Fiscal-Monetary Policy Mix: An Investigation of Political Determinants of Macroeconomic Policy Mixes. Takayuki Sakamoto

Transcription

1 Fiscal-Monetary Policy Mix: An Investigation of Political Determinants of Macroeconomic Policy Mixes Takayuki Sakamoto Department of Political Science Southern Methodist University P.O. Box Dallas, TX Phone: Fax: Abstract It has become common to probe both fiscal and monetary policies in analyzing industrialized economies. But previous studies do not tell us exactly what fiscal-monetary policy mixes are concurrently used by different governments. This paper examines the policy mixes and the factors affecting them, using cyclically adjusted measures of discretionary fiscal and monetary policies. The data suggest that fiscal and monetary policies move in response to one another. Partisan governments and central banks in tandem affect policy mixes. The findings are: left governments use a tight fiscal-loose monetary policy mix more than the center or right; independent central banks restrain party governments fiscal policy; independent central banks counter center and right governments loose fiscal policy with tight monetary policy; expansions in election years are manufactured by loose fiscal policy, and accompanied by a tightening of monetary policy; and expansions during recessions are crafted by loose monetary policy, but not so much fiscal policy.

2 It has become a more common practice to refer to the roles of both fiscal and monetary policies in the comparative political economy literature. But the fiscal-monetary policy mix has not been explicitly studied. We currently know little about how these instruments are concurrently used, and whether or not different governments use different policy mixes, not to mention whether there are systematic patterns in the way the two instruments are concurrently used or how distinct policy mixes affect economic performance. This paper is a first-cut attempt to investigate what policy mixes are used by governments and what factors affect the choice of policy mixes. The next section briefly reviews the studies touching on the issue. Then, I explain the method I use in examining policy mixes and their determinants, and present the results. In the final section, I attempt an explanation for the results and discuss implications. I use cyclically adjusted measures of discretionary fiscal and monetary policies to measure policy stances. To anticipate the findings, the data show that fiscal and monetary policies move in response to each other. Partisan governments and central banks in tandem affect what policy mixes are used. Among other findings are: (1) largely contrary to previous studies findings, left governments use a tight fiscal-loose monetary policy mix more than the center or right. (2) Independent central banks restrain the use of loose fiscal policy by party governments, and their monetary policy is often not so much tight as neutral. (3) Independent central banks counter center and right governments loose fiscal policy with tight monetary policy, and center governments run expansionary fiscal policy in the absence of independent central banks. (4) Expansions in election years are manufactured by loose fiscal policy, and accompanied by a tightening of monetary policy. (5) Expansions during recessions are crafted by loose monetary policy, but not so much fiscal policy, regardless of central bank independence. (6) Coalition governments do not necessarily run loose macroeconomic policy 1

3 more than single-party governments. Independent central banks also restrain coalition governments macroeconomic policy. REVIEW OF THE LITERATURE Research has long referred to macroeconomic policy stances such as tight or loose fiscal and monetary policies. But it is only in recent years that researchers have started to explicitly examine the interaction of the two policy tools. Economists studies have been limited almost exclusively to theoretical exposition of what policy mixes are likely to be used by policy makers, and there have been very few empirical studies. Economists typically employ noncooperative game-theoretical models, and conclude that conflict between fiscal and monetary authorities policy preferences generates a loose fiscal-tight monetary policy mix and, as a result, produces suboptimal outcomes; higher deficits and higher interest rates, or lower output and higher inflation. In their game-theoretical models, this undesirable outcome results because fiscal and monetary policy makers respectively use their own policy tool to undermine each other (Nordhaus, 1994; Bennett and Loayza, 2002; Demertzis et al., 1998; Dixit and Lambertini, 2002). The intensity of policy conflict increases when the party government is leftist, because the government s and the monetary authority s policy preferences diverge more, resulting in more undesirable outcomes. Melitz (1997) is one of the few economists who has empirically studied the responses of fiscal and monetary policies to each other, and concludes that the two policies move in opposite directions among OECD countries. (By contrast, there are a number of empirical studies on the cyclicality of discretionary fiscal policy. See, for example, Buti and van den Noord, 2003; Alesina and Perotti, 1995; Gali and Perotti, 2003; Lane, 2002.) Political scientists have also begun to speak of a fiscal-monetary policy mix in recent years, focusing on the effects of government partisanship, central bank independence, and labor 2

4 organization on policy instruments (deficits and interest rates), or on macroeconomic performance (inflation, unemployment, and GDP growth) (Boix, 2000; Clark, 2003; Clark and Hallerberg, 2000; Garrett, 1998; Iversen, 1999; Oatley, 1999; Way, 2000). While existing studies examine the effects of different political-economic institutions and policy makers on economic policy or performance, they are not exactly about fiscal-monetary policy mixes. They do not tell us what policy mixes are concurrently pursued by fiscal and monetary authorities. Further, some of these studies, instead of actually analyzing the nature of policies, tend to proceed with analysis by indirectly inferring the contents of economic policies from the policy makers institutional or partisan attributes, such as central bank independence and government partisanship, or their presumed policy preferences for output and employment versus price stability. So these studies are conducted based on such assumptions as Central banks goal is price stability, so their monetary policy is contractionary, and Left governments care more about employment and growth than price stability, so their fiscal policy is expansionary. In this case, analysis is about policy makers presumed policy preferences or their institutional attributes, not about their actual policies. Bearce (2002) has more explicitly discussed an economic policy mix. He argues that left governments adopt a loose fiscal-tight monetary policy mix, whereas right governments employ a tight fiscal-loose monetary policy. He explains that if governments use an expansionary fiscal policy to promote growth, then, they use a tight monetary policy to contain inflation, and vice versa. He argues that left governments choose fiscal over monetary policy for growth strategies because fiscal policy is more suited for targeted use to bring benefits to their core supporters, such as labor and low income groups. By contrast, right governments choose monetary policy for growth strategies, because they receive electoral support from capital interests engaged in 3

5 international business, and they both prefer a smaller state. He also claims that multiparty governments pursue a loose fiscal-tight monetary policy mix, because their need to maintain their coalitions lead them to engage in income redistribution and fiscal policy is more suited for redistribution and targeted economic growth, while single-party governments pursue a tight fiscal-loose monetary policy mix. In contrast, Iversen and Soskice (1999) suggest that governments use fiscal and monetary policies in a mutually consistent manner (both accommodating fiscal and monetary policies, or both non-accommodating policies) to maximize the efficacy of the policies. Many economists also seem to believe that coordinated, consistent policy mixes are better in riding out business cycles, as long as fiscal and monetary policy makers share the same economic goals. In their view, inconsistent policy mixes neutralize and undermine the effectiveness of respective policies. Given the state of study on this topic, what we need is an explicit study of the fiscalmonetary policy mix. We need to uncover what policy mixes governments concurrently use and what political or economic factors affect their choice of policy mixes. In examining policy mixes, we also need to be more accurate than in previous studies in identifying the discretionary component of policy stances or change, because we are interested in policy stances resulting from policy makers intentional actions. To do so requires us to remove the effects of automatic stabilizers from conventionally used indicators such as budget deficits and interest rates. EMPIRICAL ANALYSIS Data and Measurement To investigate fiscal-monetary policy mixes, I examine data from 18 industrialized democracies. 1 The period covered is approximately from 1961 to The qualifier approximately is 4

6 attached because the availability of two indicators, cyclically adjusted primary balance and potential output, varies across countries. There are different ways to measure fiscal and monetary policy stances. In the past, political scientists have used such conventional measures as fiscal deficit and interest rates which are cyclically unadjusted. 2 But in this paper, we are interested in the stance of discretionary policy that is a product of policy makers intentional actions. Thus, it is not sufficient just to look at conventional cyclically unadjusted measures. We need to examine cyclically adjusted policy stances. The reason is that (to take the case of fiscal policy) in the presence of automatic stabilizers, fiscal deficit can increase or decrease even if policy makers do nothing to change their fiscal policy stance, when business cycles induce change in the tax bases and unemployment transfers. Therefore, to tap the discretionary component of macroeconomic policy, it is necessary to remove the effects of automatic stabilizers that result from business cycles. Discretionary Fiscal Policy. For the measure of discretionary fiscal policy, we use cyclically adjusted primary balance as a percentage of potential GDP (this measure excludes interest payments). The source is OECD (2003). What represents an appropriate measure of the fiscal stance the level of or change in deficit is open to debate, and the answer depends partly on the conceptualization of the policy stance and the purpose of analysis. I decided to use the level of deficit to avoid an otherwise awkward interpretation that results from the use of change which would code a government s fiscal policy stance as neutral (neither expansionary nor contractionary) when the government s annual deficit stays 6 percent for 4 consecutive years. The OECD calculates cyclically adjusted fiscal balance as follows (van den Noord, 2000): Potential output is estimated by country-specific production functions. Elasticities of 5

7 various taxation and expenditure components to output fluctuations are calculated. Then, they obtain the cyclical component of the budget balance by using the output gap and the elasticities, and subtract it from the actual balance. This method is very similar to the ones used by the IMF and the European Commissions. One could alternatively control for the effects of economic cycles, for instance, by using regression models and entering growth and unemployment as independent variables. But the OECD s cyclically adjusted balance is a more direct and precise measure because it empirically and directly derives the country-specific responses of taxation and expenditure to economic fluctuations and use them to calculate cyclically adjusted balance. The use of potential output as a reference point is not without problems and is subject to debate, because of the issue of the reliability of its measurement. Alesina and Perotti (1995) and Buti and van den Noord (2003) also correctly point out that cyclically adjusted primary balance does not take inflation into account. But we use this measure because it is still a reasonable, useful measure of discretionary fiscal policy stance, data availability is large, and this is widely used by the OECD and others (Gali and Perotti, 2003). 3 Discretionary Monetary Policy. When measuring the stance of discretionary monetary policy, we also need to remove the effects of semi-automatic monetary policy responses by central banks that are considered normal reactions to business cycles in their effort to maintain price stability. One way to control for central banks normal, automatic responses is to use a Taylortype rule and measure the stance of their discretionary policy distinct from normal policy responses suggested by such rules (Taylor, 1993; Rothenberg, n.d.). In my empirical analysis, I calculate discretionary monetary policy by subtracting the neutral interest rates suggested by the Taylor-type rule from the actual discount rates. The Taylor-type rule I use is: Taylor-rule implied discount rate (t) = 2 + inflation (t-1) +.5*(inflation (t-1) π*) +.5*output gap (t-1) 6

8 where the constant term (2) is the assumed long-run equilibrium real rate, and π* is the central bank s inflation target rate, and it is assumed to be 2 percent. The Taylor-type rule calculates a price-stability-conforming discount rate target from the past inflation rate, central banks inflation target rate, the long-term real interest rate, and the gap between real and potential GDP. Note that I am not suggesting that central banks actually use Taylor rules in their conduct of monetary policy. All I assume here is that Taylor rules suggest a reasonable response of monetary policy to economic cycles given the goal of price stability, and that we can meaningfully study the deviations of monetary policy from a neutral stance by examining the gap between real rates and Taylor-rule rates. The use of the Taylor-type rule is not without problems. The rule assumes that both the long-run real interest rate and central banks inflation target rate are 2 percent. But the calculation of the long-run real interest rate is problematic, and it is questionable that the equilibrium real interest rate and the inflation target rate are constantly 2 percent across countries and over time (Hetzel, 2000; Kozicki, 1999). In addition, the reliable measurement of potential output is not easy. But in the absence of better measures, the measure based on the Taylor-type rule is a sensible choice. We set the definitions of loose and tight fiscal and monetary policy as follows. Loose fiscal policy if discretionary fiscal policy stance -1.0 Tight fiscal policy if discretionary fiscal policy stance 1.0 Loose monetary policy if discretionary monetary policy stance -1.0 Tight monetary policy if discretionary monetary policy stance 1.0 In deciding the cut-off points, we considered two factors. First, we need to set them in such a way that loose or tight policy is really different from neutral policy. To achieve this, the 7

9 cut-off values need to be sufficiently large. Second, we also need to set them low enough that we will obtain a sufficient number of observations for loose and tight policy for analysis. Considering these two requirements, we decided to set the thresholds at -1.0 and 1.0 percent. These thresholds are admittedly arbitrary. So in all analyses below, we also experiment with the data using and 1.5-percent cut-off points. When we raise the thresholds, the number of observations for loose and tight fiscal policy naturally decreases, and neutral policy increases. But the -1.5/1.5-percent definition does not change the main results. Therefore, we only report the results obtained using the 1-percent definition. We also experiment with lags of variables to take into account the time lag between policy process, implementation, and effect. This induces minor changes in observations, but does not change the main results. So we report the results obtained with no lags. 4 The inflation and discount rates data used for calculating the monetary policy stance are from the IMF (2003). 5 The output gap data is from OECD (2003), and so are real GDP growth, primary balance, and unemployment. The definitions of political-institutions variables are as follows. Left, Center, and Right are dummy variables representing the partisan control of government, measured by cabinet portfolios (as a percentage of all cabinet portfolios) held by left-, center-, and right-parties. Left is defined as governments where leftist parties hold more than 50 percent of cabinet portfolios. Center and Right are operationalized similarly. Center includes Christian democratic and Catholic parties and other centrist parties, and Right conservative and liberal parties. The 50- percent cut-off point is arbitrary, but we chose this definition for two reasons. First, the left, center, and right governments operationalized with the 50-percent threshold approximately match those identified by other conventional measures such as Woldendorp et al. (1998). Second, it makes the three variables all mutually exclusive. If we used a threshold of 50 percent 8

10 or below, they would not be mutually exclusive and complicate analysis because there would be many cases of left-center and right-center governments. Also, if we raised the threshold to a higher score, we would have a much reduced number of observations, and this would make analysis difficult. The original raw partisanship data are from Armingeon et al. (2002). This partisanship data are almost identical to the data by Swank (n.d.) in definitions and actual values. The only major difference is that Armingeon et al. classify the German Christian Democratic Union as a center party, and Swank as a right party. There is a legitimate concern that the partisan dummies do not capture the real nature of government partisanship. So in the supplementary regression analysis, I also use left, center, and right cabinet portfolios as a percentage of all cabinet portfolios (continuous variables), as well as dummy variables, to check if the use of raw percentage scores change results. CBI is a measure of central banks independence from political control and captures their ability to pursue monetary policy without or despite political interference (the values range from 0 to 1). The data are Cukierman s (1992) index of legal central bank independence (LVAU), updated by Bernhard and Leblang for Belgium, France, Italy, and New Zealand, and by the author for Finland, Ireland, Japan, Sweden, and the United Kingdom to incorporate the changes resulting from central bank reforms in the 1990s. 6 Except in the regression analysis, the variable CBI is turned into a categorical variable representing independent and dependent central banks, divided at the means of the CBI score (.39). 7 Coalition is the raw number of governing parties. The sources are Woldendorp et al. (1998) and Keesing s Record of World Events (various years). Fiscal and Monetary Policy Stance and Policy Mixes: Results 9

11 In this section, I present the results. Reported in the tables below are the frequency and probability of a particular policy stance pursued by each type of government. The probability is calculated as the percentage of the number of observations of a policy stance taken by a government type divided by the total number of observations for the government type. In all the tables, the observed differences among cells are independent, according to various measure of association tests. Partisan Governments and Policy Mixes Table 1a shows the fiscal policy stances pursued by partisan governments. Contrary to the traditional partisan assumption (Hibbs, 1977) and consistent with some recent studies (Boix, 1998), left governments fiscal policy is the tightest, and center governments are the loosest. By contrast, the loosest monetary policy is pursued by the left, and the center s stance is the tightest, though the differences are not large (Table 1b). Tighter monetary policy under the center and right than the left seems to be the result of the former s loose fiscal policy. Table 1c shows the fiscal and monetary policy mix. The left has a more frequent incidence of policy mixes with tight or neutral fiscal policy than the center or right, and carries out loose monetary policy slightly more often the center or right. Although Bearce (2002) argues that the left pursues a loose fiscal-tight monetary policy mix, the table shows that the left uses this policy mix least often of all partisan governments. We will look at partisan governments policy mix more closely below. -- Tables 1a, 1b, and 1c about here -- Central Banks and Policy Mixes Let us see what effect central banks have on fiscal and monetary policies. Here we observe somewhat surprising patterns. Central banks manage monetary policy, so it is understandable 10

12 that they affect monetary policy. But Table 2a shows that as central banks become independent, loose fiscal policy decreases and tight fiscal policy increases. Central banks do not control fiscal policy, so we need to interpret that they affect party governments fiscal policy, and that central bank independence restrains fiscal policy as well as monetary policy. Our supplementary regression analysis reported later also shows that central bank independence significantly depresses fiscal deficit. This runs counter to most game-theoretic analyses by economists that show that antiinflationary central banks induce fiscal expansion by party governments seeking to achieve their output and employment goals by countering the former s contractionary monetary policy. We will consider the reason for this later. Monetary policy behaves in lines with conventional understanding. Dependent central banks monetary policy is looser than independent ones (Table 2b). But the former also implements tight monetary policy more often than the latter. Thus, independent central banks monetary policy tends to be neutral. In contrast, dependent central banks policy fluctuates widely between loose and tight policy, which results probably from their lack of independence and of the ability to pursue monetary policy consistently. Table 2c shows the distribution of policy mixes. Dependent central banks tend to have loose policy mixes. In contrast, independent central banks tend to have policy mixes with tight or neutral fiscal policy, but their monetary policy is not particularly tight. It appears that they do not have to run contractionary monetary policy, because when central banks are independent, fiscal policy pursued by party governments is tight. -- Tables 2a, 2b, and 2c about here -- Party Governments-Central Banks and Policy Mixes 11

13 We now look at the policy stances pursued by different combinations of party governments and central banks (six regimes). We first examine fiscal and monetary policy stances separately and then inspect their policy mixes under these regimes. --Tables 3a and 3b about here -- Fiscal stances are reported in Table 3a. Here we see that partisan governments pursue different fiscal policy depending on the independence of central banks. When central banks are independent, center governments fiscal policy is the tightest of all partisan governments. But with dependent central banks, their loose fiscal policy increases dramatically. Left governments show a similar but much milder trend their fiscal policy is also tight with independent central banks, but becomes looser with dependent ones. The right s fiscal stances are stable between independent and dependence central banks and change little. The right s fiscal policy is also the loosest of all partisanship when central banks are independent. This is consistent with anecdotal episodes of the U.S. Reagan administration, which ran expansionary fiscal policy while an independent central bank conducted restrictive monetary policy. 8 Though the left s policy becomes loose with dependent central banks, its fiscal stance is overall slightly tighter than the right, which calls into question the traditional claim that the left s fiscal policy is expansionary. 9 Table 3b shows that when central banks are independent, the left s monetary policy is the loosest of all partisan governments. This is the opposite of previous studies argument that independent central banks counter the left s expansionary fiscal policy with contractionary monetary policy. A reasonable interpretation is that central banks do not have to run restrictive policy because the left s fiscal policy is tight. With independent central banks, the center s policy is the tightest. But with dependent central banks, the center s monetary policy becomes the loosest, and the left s the tightest. The pattern here is that if central banks are dependent, 12

14 monetary policy becomes loose under center and right governments, and tight under the left. So independent central banks restrain both the fiscal and monetary policies of center governments, and the center s policy becomes sharply loose with dependent central banks. -- Table 3c about here -- Policy mixes under the six regimes are reported in Table 3c. When central banks are independent, the left s policy stance tends to be tight fiscal-loose monetary mixes; fiscal policy tends to be either tight or neutral, and monetary policy either loose or neutral. The left s loose fiscal-tight monetary policy mix predicted by the conventional argument like Bearce (2002) has no observation. This result dispels the argument that the left runs expansionary fiscal policy and independent central banks counter it with tight monetary policy. The right s policy mixes under independent central banks do not have a clear pattern, and its fiscal and monetary policies are either tight or loose with a few neutral observations. But the policy stance used by the right most often is the loose fiscal-tight monetary mix, which again goes against the argument that the right engages in the tight fiscal-loose monetary mix or that central banks do not have to resort to contractionary monetary policy because the right prefers price stability and pursues tight fiscal policy. One might counter this result by saying that it is unduly influenced by cases of conservative governments such as the U.S. Reagan administration that carried out a loose fiscaltight monetary policy mix in the 1980s. But this does not hold because even when only the data prior to the 1980s are examined, the left engages in a tight fiscal-loose monetary mix more often than the right. Policy mixes employed by center governments under independent central banks are tighter than the left or right (either tight or neutral policy for both fiscal and monetary policies). But when central banks are dependent, the center s policy mix sharply turns loose. The center 13

15 actually engages in the loose fiscal-loose monetary mix half of the time (48%). Thus, independent central banks restrain the center s policy mix. With dependent central banks, right and left governments loose fiscal-loose monetary mix also increases, though not as much as the center. In the left s case, this happens because fiscal policy gets looser, compared to when central banks are independent. The left s loose fiscal-tight monetary policy mix also increases, but the left still runs the tight fiscal-loose monetary mix at least as much as the loose-tight mix. As for the right, loose fiscal policy decreases mildly, and loose monetary policy increases. It is interesting to note that when central banks are dependent, the right and center use the tight-tight and loose-loose mixes most often of all mixes. Considering that the left also pursues the looseloose and tight-tight mixes more often than when central banks are independent, it indicates that dependent central banks accommodate party governments fiscal policy be it expansionary or contractionary due to lack of independence. -- Table 4 about here -- Economic Cycles and Policy Mixes We now examine whether and how economic cycles affect the policy mix. Following Alesina and Perotti (1995), I define a recession as a year when the GDP growth rate is at least 1 percent below the average of the previous two years. Table 4 shows that, overall, policy mixes become looser during recessions, and are sensitive to economic fluctuations. Relatively tight policy mixes are used more often during non-recessions. There is no surprise here. But there is an interesting pattern; monetary policy is more clearly countercyclical than fiscal policy. (Though not reported here, separate tables for fiscal and monetary policies more clearly show this pattern.) During recessions, policy mixes including loose monetary policy increase, and during non-recessions, those with tight monetary policy increase. Fiscal policy is looser during 14

16 recessions than non-recessions, but does not move countercyclically as much as monetary policy does. Thus, monetary policy more than fiscal policy is actively used by governments as a countercyclical policy instrument to fight recessions. This is counterintuitive because the conventionally popular explanation is that governments (particularly, the left) respond to economic downturns by running loose fiscal policy. If central banks were not independent from political control, this would still be reasonable because in this case party governments could choose either fiscal or monetary policy to respond to recessions. By contrast, if central banks were independent, party governments would not have much control over monetary policy and might have difficulty using monetary policy countercyclically. So I examined whether this countercyclicality of monetary policy results simply from lack of central bank independence and party governments monetary control, by controlling for central bank independence (results not reported). Somewhat surprisingly, the degree of central bank independence does not change the pattern, although central bank independence makes overall policy mixes slightly tighter (mostly, fiscal policy). Thus, whether or not central banks are dependent, monetary policy more than fiscal policy is used as a countercyclical instrument to fight recessions. This is confirmed also in the regression analysis reported below. It is not the case that monetary policy just looks in our data to overreact to recessions simply because our measure of discretionary monetary policy misses the effect of central banks' response to economic downturns and takes into account only their response to inflation. In our Taylor-rule based measure, central banks loosen their monetary policy to respond to output gap as well as too low inflation. So our data suggest that even after taking into account such normal response by central banks, monetary policy becomes loose over and above their normal, neutral response during recessions. Thus, it seems reasonable to believe that central banks are 15

17 concerned as much with slow economic growth as with inflation, and actively join governments efforts to mitigate recessions. The conventional assumption of central banks monetary conservatism seems to understate their concern with growth. It is also not the case that fiscal conservatism in the last two decades deprived governments of fiscal policy as a countercyclical tool, leaving them with no choice but to rely on monetary policy. The reason is that the large countercyclicality of monetary policy is observed both before and after the early 1980s. I also checked whether and how recessions affect party governments policy mix (results not reported). As for fiscal policy, the left s policy is tighter than the center or right, particularly during recessions (which is inconsistent with the findings that the left s fiscal policy turns expansionary during economic downturns (Cusack, 1999)). Loose fiscal policy by the left actually decreases during recessions, though tight policy also decreases, leaving most observations in neutral policy. In contrast, the right s and the center s fiscal policy is sensitive to economic fluctuations in a countercyclical direction it turns very loose during recessions. The center s policy is the loosest of all partisanship regardless of economic conditions. As for monetary policy, all partisan governments use monetary policy countercyclically. During non-recessions, the left s policy is looser than the right or center. But as with fiscal policy, the center s and right s monetary policy is more countercyclical than the left. Thus, the center and the right are more countercyclical in both fiscal and monetary policies. This does not necessarily mean that the left does not carry out countercyclical policy to alleviate the adverse output and unemployment effects of recessions on its supporters (low-income or union workers), because the left can still run expansionary policy to target its supporters without creating budget deficits as long as it either increases tax revenues or cut expenditures on budget items that do not negatively affect its supporters. This needs to be probed in future research. 16

18 Electoral Cycles and Policy Mixes As shown in Table 5, fiscal policy gets loose in election years, and tight in non-election years. Electoral cycles in economic policy have been refuted in many studies, but our data indicate the existence of such cycles. It is confirmed also in regression reported below. The expansionary effect of elections on fiscal policy enter significantly all specifications we tried. -- Table 5 about here -- Monetary policy, by contrast, is not necessarily loose during election years (results not reported). If anything, it becomes tight in elections years (loose policy decreases, and tight policy also decreases). This tightening of monetary policy in election years is also confirmed in regression. It is reasonable to believe that monetary authorities counter electoral cycles in fiscal policy. Monetary policy is not sensitive to electoral conditions. This makes intuitive sense because central banks control monetary policy and do not have electoral incentives, unlike elected politicians. The frequency of policy mixes follow the same trend; in elections years, the loose fiscal-loose monetary mix increases, and tight-tight mix decreases. Multiparty Governments and Policy Mixes The conventional explanation suggests that multiparty governments fiscal policy is loose. Roubini and Sachs (1989a, 1989b) argue that multiparty governments characterized by a short tenure of office create higher budget deficits than do single-party governments (see also Grill et al., 1991). The logic is that multiple parties participate in a coalition government, have distinct electoral needs representing diverse interests, and try to protect the parts of the budget that are vital to them. Since a coalition government needs the support of its coalition parties, an easy solution for them is to cater to each party s budgetary needs, and fiscal restraint becomes hard to achieve. They also conjecture that the rapid turnover of multiparty governments reduces 17

19 the incentives for parties to cooperate on fiscal restraint, because rapid turnover reduces the time horizon for the repeated play among the parties. Similarly, Franzese (2002) shows that multiparty governments retard debt adjustments. -- Tables 6a and 6b about here -- As Table 6a shows, multiparty governments fiscal policy is not necessarily looser than single-party governments. 10 On the contrary, 2-, 3-, and 4-party governments run tighter policy than 1-party governments. As for monetary policy, 2- and 4-party governments monetary policy is also not looser than 1-party governments (Table 6b). As for the policy mix (results not reported), too, 3- and 5-party governments tend to have looser policy mixes than 1-party governments, but 2- and 4-party governments have tighter mixes than 1-party governments. A previous study (Bearce, 2002) claims that coalition governments have a loose fiscal-tight monetary mix, but the evidence here suggests that single-party governments engage in this loosetight mix more often than coalition governments. There is no clear pattern observable in the relationship between the number of governing parties and policy mixes. But one thing that is clear is that 2-party governments policy mix is tighter than 1-party governments, and the simple argument that coalition governments policy mix is loose fiscal-tight monetary receives little support. When we use the alternative 1.5% definition for loose and tight policy, 2-party governments fiscal policy is still tighter than 1-party ones, though 3-, 4-, and 5-party governments are now looser than 1-party ones. It means that coalition governments with more than 2 parties engage in larger fiscal expansion than 1-party governments (results not reported). In any event, the deficit-producing effect of coalition governments is not linear. This may be why the effect of the number of governing parties is not significant in regression analysis reported below

20 I also examined whether coalition governments policy is affected by central bank independence. The results (not reported) show that when central banks are dependent, coalition governments policy mixes may be somewhat looser than single-party governments. But when central banks are independent, coalition governments policy mixes turn tighter than single-party governments. Thus, independent central banks restrain coalition governments policy. 12 Regression Analysis To supplement the tabulation analysis above, I ran time-series cross-section regressions. The results of the regressions are generally consistent with the findings from the tabulation analysis. I ran many models to check the robustness of the effects of variables, including specifications with and without country fixed effects, but I only report the results of the basic specifications. Where different specifications create different results, they are explained. The dependent variables are, respectively, cyclically adjusted primary balance as a percentage of potential GDP (for fiscal policy models) and actual discount rates minus neutral interest rates suggested by the Taylor-type rule (for monetary policy models). The sample countries and years are the same as the above. The estimation method is ordinary least squares (OLS) with panel-corrected standard errors (PCSEs) and country dummy variables to correct for the estimation problems prevalent in panel data of this kind serial correlation, panel heteroskedasticity, and contemporaneous correlation of errors (Beck and Katz, 1995). Lagrange multiplier tests suggested that our models showed signs of autocorrelation, so lags of the dependent variable are entered in the equations as independent variables to eliminate autocorrelation (a first lag for fiscal policy models, and first to 7 th lags for monetary policy). For the variables GDP growth, inflation, and unemployment, their first differences are used to ensure stationarity. All independent variables except for the election year dummy are lagged one year to allow for the time lag between policy making and 19

21 implementation and to mitigate the endogeneity problem. The partisan variables are raw percentage scores of cabinet portfolios (0~100%). CBI is the raw continuous values of the CBI index (in the tabulation analysis above, we treated it a dummy variable divided by the means). The fiscal policy model (Table 7, Model 1) shows that independent central banks reduce deficits (increase surpluses), and that deficits increase in election years. These results are stable across different specifications, including models with and without country fixed effects. Independent central banks restrain fiscal policy, and fiscal policy becomes loose in election years. These results parallel the results of the tabulation analysis. The coefficient for the number of governing parties is not significant. The sign of the coefficient is unstable, turning negative in models with fixed effects and positive in models without them. 13 Thus, there is no empirical support for the argument that coalition governments fiscal policy is loose. -- Table 7 about here -- As for partisanship, the coefficients for the left, center, and right are similar and not significant. When we enter the interactive terms of the partisan variables and central bank independence in a model without fixed effects, we get some statistically significant results (Model 2). Though the coefficients for center are not significant, the results suggest that the center s fiscal policy is looser than the right and left in the absence of central bank independence, and that independent central banks restrain the center s fiscal policy more than the left s or right s. The left s fiscal policy is slightly tighter the right s. These results are consistent with the findings of the tabulation analysis. Here, coalition governments run tighter fiscal policy than single-party governments. But when country fixed effects are entered in the equation, these partisan and coalition variables lose significance. 20

22 The monetary policy model (Model 3) shows that as the number of governing parties increases, monetary policy becomes loose. But it achieves significance only in models with fixed effects. Monetary policy is tighter under center governments than the others, which must be a result of central banks response to the former s loose fiscal policy. But the coefficient for the center fails to achieve significance in most other specifications. Left and Right are not significant. Consistent with our tabulation analysis, monetary policy becomes tight in election years. The coefficient for election years is stable across different specifications. Tight monetary policy must be a result of loose fiscal policy pursued by party governments in election years. The sign of central bank independence is negative but is never statistically significant. Alternative specifications where partisan variables and central bank independence are interacted suggest that when central banks are dependent, monetary policy becomes loose under all partisan governments, but independent central banks make it tighter. But these coefficients are not significant. In the above regressions, I also entered the recession dummy to see how fiscal and monetary policy behaves across business cycles. Though I do not report the results, they show that monetary policy responds countercyclically to economic fluctuations (statistically significant) and fiscal policy is not affected by cycles. This is consistent with the results of the tabulation analysis that shows that monetary policy is used as a countercyclical tool. INTERPRETATION AND DISCUSSION Our empirical analysis yields several observations about governments use of fiscal and monetary policies. First, left governments use the tight fiscal-loose monetary policy mix more than the center or right. Second, independent central banks restrain fiscal policy by party governments and serve to reduce deficits. Though central bank independence certainly induces 21

23 monetary policy restraint, it tends to produce not so much tight as neutral monetary policy, as dependent central banks produce loose as well as tight monetary policy more often than independent ones. Thus, independent central banks provide monetary policy neutrality and stability. They do not have to run contractionary monetary policy as often because in their presence, party governments fiscal policy is restrained. They do counter center and right governments loose fiscal policy with tight monetary policy, and the center runs expansionary fiscal policy in the absence of independent central banks. Independent central banks restraining effect on fiscal and monetary policies is the strongest when governments are center-led. Monetary policy is tighter for center than left or right governments, because of the center s loose fiscal policy. When central banks are dependent, right and center governments use loose fiscalloose monetary or tight fiscal-tight monetary policy mixes most often of all mixes. Thus, dependent central banks accommodate the center s and the right s fiscal policy. 14 Third, economic policy mixes are sensitive to economic cycles and looser in recession years. Economic expansions during recessions are mainly crafted by loose monetary policy, but not fiscal policy, regardless of the independence of central banks. Monetary policy is more countercyclical than fiscal policy. In addition, both fiscal and monetary policies are more countercyclical under the center and right than the left. Fourth, fiscal policy becomes looser in election years, supporting electoral cycle theory. Economic expansions in election years are manufactured by loose fiscal policy, and accompanied by a tightening of monetary policy. Central banks counter electoral fiscal expansion by party governments. Fifth, coalition governments do not necessarily run loose macroeconomic policy more than single-party governments. Particularly, 2-party governments economic policy is more restrained than oneparty governments. Further, when central banks are independent, coalition governments policy 22

24 mixes turn tighter than single-party governments. Lastly, game-theoretic analyses by economists emphasize the conflictual nature of the policy game between party governments and central banks, and predict that the conflict produces the loose fiscal-tight monetary policy mix. But my results show that this particular policy mix does not occur often empirically. Overall, the patterns of the data suggest that fiscal and monetary authorities make respective policies with an eye toward the other. While we cannot tell whether it is coordination or just simple action and reaction, the data show that fiscal and monetary policies move relatively systematically in response to each other. Further, fiscal and monetary policies (and their mixes) depend on partisan governments and central banks. Unless governments are centerled and central banks are dependent, when fiscal policy gets loose, monetary policy does not, most of the time (as in election years). When monetary policy becomes loose, fiscal policy does not (as in recession years). With independent central banks, the left s fiscal policy is tight, but monetary policy gets somewhat loose. With dependent central banks, the left s fiscal policy becomes loose, but monetary policy gets tight. The right s monetary policy becomes loose with dependent central banks, but fiscal policy gets a bit tighter. Fiscal and monetary policies do move in opposite directions, especially when fiscal policy is expansionary. But the relationship does not look so much like the noncooperative policy conflict between fiscal and monetary authorities envisioned by economists as the policy adjustments conducted by central banks when fiscal policy becomes loose. In fact, the tightening of monetary policy takes place when fiscal policy is loose under center governments or during election years. Party governments fiscal policy also reacts to central banks, as when the former s fiscal policy becomes tight when the latter is independent. 23

25 We also observe that party governments take advantage of their control of monetary policy. When central banks are dependent, monetary policy is often made in a way to accommodate fiscal policy (loose fiscal-loose monetary mix or tight fiscal-tight monetary mix). This pattern is observed more in center and right governments, but is also present in the left. What explains left governments frequent use of a tight fiscal-loose monetary policy mix? There are grounds to believe that their policy preferences became more compatible with those of central banks in the 1980s and 1990s, and they together were able to fashion a policy mix conducive to economic performance. Left governments in the 1970s conducted expansionary monetary policy and Keynesian demand management policy. But when their economic measures were unsuccessful, they reversed Keynesian policy in the 1980s (Boix, 2000), and their economic policy shifted rightward (anti-inflationary). Concurrent with this were the globalization of capital and trade and the rise of the neoclassical policy orthodoxy among industrialized countries. Capital became increasingly mobile, and governments of all stripes came to feel the need to make their economic policy consistent with the goals of attracting capital and maximizing the competitiveness of their national economies. Governments latitude in choosing economic policy contracted, and they had little choice but to attach more importance to price stability and fiscal restraint, because capital interests are averse to inflation and because fiscal expansionism leads to large deficits (equally disliked by capital interests) which lead to high taxes and inflation, and to high interest rates and low savings which in turn depress investment, growth, and employment. Inflation also creates price distortions and resource allocation inefficiencies and increases economic uncertainty, further discouraging investment and impairing growth. As a result, industrial countries economic policy became increasingly anti-inflationary (Franzese, 1999) as well as fiscally conservative. 24

26 In order to attract mobile capital and promote international competitiveness, particularly left governments needed to show their commitment to price stability and to restrain their fiscal policy to build credibility to their anti-inflationary stance, because of their reputation for fiscal expansionism. As evidence of their efforts to increase credibility, many of the governments in which left parties participated granted independence to their central banks (Belgium, 1993; Finland, 1998; France, 1993; Ireland, 1998; Italy, 1981, 1992; New Zealand, 1990; Sweden, 1998; the United Kingdom, 1997), adopt fixed exchange rate regimes (Austria, 1996; Belgium, 1979; Denmark, 1979; Finland, 1996, Maastricht; France, 1991, Maastricht; Germany, 1979; Italy, 1991, Maastricht, 1996; the Netherlands, 1991, Maastricht; Norway, 1994; Sweden, 1996), and abandon monetary policy autonomy. 15 Left governments relatively conservative fiscal policy releases monetary policy for use as a countercyclical tool rather than as a tool to maintain price stability. Then, independent central banks do not have to run contractionary monetary policy which could otherwise produce recessionary pressure, and governments do not have to create high deficits that could lead to high interests, low investment, and low growth. Put differently, central banks can more comfortably use monetary policy to promote economic growth in exchange for party governments fiscal restraint. Further, left governments fiscal restraint gives them the fiscal maneuverability to carry out its supply-side economic policy (Boix, 1998). Next, the question remains why and how independent central banks restrain party governments fiscal policy in general. There are a couple of possibilities. First, central banks may restrain party governments fiscal policy in the same way they restrain labor s militant wage behavior like in Germany, where labor s anticipation that the Bundesbank would tighten monetary policy to counter inflationary wage increases deters labor from demanding high wage 25