capital, the exchange rate regime constrains the ability of policymakers to use one of the two macroeconomic instruments before elections. When governments change their exchange rate regime, one should expect a change in the incumbents' behavior before an election. If a government fixes its currency when capital is mobile, for example, one would expect a shift from monetary to fiscal policy instruments. It can be argued that such a shift is occurring in Britain. Under Stage III of EMU, EU members anticipate the eventual creation of Exchange Rate Mechanism II for countries that do not join the common currency. The expectation is that the band around the euro for these currencies will progressively narrow, so that these countries will also experience essentially fixed exchange rates. Preelection monetary expansions will become less effective than fiscal expa nsions.

There also are implications for the selection of institutions by governments. When considering the tradeoffs involved in enhancing central bank independence, a government may consider what it gives up in terms of survival in the next election if it relinquishes monetary policy autonomy in exchange for greater general price stability with a more independent central bank. A government in a flexible exchange rate economy will lose its ability to influence outcomes with macroeconomic policy before an election if it makes the bank more independent. If the government moves the country to a fixed exchange rate regime, however, it can gain some price stability and still maintain the ability to manipulate the economy before an election. Our work suggests, for example, that Chancellor of the Exchequer Brown would have lost little maneuverability for his government before the next election by granting the Bank of England greater independence in May 1997. The government presumably can initiate a fiscal expansion before the next election so long as Britain has joined Exchange Rate Mechanism II by then. Counter to Bernhard and Leblang (1999), in the absence of central bank independence, governments that peg their exchange rates when capital is mobile do not sacrifice their ability to pursue electoral goals. Instead, they change the instrument with which they can do so.

Our study suggests why previous empirical research on fiscal cycles has yielded ambiguous results. Few studies consider the effects of different exchange rate regimes on the likelihood of fiscal cycles. States engineer fiscal expansions when exchange rates are fixed but not when they are flexible, and works that do not include the exchange rate regime are missing this critical variable. One article that does consider the exchange rate factor, by Clark and Reichert (1998), appears to contradict the results reported here. The authors find that states with either fixed exchange rates or independent central banks are less likely to experience opportunistic political business cycles. Yet, they concentrate on outcomes, such as the unemployment rate and economic output, rather than on policy instruments, such as the money supply or budget balances. Our research produces an interesting puzzle: If fiscal cycles exist when monetary policy is constrained by fixed exchange rates or a highly independent central bank, why do Clark and Reichert find no evidence of cycles in unemployment or growth under precisely these conditions? A possible explanation is that politicians use monetary policy before elections to affect the general economy but manipulate fiscal policy to sway specific constituencies. Indeed, although an increase in the money supply may help certain groups--such as home buyers--more than others, it is a blunt instrument for cultivating specific clienteles. Fiscal policy, in contrast, is more suited to targeted use, whether through greater spending, tax cuts, or both. The implication is that these different strategies have markedly different macroeconomic consequences.

Finally, it is clear that the choice of exchange rate regime and the degree of capital mobility have complex effects on the ability of politicians to influence the economy for electoral purposes. Our study does not necessarily contradict the growing literature on the effects of increased capital mobility on national policy autonomy, but it shows that the effects of capital mobility vary a great deal across cases, both because they are refracted by the institutions through which they are transmitted, such as central banks, and because exchange rate regimes interact with capital mobility in important ways.

William Roberts Clark is Assistant Professor of Politics, New York University, 715 Broadway, New York, NY 10003-6806 (william.clark@nyu.edu). Mark Hallerberg, is Assistant Professor of Political Science, University of Pittsburgh, 4t23 Forbes Quadrangle, Pittsburgh, PA 15260 (hallerb+@pitt.edu).

The paper was first drafted while the authors were members of the Sam Nunn School of International Affairs at the Georgia Institute of Technology, and the authors thank colleagues and students there, as well as at New York University and the University of Pittsburgh, for helpful advice and comments. Revisions were supported by the MacArthur Foundation through the Center of International Studies at Princeton University. We also thank David Andrews, Peter Brecke, Kirk Bowman, Benjamin Cohen, David Denoon, Herbert Doering, Martin Edwards, Jesus Felipe, Jeifry Frieden, Fablo Ghironi, Joanne Gowa, Peter Kenen, Jeffry Lewis, Bill Mabe, Jonathon W. Moses, Shanker Satyanath, Beth Simmons, Alberta Sbragia, Rolf Strauch, Michael Wallerstein, Katja Weber, Tom Willett, anonymous reviewers, and the editorial staff at APSR for helpful and insightful comments. The paper also benefited from the comments of seminar participants at the University of California, Berkeley; the Mannheim Centre for European Social Research; the M ax Planck Institute in Cologne, Germany; McGill University; Princeton University; Rutgers University; the University of Potsdam; and Washington University, St. Louis. Stephen Flanders provided research assistance. We thank Jakob De Haan for sharing his data.

(1.) The authors thank Lawrence Broz, Benjamin Cohen, Peter Hall, and Louis Pauly for their thoughts on this point.

(2.) Discussion of the Mundell-Fleming framework can be found in any open economy macroeconomics text. The foundational works are Fleming 1962 and Mundell 1963. Frieden 1991 and Cohen 1993 are two important extensions of the framework into political economy. The standard Mundell-Fleming model has been criticized for lacking adequate microfoundations (Dornbusch 1976). Attempts to correct that problem (Corsetti and Pesenti 1997; Dornbusch 1976; Obstfeld and Rogoff 1996) have produced positive conclusions broadly consistent with the standard model. At any rate, it is clear that the standard model has guided policy for the last thirty years. See Ghironi and Giavazzi 1998 for a model of interdependence that emphasizes the size of the economy.

(3.) This creates the opportunity for conflict between government and even the most independent central bank. Berger and Thum (n.d.) analyze this friction, and Berger and Schneider (n.d.) provide evidence that such conflict occurs in Germany.

(4.) Fiscal policy has direct distributional consequences, but we will focus on macroeconomic effects in the main body of the article, then return to distributional consequences in the conclusion.

(5.) The same may be true for ministers of finance, who tend to be more concerned about the general health of the economy than are other ministers (Hallerberg and von Hagen 1998).

(6.) Two ways to capture cross-actor differences are found in the literature. Actors may differ in ideal points for inflation and/or output ([[phi].sup.*i] [neq] [[phi].sup.*-i] and [y.sup.*i] [neq] [y.sup.*-i]) (Svensson 1995), or they may differ in the importance they place on inflation stabilization relative to achieving output targets ([[alpha].sup.t] [neq] [[alpha].sup.-1]) (Lohmann 1992; Rogoff 1985). We adopt the former approach.

(7.) If one is uncomfortable with deriving explanatory power through arbitrary assumptions about changing preferences (and good reasons are provided in Stigler and Becker 1977), then assume that the government is engaged in strategic behavior--it is acting "as if" it were minimizing a loss function, where k [greater than] 1 in preelectoral periods in order to maximize its utility in the game it is playing with voters.

(8.) This is the flip side of the logic employed by Blinder (1998), who recommends that politicians instruct the central bank to act as if their ideal point were the natural rate of growth.

(9.) For simplicity, we assume that central banks are either dependent or not.

(10.) This is merely a cursory way of saying they manipulate the money supply and/or short-term interest rates in an attempt to effect changes in inflation.

(11.) "See Branson and Buiter 1983, Dornbusch 1976, and Mundell 1963. The reasoning is that fiscal expansion (or any other increase in autonomous expenditure) leads to an increase in both income and interest rates. When capital is mobile, the rise in interest rates attracts capital, which leads to a currency appreciation. When the exchange rate is fixed, the central bank has to expand the money supply to offset the effects of the capital inflow on the exchange rate. Thus, under fixed exchange rates and mobile capital, a fiscal expansion induces a reinforcing monetary expansion. Dooley (1996) suggests in his literature review of capital controls that they be removed in some cases to make fiscal policy more effective in stabilization programs.

(12.) These stark conclusions are drawn from a version of the Mundell-Fleming model in which prices are held constant, capital is fully mobile, and the short-term effects of policy are ignored. More measured and realistic conclusions are produced by relaxing these strong assumptions. Nevertheless, the strong conclusions are adopted as a starting point.

(13.) Using a similar model, Berger (1997b) argues that the West German government exploited just such a position as a Stackelberg leader during the 1950s. The combination of a closed economy and a conservative central bank would give the government the large budgets that it wanted but would force the Bundesbank to fight inflation through monetary policy. The only way the bank could escape this trap was to find a credible external commitment device in the return to convertibility in the context of the Bretton Woods pegged exchange rate system (Berger 1997a).

(14.) Chronic deficits may induce capital flight and undermine the fiscal expansion. This flight is expected only when investors lose confidence in the government's ability to repay. Short-term increases in the debt due to electoral cycles are unlikely to cause such flight as long as the government is expected to bring the deficit back to its level in the previous nonelectoral period. See Canzoneri and Diba 1999 for a similar argument applied to the Stability and Growth Pact within Europe.

(15.) A Stackelberg equilibrium is a set of strategies such that each player chooses its best response given that one player (in this case the government) chooses first.

(16.) We are assuming that these variables are exogenous. It is possible, of course, that capital mobility, the exchange rate, and even central bank independence are themselves choice variables and are endogenous to the model. If the exchange rate regime prevents a policymaker from using an instrument for electoral purposes, and if the exchange rate regime is also under the control of the politician, why would not an opportunistic politician just drop the commitment to the exchange rate regime when it becomes inconvenient? The answer provided by the Mundell-Fleming model is that doing so would not mean gaining an electorally useful instrument but a change in which instrument would be effective for electoral purposes. An election may make policymakers more eager to manipulate the economy, but that alone is not enough to create the kind of endogeneity problem that would bias our tests of the link between policy and elections. More generally, we simply do not observe instances in which governments systematically change the exchange rate regime, the level of capital mobility, or central bank independence before elections. We thank two anonymous reviewers for their comments on this issue.

(17.) Clark and Reichert based their codings of exchange rate restrictions on Coffey 1984; IMF, International Financial Statistics (various years b); and OECD 1985. Their dummy variable for central bank independence equals 1 when the score for legal independence is above the median, (see Cukierman, Webb, and Neyapti 1992 for an extensive discussion of this measure).

(18.) See Appendix B for data sources and descriptions of indicators.

(19.) Unlike M1, quasimoney includes such assets as savings accounts, small-denomination time deposits, money markets accounts, and money market mutual fund shares. When added to M1, their sum is a broader monetary aggregate, often referred to as M2. Given the tendency of asset holders to substitute among them, some scholars argue that the broader indicator is a better measure of the money supply. We use both indicators as a robustness test. Also, use of the broader aggregate allows us to include a wider set of countries in our analysis than would have been possible for M1 alone.

(20.) Unemployment is measured as the quarterly rate of change in the number of persons unemployed. Inflation is the year-to-year rate of change in the consumer price index. See Appendix B.

(21.) To create a continuous measure of eroding monetary policy autonomy as a result of the interaction between exchange rates and capital mobility, we averaged, inverted, and rescaled on the interval from 0 to 1 the eight capital control variables used in Simmons (1996a) and multiplied this figure by the dummy variable for fixed exchange rates. The latter equals one when the exchange rate is fixed, zero otherwise, so the resulting variable equals zero when the exchange rate is flexible and is an increasing function of the degree of capital mobility when the exchange rate is fixed. This variable and the Cukierman, Webb, and Neyapti legal measure of central bank independence appear as Fired and CBI in columns B and D of Table 3, respectively.

(22.) See Friedrich 1982 and Jaccard, Turrisi, and Wan 1990 for a useful introduction to the conditional interpretation of multiplicative interaction models.

(23.) We are indebted to Tom Willett on this point.

(24.) This automatic monetary expansion makes fiscal policy particularly effective when capital is mobile and the exchange rate is fixed. As Fabio Ghironi points out (personal communication), an examination of foreign exchange interventions may allow future researchers to distinguish between electorally induced monetary cycles (as understood in this article) and the monetary consequences of electorally induced fiscal cycles.

(25.) Alesina and Roubini (1997) find evidence of electorally induced monetary expansion in their sample of 18 OECD countries, but they assume that the occurrence of monetary cycles is uniform within their sample. Our evidence suggests that they inappropriately pooled cases where monetary cycles are expected to occur with those where they are not. Although we did not conduct country-specific tests, our finding is in tension with evidence of an electoral monetary cycle in the United States (Grier 1989) and consistent with studies that find a lack of evidence for such a cycle (Beck 1987). For more on the U.S. case, see contributions to Mayer 1990.

(26.) The somewhat different findings with the two monetary measures complicate the interpretation of the results, but the predictions of the model are broadly supported. First, our model predicts that central bank independence will influence electorally motivated monetary, but not fiscal, policy. The lower part of Table 4 supports the notion that central bank independence disturbs the link between elections and monetary policy when the exchange rate floats but not when the exchange rate is fixed, which is consistent with the contention that the observed link under fixed exchange rates is driven by fiscal policy. Second, the conditional electoral coefficients are smaller and their standard errors are larger when the exchange rate is fixed than when the exchange rate floats and the central bank is dependent. This is consistent with the idea that the results for the case of floating exchange rates and a dependent central bank reflect the deliberate manipulation of monetary instruments for electoral purposes, wh ereas the fixed exchange rate results reflect reverberations from fiscal policy. Finally, the coefficients calculated from tests based on a broader monetary aggregate, which is less likely to reflect exchange rate interventions driven by fiscal policy, show no evidence of electorally induced fiscal policy when the exchange rate is fixed.

(27.) Quarterly data on expenditures and tax collections can have a seasonal element both within and across countries that is difficult to control for, even with sophisticated econometric techniques.

(28.) The original Hallerberg and von Hagen data set covers 1981-94. We thank Rolf Strauch for providing updated data for the economic variables. See Appendix B for original sources.

(29.) De Haan and Sturm rely on OECD Economic Outlook for most of their data, whereas Hallerberg and von Hagen drew mostly from Statistical Annex of European Economy (see Appendix B). There is a difference in particular in the debt coding rules. The European Commission uses Maastricht definitions for debt and inflation, which have minor accounting differences with the OECD data.

(30.) There is some controversy about including growth in GDP as an independent variable because real GDP appears in the denominator on the left-hand side of the equation (see also Borelli and Royed 1995). We are not interested in this controversy per se, so we follow the practice of the respective authors.

(31.) These expectations are only valid for industrialized nations. Talvi and V[acute{e}]gh (1997) find that the sign on economic growth is reversed in Latin American countries. They hypothesize that governments can only justify painful cuts in expenditures or increases in taxes to their constituencies when economic conditions worsen.

(32.) Hallerberg and von Hagen (1999) refer to a strong finance minister as a "delegation" approach and to negotiated targets as a "contract" approach.

(33.) In particular, Franzese codes elections for relevance to economic policy. Policy-relevant elections can include, for example, presidential elections in Finland, France, and the United States as well as upper house elections. He then divides the value of the variable between those separate elections, so that, for example, a French presidential election and a French parliamentary election are each coded .5.

(34.) In unreported results, we restricted the De Haan and Sturm data just to European Union countries to test whether the differences between the two data sets were due to regional factors, but the gap remained.

(35.) Oatley notes that his results disappear in the 1990s, which, because that is the period when capital arguably became most mobile, would seem to be prima facie evidence against his argument. More generally, he finds supporting evidence for approximately half the hypotheses.

(36.) Note that the inclusion of the central bank as an important actor leads to somewhat different predictions from those of Oatley (1999). In particular, we expect independent central banks to prevent monetary expansions, and they even may lead to monetary contractions under leftist governments when capital is immobile. Moreover, under conditions of immobile capital and a dependent central bank, the partisan cycle shows up in our model only on the fiscal side.

(37.)We want to be clear about one important difference between the EMU case and the more general cases examined in the empirical section of the article. As explained in note 10, when countries control both fiscal and monetary policy, a fiscal expansion under a fixed exchange rate leads to a reinforcing monetary expansion to maintain the exchange rate. With monetary policy controlled by the European Central Bank, however, states face a "fixed" exchange rate, but they do not anticipate a similar expansion of the money supply. Therefore, the effect of the fiscal expansion may be less pronounced in EMU countries than in others that still have a central bank that controls the money supply. We thank Fabia Ghironi for thoughtful comments on this issue.

(38.)Eichengreen and Wyplosz (1997) provide a good discussion of the different rationales for the Stability and Growth Pact.

(39.) The text of the agreement appears in Committee on Economic and Monetary Affairs and Industrial Policy 1997.

(40.) Personal interviews with two members of Directorate General II, June and July 1997.

(41.) One additional tool that could be used to restrict governmental abuses will also be absent under EMU. In this article we have assumed that the decisions to pursue capital mobility or fixed exchange rates are exogenous. One could argue, of course, that when banks are independent and governments are especially fiscally irresponsible the bank may decide to stop defending the currency and simply let it float, putting the country in the flexible exchange rate capital mobility scenario that is advantageous to the bank. Under EMU, however, this is extremely unlikely.

(42.) The ultimate form these institutions should take is still open to discussion. Strauch (1998) finds that numerical restrictions on the size of the deficit do restrict the size of the deficit. Von Hagen (1998) notes, however, that such numerical restrictions can merely encourage politicians to be more creative with their accounting, and he suggests more centralized procedures to ensure that someone monitors the actors.

REFERENCES

Alesina, Alberto, and Nouriel Roubini, with Gerald Cohen. 1997. Political Cycles and the Macroeconomy. Cambridge, MA: MIT Press.

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

Al Marhubi, Fahim; and Thomas D. Willett. 1995. "The Antiinflationary Influence of Corporatist Structures and Central Bank Independence: The Importance of the Hump-Shaped Hypothesis." Public Choice 84 (July): 153-62.

Alvarez, R. Michael, Geoffrey Garrett, and Peter Lange. 1991. "Government Partisanship, Labor Organization, and Macroeconomic Performance." American Political Science Review 85 (June):539-56.

Andrews, David M. 1994. "Capital Mobility and State Autonomy: Toward a Structural Theory of International Monetary Relations." International Studies Quarterly 38 (June): 193-218.

Beck, Nathaniel. 1987. "Elections and the Fed: Is There a Political Monetary Cycle?" American Journal of Political Science 31 (February): 194-216.

Beck, Nathaniel, and Jonathan N. Katz. 1995. "What To Do (and Not To Do) with Time-Series Cross-Section Data." American Political Science Review 89 (September): 634-47.

Berger, Helge. 1997a. "The Bundesbank's Path to Independence: Evidence from the 1950s." Public Choice 93 (December): 427-53.

Berger, Helge. 1997b. Konjunkturpolitik im Wirtschaftswunder, Hand-lungsspielr[ddot{u}]ume und Verhaltensmuster von Bundesbank und Regierung in den l950er Jahren (Stabilization Policy during the Wirtschaftswunde: Scope of Action and Behavioral Patterns of the Bundesbank and the Government). T[ddot{u}]bingen: J. C. B. Mohr (Paul Siebeck).

Berger, Helge, and Marcel Thum. N.d. "News Management in Monetary Policy: When Central Banks Should Talk to the Government." German Economic Review. Forthcoming.

Berger, Helge, and Friedrich Schneider. N.d. "The Bundesbank's Reaction to Policy Conflicts." In 50 Years of Bundesbank: Lessons for the ECB, ed. Jakob de Haan. London: Routledge. Forthcoming.

Bernhard, William, and David LeBlang. 1999. "Democratic Institutions and Exchange Rate Commitments." International Organization 54 (Winter): 71-97.

Blinder, Alan S. 1998. Central Banking in Theory and Practice. Cambridge, MA: MIT Press.

Borrelli, Stephen A., and Terry J. Rayed. 1995. "Government 'Strength' and Budget Deficits in Advanced Democracies." European Journal of Political Research 28 (September): 225-60.

Branson, William H., and Willem H. Buiter. 1983. "Monetary and Fiscal Policy with Flexible Exchange Rates." In Economic Interdependence and Flexible Exchange Rates, ed. Jagdeep S. Bhandari and Bluford H. Putnam. Cambridge, MA: MIT Press. Pp. 251-85.

Burdekin, Richard C. K., and Leroy O. Laney. 1988. "Fiscal Policymaking and the Central Bank Institutional Constraint." Kyklos 41 (4): 647-62.

Canzoneri, Mathew B., and Behzad T. Diba. 1999. "The Stability and Growth Pact: A Delicate Balance or an Albatross." Georgetown University. Typescript.

Cerny, Phillip. 1994. "The Dynamics of Financial Globalization: Technology, Market Structure, and Policy Response." Policy Sciences 27 (4): 319-33.

Clark, William Roberts. N.d. Capitalism, Not Globalism: Capital Mobility, Central Bank Independence and the Political Control of the Economy. Ann Arbor: University of Michigan Press. Forthcoming.

Clark, William Roberts, and Usha Nair Reichert. 1998. "International and Domestic Constraints on Political Business Cycle Behavior." International Organization 52 (Winter): 87-120.

Cohen, Benjamin J. 1993. "The Triad and the Unholy Trinity: Lessons for the Pacific Region." In Pacific Economic Relations in The 1990s: Cooperation or Conflict?, ed. Richard Higgot, Richard Leaver, and John Ravenhill. Boulder, CO: Lynne Rienner. Pp. 133-58.

Coffey, Peter. 1984. The European Monetary System: Past, Present, and Future. Hingham, MA: Kluwer Academic.

Committee on Economic and Monetary Affairs and Industrial Policy. 1997. "Text of the Two Council Regulations on the Stability and Growth Pact Approved by the European Council in Amsterdam 16 June 1997." Luxembourg: The European Commission.

Corsetti, Giancarlo, and Paolo Pesenti. 1997. "Welfare and Macroeconomic Interdependence." National Bureau of Economic Research Working Paper No. 6307.

Cukierman, Alex, Pantelis Kalaitzidakis, Lawrence H. Summers, and Steven B. Webb. 1993. "Central Bank Independence, Growth, Investment, and Real Rate." Carnegie-Rochester Conference Series on Public Policy 39 (Autumn): 95-140.

Cukierman, Alex, and Steven B. Webb. 1994. "Political Influence on the Central Bank: International Evidence." Paper presented at the annual meeting of the American Political Science Association, New York.

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 (September): 353-98.

De Haan, Jakob, and Jan-Egbert Sturm. 1994a. "The Case for Central Bank Independence." In The Theory of Inflation, ed. Michael Parkin. Aldershot: Edward Elgar. Pp. 627-49.

De Haan, Jakob, and Jan-Egbert Sturm. 1994b. "Political and Institutional Determinants of Fiscal Policy in the European Community." Public Choice 80 (July): 157-72.

De Haan, Jakob, and Jan-Egbert Sturm. 1997. "Political and Economic Determinants of OECD Budget Deficits and Government Expenditures: A Reinvestigation." European Journal of Political Economy 13 (December): 739-50.

De Long, Bradford, and Lawrence H. Summers. 1992. "Macroeconomic Policy and Long-Run Growth." Economic Review 4: 5-29. Federal Reserve Bank of Kansas City.

Dooley, Michael. 1996. "A Survey of Literature on Controls over International Capital Transactions." IMF Staff Papers 43 (September): 639-87.

Dornbusch, Rudiger. 1976. "Expectations and Exchange Rate Dynamics." Journal of Political Economy 84 (December): 1161-76.

Edin, Per-Anders, and Henry Ohlsson. 1991. "Political Determinants of Budget Deficits: Coalition Effects Versus Minority Effects." European Economic Review 35 (December): 1597-603.

Eichengreen, Barry, and J[ddot{u}]rgen von Hagen. 1996a. "Federalism, Fiscal Restraints, and European Monetary Union." American Economic Review 86 (May): 134-8.

Eichengreen, Barry, and J[ddot{u}]rgen von Hagen. 1996b. "Fiscal Restrictions and Monetary Union: Rationales, Repercussions, Reforms." Empirica: Journal of Applied Economics and Economic Policy 23 (1): 3-23.

Eichengreen, Barry, and Charles Wyplosz. 1997. "The Stability Pact: More than a Minor Nuisance?" University of California at Berkeley. Working paper.

Eijffinger, Sylvester C., and Jakob de Haan. 1996. "The Political Economy of Central-Bank Independence." Special Papers in International Economics 19. Princeton, NJ: Princeton University.

Europa World Year Book. Various years. London: Europa Publications.

Fleming, J. Marcus. 1962. "Domestic Financial Policies under Fixed and Floating Exchange Rates." IMF Staff Papers 9: 369-80.

Franzese, Robert J. 1996. "The Political Economy of Public Debt: An Empirical Examination of the OECD Post-War Experience." Paper presented at the annual meeting of the Midwest Political Science Association, Chicago.

Franzese, Robert J., and Peter A. Hall. 1998. "Mixed Signals: Central Bank Independence, Coordinated Wage-Bargaining, and European Monetary Union." International Organization 52 (Summer): 505-35.

Frieden, Jeffry A. 1991. "Invested Interests: The Politics of National Economic Policies in a World of Global Finance." International Organization 45 (Autumn): 425-51.

Friedrich, Robert J. 1982. "In Defense of Multiplicative Terms in Multiple Regression Equations." American Journal of Political Science 26 (November): 797-833.

Garrett, Geoffrey. 1995. "Capital Mobility, Trade, and the Domestic Politics of Economic Policy." International Organization 49 (Autumn): 657-87.

Garrett, Geoffrey. 1998. Partisan Politics in tire Global Economy. Cambridge: Cambridge University Press.

Garrett, Geoffrey, and Peter Lange. 1995. "Internationalization, Institutions, and Political Change." International Organization 49 (Autumn): 627-55.

Garrett, Geoffrey, and George Tsebelis. 1996. "An Institutional Critique of Intergovernmentalism." International Organization 50 (Spring): 269-99.

Garrett, Geoffrey, and Christopher Way. 1995. "Labor Market Institutions and the Economic Consequences of Central Bank Independence." Paper presented at the annual meeting of the American Political Science Association, Chicago.

Ghironi, Fabio, and Fancesco Giavazzi. 1998. "Currency Areas, International Monetary Regimes, and the Employment-Inflation Tradeoff." Journal of International Economics 45 (August): 259-96.

Grier, Kevin B. 1989. "On the Existence of a Political Monetary Cycle." American Journal of Political Science 33 (May): 376-89.

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

Hall, Peter A. 1994. "Central Bank Independence and Coordinated Wage Bargaining: Their Interaction in Germany and Europe." German Politics and Society 31 (Spring): 1-23.

Hallerberg, Mark, and Scott Basinger. 1998. "Internationalization and Changes in Tax Policy in OECD Countries: The Importance of Domestic Veto Players." Comparative Political Studies 31 (June): 321-53.

Hallerberg, Mark, and William Roberts Clark. 1997. "How Should Political Scientists Measure Capital Mobility?" Paper presented at the annual meeting of the American Political Science Association, Washington, DC.

Hallerberg, Mark, and J[ddot{u}}rgen von Hagen. 1998. "Electoral Institutions and the Budget Process." In Democracy, Decentralization and Deficits in Latin America, ed. Kiichiro Fukasaku and Ricardo Hausmann. Paris: Organisation for Economic Cooperation and Development. Pp. 65-94.

Hallerberg, Mark, and J[ddot{u}]rgen von Hagen. 1999. "Electoral Institutions, Cabinet Negotiations, and Budget Deficits within the European Union." In Fiscal Institutions and Fiscal Performance, ed. James Poterba and J[ddot{u}]rgen von Hagen. Chicago: University of Chicago Press. Pp. 209 -32.

Hibbs, Douglas A. 1977. "Political Parties and Macroeconomic Policy." American Political Science Review 71 (December): 1467-87.

International Monetary Fund. Various years a. Exchange Arrangements and Exchange Restrictions. Washington, DC: International Monetary Fund.

International Monetary Fund. Various years b. International Financial Statistics. Washington, DC: International Monetary Fund.

Iversen, Torben. 1998. "Wage Bargaining, Central Bank Independence and the Real Effects of Money." International Organization 52 (Summer): 469-504.

Jaccard, James, Robert Turrisi, and Choi K. Wan. 1990. Interaction Effects in Multiple Regression. Berkeley, CA: Sage.

Katzenstein, Peter. 1985. Small States in World Markets. Ithaca, NY: Cornell University Press.

Keech, William R. 1995. Economic Politics: The Costs of Democracy. New York: Cambridge University Press.

Kurzer, Paulette. 1993. Business and Banking: Political Change and Economic Integration in Western Europe. Ithaca, NY: Cornell University Press.

Lohmann, Susanne. 1992. "Optimal Commitment in Monetary Policy: Credibility versus Flexibility." American Economic Review 82 (March): 273-86.

Mackie, Thomas T., and Richard Rose. 1982. The International Almanac of Electoral History. 2d ed. New York: Facts on File.

MacRae, C. Duncan. 1977. "A Political Model of the Business Cycle." Journal of Political Economy 85 (April): 239-63.

Mayer, Thomas, ed. 1990. The Politics of American Monetary Policy. Cambridge: Cambridge University Press.

McNamara, Kathleen R. 1998. The Currency of Ideas: Monetary Politics in the European Union. Ithaca, NY: Cornell University Press.

Moses, Jonathon W. 1994. "Abdication from National Policy Autonomy: What's Left to Leave?" (response to Ton Notermans). Politics & Society 22 (June): 125-73.

Mundell, Robert A. 1963. "Capital Mobility and Stabilization Policy under Fixed and Flexible Exchange Rates." Canadian Journal of Economics and Political Science 29 (November): 475-85.

Nordhaus, William D. 1975. "The Political Business Cycle." Review of Economic Studies 42 (April): 169-90.

Notermans, Ton. 1993. "The Abdication from National Policy Autonomy: Why the Macroeconomic Policy Regime Has Become So Unfavorable to Labor." Politics & Society 21 (June): 133-67.

Oatley, Thomas. 1999. "How Constraining Is Capital Mobility? The Partisan Hypothesis in an Open Economy." American Journal of Political Science 43 (October): 1003-27.

Obstfeld, Maurice, and Kenneth Rogoff. 1996. Foundations of International Macroeconomics. Cambridge, MA: MIT Press.

OECD. 1985. Exchange Rate Management and the Conduct of Monetary Policy. Paris: Organization for Economic Cooperation and Development.

OECD. Various years a. OECD Economic Outlook. Paris: Organization for Economic Cooperation and Development.

OECD. Various years b. OECD Main Economic Indicators. Paris: Organization for Economic Cooperation and Development.

Pauley, Louis W. 1995. "Capital Mobility, State Autonomy and Political Legitimacy." Journal of International Affairs 48 (Winter): 369-88.

Pollard, Patricia S. 1993. "Central Bank Independence and Economic Performance." Review 75 (July/August): 21-36. Federal Reserve Bank of St. Louis.

Poterba, James M., and Kim S. Rueben. 1999. "State Fiscal Institutions and the U.S. Municipal Bond Market." In Fiscal Institutions and Fiscal Performance, ed. James Poterba and J[ddot{u}]rgen von Hagen. Chicago: University of Chicago Press. Pp. 181-207.

Powell, G. Bingham, and Garry D. Whitten. 1993. "A Cross-National Analysis of Economic Voting: Taking Account of the Political Context." American Journal of Political Science 37 (May): 391-414.

Rodrik, Dani. 1997. Has Globalization Gone Too Far? Washington, DC: Institute for International Economics.

Rogoff, Kenneth. 1985. "The Optimal Degree of Commitment to an Intermediate Monetary Target." Quarterly Journal of Economics 100 (November): 1169-89.

Roubini, Nouriel, and Jeffrey D. Sachs. 1989. "Political and Economic Determinants of Budget Deficits in the Industrial Democracies." European Economic Review 33 (May): 903-33.

Simmons, Beth A. 1996a. "Capital Mobility and National Politics: The Capital Constraint and Fiscal Policy, 1967-1990." Paper presented at the annual conference of the International Studies Association, San Diego, CA.

Simmons, Beth A. 1996b. "Rules of the Game: Central Bank Independence During the Interwar Years." International Organization 50 (Summer): 407-43.

Sinn, Hans-Werner. 1992. "The Case for European Harmonization." In Tax Harmonization and Financial Liberalization in Europe, ed. Georg Winkler. New York: St. Martin's. Pp. 3-7.

Statistical Annex of European Economy. Various years. European Commission: Luxembourg.

Stigler, George J., and Gary S. Becker. 1977. "De Gustibus Non Est Disputandum." American Economic Review 67 (March): 76-90.

Strauch, Rolf. 1998. "Budget Processes and Fiscal Discipline: Evidence from U.S. States." University of Bonn. Typescript.

Strange, Susan. 1996, The Retreat of the State. Cambridge: Cambridge University Press.

Svensson, Lars E. O. 1995. "Optimal Inflation Targets, Conservative Central Banks, and Linear Inflation Contracts." National Bureau of Economic Research Working Paper No. 5251.

Talvi, Ernesto, and Carlos V[acute{e}]gh. 1997. "Can Optimal Fiscal Policy Be Procyclical?" Inter-American Development Bank. Typescript.

Tsebelis, George. 1995. "Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism, Multicameralism and Multipartyism." British Journal of Political Science 25 (July): 289-325.

Tufte, Edward R. 1978. The Political Control of the Economy. Princeton, NJ: Princeton University Press.

Von Hagen, J[ddot{u}]rgen. 1998. "Budgeting Institutions for Aggregate Fiscal Discipline." Zentrum f[ddot{u}]r Europ[ddot{a}]isehe Integrationsforschung Working Paper B98-01.

Wallerstein, Michael, and Adam Przeworski. 1995. "Capital Taxation with Open Borders." Review of International Political Economy 2 (3): 425-45.

Webb, Michael C. 1995. The Political Economy of Policy Coordination: International Adjustment since 1945. Ithaca, NY: Cornell University Press.

Woldendorp, Jaap, Hans Keman, and Ian Budge, eds. 1993. "Special Issue: Political Data 1945-1990." European Journal of Political Research 24 (July): 1-119.

Woldendorp, Jaap, Hans Keman, and Ian Budge. 1998. "Party Government in 20 Democracies: An Update (1990-1995)." European Journal of Political Research 33 (January): 125-64.

APPENDIX A. DERIVATION OF PROPOSITION 1

The reaction functions in proposition 1 follow directly from the actor's loss functions and the Phillips curve mechanism. The government's problem is to choose g so as to minimize its loss function. If we substitute in the Phillips curve process of equation 4, which determines y, and the right-hand side of equation 2 for the government's ideal point for output, and if we assume (without loss of generality) that [[pi].sup.*] = 0, the loss function becomes:

[L.sub.i] = [([y.sup.n] + [mu]([pi] - [[pi].sup.e]) + [phi]g - [k.sup.gov][y.sup.n]).sup.2] + [alpha][[pi].sup.2]. (A-1)

To find the minimum, we take the partial derivative with respect to g:

[frac{[partial]L}{[partial]g}] = 2[phi]([y.sup.n] + [mu]([pi] - [[pi].sup.e] + [phi]g - [k.sup.gov][y.sup.n]), (A-2)

which, when set equal to zero and solved for g, becomes

g = [frac{1}{[phi]}][[y.sup.n]([k.sup.gov] - 1) - [mu]([pi] - [[pi].sup.e])] (A-3)

as proposed.

An analogous process can be used to determine the central bank's optimal response. The problem is to find the level of inflation that minimizes its loss function. Once again, substituting the Phillips curve process of equation 4 as well as the right-hand side of equation 3 for the bank's ideal point for output, and assuming (without loss of generality) that [[pi].sup.*] = 0, the bank's loss function is:

[L.sub.cb] = [([y.sup.n] + [mu]([pi] - [[pi].sup.e]) + [phi]g - [k.sup.cb][y.sup.n]).sup.2] + [alpha][[pi].sup.2]. (A-4)

Differentiating with respect to [pi] yields:

[frac{[partial]L}{[partial][pi]}] = 2[mu]([y.sup.n] + [mu]([pi] - [[pi].sup.e]) + [phi]g - [k.sup.cb][y.sup.n]) + 2[alpha][pi], (A-5)

which, when set to zero and solved for [pi], becomes

[pi] = [frac{1}{[mu] + [frac{[alpha]}{[mu]}]}][[mu][[pi].sup.e] + [y.sup.n]([k.sup.cb] - 1) - [phi]g] (A-6)

as proposed.

APPENDIX B

We will briefly describe indicators used in the empirical section. These fall into two categories: institutional/political and macroeconomic.

Institutional/Political Variables

Given the qualitative nature of our theoretical discussion, we elected to classify our observations primarily according to categorical distinctions. Central banks are viewed as independent or not; exchange rates are fixed or flexible. We have neither strong theoretical arguments about such operational choices nor reason to believe that the strategic calculations of policymakers are influenced in a linear fashion by changing degrees of capital mobility or central bank independence. Table B-1 reports the classifications we used.

Central Bank Independence: The Cukierman, Webb, and Neyapti (1992) measure of legal independence was used to construct a categorical variable that equals 1 if the country's score is above the sample median, 0 otherwise.

Elections: For monetary tests, quarters in which a general election is held, and the three preceding quarters, are coded as electoral periods. Dates for elections are taken from Mackie and Rose (1982) and Europa World Year Book (various years). For fiscal tests, we used either the year. in which an election was held, with the data provided in Hallerberg and von Hagen (1999) and based on the yearly data appendix in European Journal of Political Research (various years), or the percentage of time in a given year before an election for a body important for macroeconomic policy as coded by Franzese (1996) and updated by the authors May 1998. (More details for this data source are provided in note 33.)

Fixed Exchange Rates: This categorical variable equals 1 when a country either belongs to a fixed exchange rate regime or pegs its currency to another national currency, 0 otherwise. Codings are based on the IMF's Exchange Arrangements and Exchange Restrictions (various years a).

Fiscal Institutions--Strong Finance Minister and Negotiated Targets: These variables are coded 1 when the institution is present, 0 otherwise. Strong finance ministers generally serve as agenda-setters on the budget, have monitoring functions over the budgets of other ministries, and can strike out spending on some occasions when it is deemed excessive. Negotiated targets are present when coalition partners negotiate budgets for every ministry. The alternative, termed the "fiefdom" approach, is for coalition partners to negotiate the distribution of cabinet portfolios and cabinet meetings among all ministers who determine the budget. Data appear in Hallerberg and von Hagen (1999).

Left: This variable is the percentage of cabinet seats held by leftist parties. It is coded from the data on cabinet compositions in Woldendorp, Keman, and Budge (1993) and from the yearly data appendix in European Journal of Political Research (various years).

Government Type: Roubini and Sachs (1989) code this variable as follows: 0 = one-party majority government, 1 = 2-3-party majority government, 2 = 4-5-party majority government, 3 = minority government. Our data for this variable come from De Haan and Sturm (1997), who update the data set of Roubini and Sachs. Edin and Ohlsson (1991) break this variable into three separate dummies. Hallerberg and von Hagen (1999) use this formulation, and we received the data which are based on the yearly data appendix in European Journal of Political Research (various years), from them.

Partisanship: Data for this variable come from Woldendorp, Keman, and Budge (1998). The coding they use is for the political complexion of parliament and government. The values range from 1 for right-wing dominance to 5 for left-wing dominance. Note that they do not have data for Greece, Portugal, Spain, and the United States. With the exception of Portugal, these countries have governments that are clearly on the Left or Right, and we code them ourselves (1 if a right-wing government was in place, 5 if a left-wing government was in place).

Macroeconomic Variables

Change in Debt Costs: The change in debt servicing costs is computed as the change in the real interest rate minus the change in the growth rate times the gross deficit in the previous year. Real interest rate data for the De Haan and Sturm regressions came from OECD Economic Outlook (various years a), while the data for the Hallerberg and von Hagen regressions came from Statistical Annex of European Economy (various years).

Change in Gross Debt: This variable is expressed in terms of gross government debt over GDP. The data for the De Haan and Sturm regressions came from OECD Economic Outlook (various years a), and the data for the Hallerberg and von Hagen regressions came from Statistical Annex of European Economy.

Growth: For monetary tests, growth in output was measured by seasonally adjusted total industrial production, except for Canada (where data were not seasonally adjusted) and Australia (where seasonally adjusted real gross product volume was used), from OECD Main Economic Indicators (various years b). Other studies have used change in real GDP, but seasonally adjusted quarterly data were unavailable for several countries. For fiscal tests, figures for the De Haan and Sturm regressions came from OECD Economic Outlook (various years a), and data for the Hallerberg and von Hagen regressions came from Statistical Annex of European Economy (various years).

Inflation: This variable was based on the consumer price index for all goods, except for Japan (all goods less food). It was calculated as ([Cpi.sub.t]/[Cpi.sub.t], - [Cpi.sub.t-4]). The source was International Monetary Fund, International Financial Statistics, CD-ROM version.

Long-Term Interest Rates: Data for the De Haan and Sturm regressions came from OECD Economic Outlook (various years a), and data for the Hallerberg and von Hagen regressions came from Statistical Annex of European Economy (various years).

Money Supply: M1 and M1 + quasimoney came from OECD, OECD Main Economic Indicators (various years b).

Unemployment: For monetary tests, the quarterly percentage change in the seasonally adjusted number of unemployed was used, from OECD, OECD Main Economic Indicators (various years b). For fiscal tests, unemployment figures for the De Haan and Sturm regressions came from OECD Economic Outlook (various years a), and data for the Hallerberg and von Hagen regressions came from Statistical Annex of European Economy (various years).
COPYRIGHT 2000 Cambridge University Press
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2000 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:CLARK, WILLIAM ROBERTS; HALLERBERG, MARK
Publication:American Political Science Review
Geographic Code:00WOR
Date:Jun 1, 2000
Words:18977
Previous Article:Free to Trade: Democracies, Autocracies, and International Trade.
Next Article:Reflections on Fear: Montesquieu in Retrieval.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters