Fiscal Politics
Chapter

Chapter 4. Economic and Political Determinants of Tax Policies in OECD Countries

Author(s):
Vitor Gaspar, Sanjeev Gupta, and Carlos Mulas-Granados
Published Date:
April 2017
Share
  • ShareShare
Show Summary Details
Author(s)
Mark Hallerberg and Jürgen von Hagen 

This chapter explores the economic and political causes of tax policy changes in Organisation for Economic Co-operation and Development (OECD) countries over the period 1990–2015. The analysis considers rate changes of the most important taxes in developed economies, that is, corporate income taxes (CIT), personal income taxes (PIT), and value-added taxes (VAT). The results show that banking crises affect tax rates, but that there is a political dimension to their effects—left-leaning governments are less likely to increase the VAT rate, and more likely to increase the top PIT rate, than right-leaning governments. There is also evidence that the left is more likely to raise the top income tax rate as part of pronounced fiscal consolidation. Perhaps surprisingly, the research finds no evidence of partisan effects during more “normal,” noncrisis times. The effects of elections are also generally muted, with only a CIT rate increase more likely the year after an election. Changes in other countries make it more likely that a given government changes its own rates. Increases in both the VAT rate and the top PIT rate are more likely when governments are consolidating their cyclically adjusted budget balances. Among European Union (EU) member states, VAT rate changes are more likely if a given government is in the euro area and in the period after 2011. Small EU member states also are more likely to raise PIT and CIT rates, which is consistent with an argument that pressure to consolidate under the Stability and Growth Pact was higher for small states during the period under study.

Introduction

Tax policy is one of the core activities of government. It determines how much revenue the government has to finance its activities. It also determines which groups in society contribute how much to this revenue, which means that it redistributes income and wealth. It can encourage or discourage certain economic activities and the choice of their geographical location and, thus, regulate and cause distortions to economic activities. As a result, tax policy is the object of the activities of political parties and interest groups. In democratic societies, it is conducted on the basis of economic considerations and political motivations.

This chapter explores the economic and political causes of changes in tax policies in OECD countries over the period 1990–2015. More specifically, it studies changes in tax rates, which are taken as a main indicator of a government’s desired changes in tax policies. The chapter explores the case that development of a given tax base over time is heavily influenced by economic activity and, therefore, by forces and factors other than tax policy. Thus, measures of tax revenues (or changes thereof) are noisier indicators of policy changes than are tax rates. The chapter considers rate changes to the most important taxes in developed economies—CIT, PIT, and VAT. Rates are the principal instruments of tax policy for several reasons. Tax legislation defines tax rates and tax bases, but changes in the definition of the latter are less common and are difficult to observe. Moreover, changes in tax rates usually lead to changes in revenues in the same direction.1 Furthermore, tax rates affect tax competition across jurisdictions. For example, Devereux and Griffith (1998) find that, while differences in tax rates did not affect American firms’ decisions to enter the European market, they did affect where firms located within Europe. Devereux, Lockwood, and Redoano (2008) argue that international competition in the dimension of corporate tax rates is intensified by the absence of capital controls.

During the period investigated in this chapter, several notable developments occurred that might have affected the likelihood of changes in tax policies. First, it was a period of high and increasing international capital mobility and thus one of intensifying international tax competition. Noticing its potentially harmful effects, the OECD launched a program combating harmful tax competition in 1998.2 A second development was the spread of democracy. The transition to democracy in Central Europe and in some Latin American countries in the sample, such as Chile and Mexico, introduced new voters, and potentially new interests, into the political arena, and these new players may have wanted changes in tax policies. As Gehlbach (2008) notes for Eastern Europe, this should spur more “normal” politics in countries where the ease of collection across different types of taxes (which is the case in OECD countries) is similar. Third, most OECD countries had high and rising levels of public debt during this period, while only a few managed to reduce their debt burdens significantly. As a result of the financial crisis that started in 2008, high debt burdens became a pressing issue, especially toward the end of the sample period. Fourth, the period contains several episodes of financial crises that put pressure on governments in opposite ways. On the one hand, governments wished to respond to incipient crises with fiscal stimulus packages that included tax cuts. On the other hand, governments had debt-sustainability concerns that forced them to consolidate their budgets and reduce their debt, and some of them introduced tax increases for that purpose. These contrasts were particularly prominent in Europe toward the end of the sample period, where the crisis was especially acute and where some countries received IMF- and EU-sponsored programs to stabilize their economies. Fifth, the countries belonging to the EU and in particular to the Economic and Monetary Union introduced new kinds of fiscal rules to keep government budget deficits and debt in check. This chapter explores the impacts of these developments on the tax policies of OECD countries. Furthermore, it asks how political factors such as elections and the ideological orientation of governments shaped these policies.

The chapter is structured as follows: The second section describes the developments of CIT, PIT, and VAT rates over the sample period, that is, the dependent variables. The third section develops the main hypotheses. The fourth section presents the empirical analysis and results. The analysis finds that banking crises affect tax rates, but that there is a political dimension to their effects—left-leaning governments are less likely to increase the VAT rate, and more likely to increase the top PIT rate, than right-leaning governments. Changes in other countries make it more likely that a given government changes its own rates. Increases in both the VAT rate and the top PIT rate are also more likely when governments are consolidating their cyclically adjusted budget balances. Among EU member states, VAT rate changes are more likely if a given government is in the euro area and in the period after 2011. The only political variable that affects the CIT rate is the occurrence of an election, with rate increases more likely the year after an election has been held. The final section concludes.

Describing Tax Policies in Oecd Countries

The main source of data for tax rates is the OECD tax database. In addition, information is taken from OECD Consumption Tax Trends and from the European Commission and national sources. Because of missing values in the data, an unbalanced panel of 34 OECD countries results.3

The chapter describes tax policies with an indicator variable equal to one for a given country and given year when an increase in the relevant tax rate occurs, zero when there is no change, and negative one when there is a decrease in the tax rate.

Corporate Income Taxes

Figure 4.1 describes the development of CIT rates in the OECD countries during the sample period. A first, important observation comes from Figure 4.1, panel 1, which shows that the unweighted average tax rate declined steadily during the sample period by a total of 15 percentage points, from 40 percent to 25 percent. Governments in OECD countries generally lifted the tax burden on corporate income by a very large margin.

Figure 4.1.Corporate Income Taxes in OECD Countries, 1991–2015

(Percent)
(Percent)

Sources: Authors’ calculations based on OECD Tax Revenue Statistics (http://www.oecd.org/ctp/tax-policy/tax-database.htm), June 2016; and OECD Consumption Tax Trends 1999, 2001, 2004, 2006.

Note: OECD = Organisation for Economic Co-operation and Development.

Panel 2 of Figure 4.1 looks at this average trend in more detail by showing the change in CIT rates between 1990 and 2000 and between 2001 and 2015. Austria, Belgium, and Canada were the only countries that increased rates in the first period; Hungary and Chile were the only countries that increased rates in the second period. On average, over all countries, the rate cuts were almost equal in both periods, but there are substantial differences between individual countries in both periods.

Panel 3 of Figure 4.1 shows the share of OECD countries that changed CIT rates in each year, that is, the number of countries with rate changes divided by the total number of countries for which data are available for that year. In 1991, 60 percent of all OECD countries implemented rate changes. Rate changes became rarer in the mid-1990s, but toward the end of the decade they became more common again, peaking at 50 percent of countries in 2000. After the mid-2000s, rate changes again became less common among the OECD countries.

Panel 4 of Figure 4.1 shows the balance of rate changes in the OECD, that is, the number of countries with rate increases minus the number of countries with rate decreases divided by the number of countries in the sample in that year. It indicates the dominant policy trend among the OECD countries in each year. This dominant trend was clearly negative in the early 1990s and from 1998 to 2008, suggesting that pressures on CIT rates from tax competition were strong for OECD governments during these years. Only in the wake of the financial crisis that started in late 2008 did this pressure weaken.

Panel 5 of Figure 4.1 shows the average frequency of CIT rate changes for each country during the sample period. This is the number of years in which a rate change occurred divided by the total number of years for which observations are available. The figure shows a large degree of variation in frequency across countries. Canada and the United States changed rates almost every year. In contrast, the frequency in Austria, New Zealand, and Norway was about 10 percent, indicating that tax rate changes occur only once in 10 years, on average. The average OECD frequency was 0.34, corresponding to a tax rate change every three years.

Personal Income Taxes

PITs are progressive in all OECD countries, implying that rates for different income brackets can change in different ways. This analysis represents PIT policy by changes in rates for the top bracket.

Panel 1 of Figure 4.2 shows that the average top PIT rate in the OECD declined from 43 percent to 35 percent during the sample period, with a slight increase after the financial crisis that started in 2008. Panel 2 of Figure 4.2 shows that cumulative PIT rate changes are more heterogeneous across countries than CIT rate changes. Mexico, New Zealand, Sweden, the United Kingdom, and the United States in the 1990s, and Japan and Portugal after 2001, saw noticeable increases in PIT rates, while all other countries had decreases in PIT rates. The share of countries with tax rate changes, shown in panel 3 of Figure 4.2, was generally lower than for CIT rates. Starting at 40 percent in 1990, the top rate oscillated around 20 percent for the rest of that decade. The early 2000s saw an increase in tax policy activism in the OECD. Following the financial crisis of 2008, the share of countries with tax rate changes hovered around 25 percent.

Figure 4.2.Personal Income Taxes in OECD Countries, 1990–2015

(Percent)
(Percent)

Sources: Authors’ calculations based on OECD Tax Revenue Statistics (http://www.oecd.org/ctp/tax-policy/tax-database.htm), June 2016; and OECD Consumption Tax Trends 1999, 2001, 2004, 2006.

Note: OECD = Organisation for Economic Co-operation and Development.

Panel 4 of Figure 4.2 shows the balance of rate changes in the OECD countries. It was positive only in two brief periods—from 1993 to 1995 and from 2009 to 2013, that is, immediately after the financial crisis of 2008. The development over time resembles the balance of CIT rate changes. Considering the frequency of PIT rate changes in individual countries (panel 5 of Figure 4.2), a large degree of heterogeneity again is evident. In contrast to CIT rate changes, Canada and the United States are among the countries with the lowest frequencies, while Denmark, Finland, and Norway have particularly high frequencies.

Value-Added Taxes

Panel 1 of Figure 4.3 shows that the average VAT rate increased in the OECD during the sample period from slightly higher than 16.5 percent to slightly higher than 19 percent. Overall, then, the burden of taxation shifted from incomes to consumption during this period. Because the elasticity of the tax base for corporate and personal income taxes with regard to tax rate changes is probably higher than the elasticity of consumption, this shift may reflect the pressures of international tax competition and the need of growing government sectors to secure a stable revenue base. Panel 2 of Figure 4.3 indicates that the Czech Republic, Ireland, the Netherlands, and the Slovak Republic in the first decade, and Canada, the Czech Republic, Iceland, and the Slovak Republic in the period 2001–15 were the only countries implementing VAT rate cuts. All other countries experienced rate hikes of various sizes. Estonia and Turkey had the largest rate hikes in the 1990s, while Greece and Portugal, which both experienced fiscal crises after 2010, had the largest cumulative rate hikes during 2001–15.

Figure 4.3.Value-Added Tax Rates in OECD Countries, 1990–2015

(Percent)
(Percent)

Sources: Authors’ calculations based on OECD Tax Revenue Statistics (http://www.oecd.org/ctp/tax-policy/tax-database.htm), June 2016; and OECT consumption Tax Trends 1999, 2001, 2004, 2006.

Note: OECD = Organisation for Economic Co-operation and Development.

Empirical Hypotheses and Operationalization

The previous section shows that tax rates and, hence, tax policies, vary substantially across the three types of taxes, across countries, and across time. This section develops some empirical hypotheses derived from standard economic as well as politico-economic considerations explaining the variation. It also indicates which explanatory variables are chosen for the empirical section that follows. The hypotheses are grouped into five categories. The first group is related to “normal” tax policies, linking them to economic developments in a conventional way. It sets the benchmark for the subsequent effects. The second group considers government responses to crisis situations, defined as the incidence of financial crises. The third group represents political factors of partisanship and electoral effects. The fourth group studies the effects of fiscal rules. The final group checks whether EU governments that now operate under the common framework of the Excessive Deficit Procedure (Maastricht Treaty), the Stability and Growth Pact (introduced in 1998), and the Fiscal Compact (introduced as a response to the European debt crisis that started in 2010) behave differently compared with other OECD governments.

Normal Tax Policies

Conventional public finance theory suggests that governments use tax policies for macroeconomic stabilization, cutting rates during recessions and raising rates during expansions. While some of this theory in practice is embedded in progressive tax schedules, one may expect that governments use discretionary changes in tax rates for taxes that are less progressive. Furthermore, unless tax schedules are indexed to the price level, tax policy may respond to increases in inflation to correct for “bracket creep” in progressive taxes. This could lead to pressures to reduce top PIT rates in particular. In this vein, Mahon (2004) finds that higher inflation in previous periods is correlated with tax rate reductions in Latin American countries during the period 1977–95. This gives the following hypotheses:

  • H1: Governments are more likely to cut tax rates following periods of low growth and more likely to raise tax rates following periods of high growth.4

  • Empirical operationalization: Lagged real GDP growth.

  • H2: Governments are more likely to cut PIT rates following years of high inflation. Empirical operationalization: Lagged inflation rates.

Debt sustainability is a second main concern of normal tax policies. To maintain the sustainability of public debt, primary budget surpluses must respond positively to rising public debt (for example, Bohn 1998). Although increases in tax rates may not be required because the adjustment could come through primary spending cuts, rising public debt makes increases in tax rates more likely, while declining public debt relieves pressures for tax rate hikes. This argument is captured empirically by including lagged ratios of government debt to GDP as explanatory variables.

This gives the following hypothesis:

  • H3: A low debt burden in the previous year makes tax rate cuts more likely, while the reverse is true for a high debt burden.

  • Empirical operationalization: Lagged government debt as percentage of GDP.

Episodes of persistent and large government deficits and rising public debt may force governments to implement large fiscal consolidations to maintain the sustainability of their debts. When this occurs, governments must decide on the appropriate mix between tax and expenditure adjustments. Based on OECD data from the early 1970s to the mid-1990s, Alesina and Perotti (1997); Perotti, Strauch, and von Hagen (1998); Hughes-Hallett, Strauch, and von Hagen (2002); and Alesina and Ardagna (2010) show that consolidations based on spending cuts are more likely to be successful in the sense that they lead to lasting reductions in deficits and debts.5 The suggested reason is that governments are likely to give in to political pressures for more spending as soon as larger revenues start flowing in following an increase in tax rates. Perotti, Strauch, and von Hagen (1998) show that successful fiscal consolidations are also associated with increases in revenues from taxes on labor rather than on business.6 Based on these considerations, one may expect that large and successful fiscal consolidations are associated with PIT rate increases but not with CIT rate increases. Finally, since VAT revenues in practice respond to rate changes faster than most other taxes, large and successful consolidations are expected to involve VAT rate increases.

Following this literature, the analysis identifies episodes of large and successful fiscal consolidations as periods during which the primary structural budget balance increased by at least 1.5 percent of GDP in two consecutive years. The data for the ratio of primary structural balances to GDP are taken from the OECD. The exercise uses a dummy variable for the existence of such episodes as an explanatory variable. Note that this approach implicitly assumes a two-stage decision-making process, with the decision to implement a large consolidation at the first stage and the choice of consolidation strategy and instruments at the second stage. The analysis takes the decision at the first stage as given and considers the tax policy aspects at the second stage. This gives the following hypothesis:

  • H4: Large, successful consolidations are associated with VAT and PIT rate increases.

  • Empirical operationalization: A dummy variable for episodes of large and successful fiscal consolidations.

Finally, the importance of tax competition among OECD countries is also considered. Tax competition occurs when the tax base is mobile across country borders and responds to international differences in tax rates. One may expect that tax competition is strongest in taxes falling on capital, less strong in taxes falling on labor, and least strong in taxes falling on consumption. The intensity of tax competition in the OECD is measured by including the balance of tax rate changes in the OECD from the previous section. The more countries that cut tax rates in a given year, the stronger the pressure on an individual government to do the same.7 At the same time, the literature has argued that large countries are less exposed to tax competition than small countries because size translates into market power in international capital markets and investors are more reluctant to leave large markets than small markets in response to increases in tax rates (for example, Haufler and Wooton 1999; Wilson 1999). The analysis includes the natural log of population to account for country size.

This gives the following hypotheses:

  • H5: A country is more likely to change its tax rate if other countries are changing their tax rates. A tax increase is more likely if others are raising taxes and a tax cut is more likely if others are cutting taxes.

  • Empirical operationalization: Balance of tax changes in the OECD in a given year.

  • H6: Small countries are more likely to cut rates.

  • Empirical operationalization: Natural log of population.

Tax Policy in Crisis Mode

This section turns to tax policies in crisis mode. We consider two types of crises. The first are banking crises, which represent important fiscal shocks, since, when a sizable part of the banking system fails, only the government has the ability to provide liquidity and to deal with insolvency issues. Whatever tools governments choose, the fiscal costs of the crisis are likely to be expensive and call for additional tax revenues (see Honohan and Klingebiel 2000; Amaglobeli and others 2015). Moreover, markets may be skeptical about the ability of governments to repay what they borrow unless they raise tax rates to show their determination to avoid a fiscal crisis coming out of the banking crisis. Furthermore, Wälti (2016) argues that government reforms to the financial sector are more likely when there is a sudden stop in international financing. Such a scenario may put pressure on government to raise funds domestically and to do so quickly. Alesina and Drazen (1991) argue that the necessary fiscal response to a crisis may be delayed by “wars of attrition” that arise when different groups in society are able to block the required policy changes and shift the burden of adjustment to other groups. As time passes, however, the burden of adjustment becomes larger and, as a result, the likelihood of tax rate changes increases. Consistent with this argument, Hallerberg and Scartascini (forthcoming) find that the governments of Latin American countries that experienced a banking crisis were more likely to increase tax rates, particularly of the VAT.

Following from this research, expectations are that banking crises in OECD countries make tax cuts less likely and tax increases more likely, which leads to the following hypothesis:

  • H7: Tax cuts are less likely, and tax increases are more likely, during banking crises.

  • Empirical operationalization: The standard IMF data set for banking crises, which is Laeven and Valencia (2013).8

The analysis also examines whether there was anything specific to the 2008–09 financial crisis. Both the European Commission and the G20 urged countries to engage in fiscal expansions if they had the fiscal room to do so in this period. Therefore, in the initial years of the crisis, the effect might be the opposite of what the analysis first posited, that is, there should be a fiscal expansion, which can be paid for through a tax increase.

  • H8: Tax cuts were more likely in the years 2008–09.

  • Empirical operationalization: Include a dummy variable for these years.

Countries facing fiscal crises may turn to the IMF for financial assistance, which, if granted, comes with conditions specified by IMF adjustment programs. The incidence of an IMF adjustment program is, therefore, another indicator for a fiscal crisis. Mahon (2004) finds that governments made more tax changes when under a program. The logic is that the IMF expects changes that particularly raise additional revenues. In a more general paper on the effects of IMF conditionality, however, Biglaiser and Rouen (2011) do not find any effects specifically on taxation. To identify the incidence of IMF adjustment programs, the analysis uses the data provided by Dreher (2006), as updated by that author through 2012 and by the current authors through the end of 2015.9 Based on those data, a dummy variable is defined that equals one when a country falls under an IMF program during a given year and zero otherwise. This gives the following hypothesis:

  • H9: Countries under IMF adjustment programs are more likely to increase tax rates.

  • Empirical operationalization: A dummy variable for the incidence of IMF adjustment programs.

Finally, tax responses to fiscal crises can be part of broader reforms that combine changes on the expenditure and revenue side of the budget with changes in other policy domains such as social insurance or labor market regulation. For example, tax rates may go up because a policy reform extends social insurance coverage to a group in society that was previously excluded. Alternatively, efforts may be made to reduce the distortions that taxation imposes on the economy. Tax rate changes, therefore, could occur in the context of changes in labor market regulation. The analysis uses an IMF data set indicating the incidence of public wage bill and employment reforms in a given country and year to identify periods of reform and define a dummy variable accordingly (IMF 2016).

This gives the following hypothesis:

  • H10: Tax changes are combined with reforms of the public wage bill and employment.

  • Empirical operationalization: A dummy variable for whether there was a reform in a given year.

Political Determinants of Tax Policy

How do politics affect tax policies? Political parties run on campaign planks that suggest they will change the tax burden for certain constituencies if they win the election. The standard analysis, which goes back to Hibbs (1977), is that the political right is most likely to favor “capital” while the political left is more likely to support “labor.” Looking at the tax burden in Europe, Osterloh and Debus (2012) find that left party control of government is associated with corporate tax increases in 32 European countries from 1980 to 2006; Angelopoulos, Economides, and Kammas (2012) similarly find that left-leaning governments rely more on taxation of capital than on labor taxation in a sample of 16 OECD countries over the period 1970–2000. Looking at revenues rather than tax rate changes, Stein and Caro (2013) find that the left raises overall tax levels as well as income tax levels. They find no partisan effects on VAT revenues. Building on this work and applying it to the data in this analysis, expectations are that the right cuts taxes on income and corporations and raises the VAT, which is regressive and spreads the tax burden to a greater part of the population than pays the top marginal income tax rate. Left-leaning governments are expected to want the reverse policies, that is, they are more likely to cut the VAT but increase the top rate of the PIT and the CIT.

This exercise uses the Database of Political Institutions (DPI) provided by Cruz, Keefer, and Scartascini (2016) to identify the political partisanship of governments. In this data set, governments are either right or left leaning. A government that has a right-leaning orientation is coded as one while left-leaning governments are coded as zero.

This yields the following hypothesis:

  • H11: Right-leaning governments are more likely to cut CIT and PIT rates and raise VAT rates, while left-leaning governments are more likely to do the reverse.

  • Empirical operationalization: Partisanship dummy for right-leaning governments.

Based on Latin American evidence, however, Capello (2014) argues that partisan differences dissolve in crisis settings. In such settings, governments have to react in any way they can, and there is simply no room or time to take partisan issues into account. This line of reasoning suggests the relevance of an interaction between crises and partisanship. For taxation, the inclusion of such a term could also be justified based on another argument: A crisis puts strong pressure on governments to raise taxes, and when that happens, they are more likely than under normal circumstances to raise taxes on their opponents. This suggests that right-leaning governments will increase the VAT rate and left-leaning governments will increase PIT and CIT rates during a banking crisis.

This yields the following hypothesis:

  • H12: In a financial crisis, right-leaning governments are more likely to increase the VAT rate while left-leaning governments are more likely to increase CIT and PIT rates.

  • Empirical operationalization: A term that interacts the banking crisis dummy with the partisanship dummy.

Political theory considers the effects of political fragmentation on the ability of governments to adjust tax and other policies. The more fragmented a political setting, the more difficult it is for the government to overcome the resistance of organized political interest groups to changes in tax policies. A good measure of political fragmentation is the number of “veto players” in the relevant decision-making process. A veto player is an actor in the policy process who can block changes to the tax policy status quo and whose assent is therefore required to implement such changes. For example, there are three institutional veto players in the United States when passing tax legislation—the two houses of Congress and the president. Tsebelis (2002) contends that the further apart are the most preferred policies of such players from each other, the more difficult it is for any one of them to achieve a change in the status quo. Thus, changes in tax policy are less likely if a U.S. president faces opposing majorities in one or both houses of Congress. Indeed, Basinger and Hallerberg (2004) find that the greater the ideological distance between veto players, the less likely is change in tax systems to occur.10

This analysis uses the variable “checks” from the DPI data set (Cruz, Keefer, and Scartascini 2016) to measure fragmentation. This variable counts the number of institutional and party veto players whose assent is required for change in policy, leading to the following hypothesis11:

  • H13: Countries with more institutional veto players have fewer changes in tax rates than countries with fewer institutional veto players.

  • Empirical operationalization: The variable “checks” from the DPI data set.

Another important political factor that could affect tax policies is the occurrence of elections. Political economy literature argues that governments are likely to cut tax rates and unlikely to increase them before elections, even if the arguments from normal tax policies would call for an increase. Several lines of reasoning support this conjecture. In a simple Keynesian framework of political business cycles, cutting tax rates is an attractive way to try to stimulate the macroeconomy, with many voters benefiting from increased economic growth and voting for the incumbent government (Nordhaus 1975; Hibbs 1977). In a rational-expectations version of political business cycles (Rogoff and Sibert 1988), cutting taxes is a signal to voters that the incumbent government has greater competence in economic policy than its opponents. In a distributional politics version of the argument, tax rate cuts are targeted at the constituencies of the ruling party; any growth effects are secondary. Von Hagen and Brückner (2002), Buti and van den Noord (2003), and von Hagen (2006) find electoral budget cycles in the euro area, especially in the early years of the introduction of the euro. At the same time, the political business cycle literature anticipates tax increases in the year after an election, which would correct for the increased deficits in an electoral year.

This yields the following hypothesis:

  • H14: Tax rate increases are less likely in years preceding elections and more likely in years following elections.

  • Empirical operationalization: A dummy for the year of the election and a lagged dummy variable for a legislative election year. Data from DPI.

Fiscal Rules

Fiscal rules are numerical restrictions on budgetary aggregates such as total government spending, revenues, budget balances, primary or structural balances, or on the ratio of public debt to GDP. Such rules aim to increase fiscal discipline and strengthen the credibility of the government’s commitment to stable fiscal policies.12 Rules restricting budget balances may, however, induce a bias toward higher taxes. On the one hand, they create pressures for tax rate hikes in times of low growth when government spending increases because of automatic stabilizers, such as unemployment insurance benefits or other types of welfare spending.13 On the other hand, budget balance rules can force governments to increase taxes if they face spending demands from their constituencies that they cannot resist and the additional tax burden falls on other social groups.

To identify countries and periods in the sample in which budget balance rules existed, the fiscal rules data set provided by the IMF is used to construct a dummy variable that is equal to one if a rule existed and zero otherwise.14

This leads to the following hypothesis:

  • H15: Countries with budget balance rules are more likely to increase tax rates than countries that do not have these rules.

  • Empirical operationalization: A dummy variable is included for countries that have budget balance rules in place.

European Union Effects

An extension of this framework examines EU member states. Although taxation is mostly a competence for the member states and not for the Union, minimum rates are in place for the VAT. Stage III of Economic and Monetary Union introduced a currency union, the euro area, in 11 member states in 1999. Today, the euro area has 19 members. All EU members are today subject to a common economic governance framework, which, however, is stricter for members of the euro area. Within this framework, the European Commission can recommend that a member state be declared as having an “excessive deficit” if the commission perceives that there are serious concerns about the sustainability of that state’s public finances. If the European Council of Ministers approves the European Commission’s recommendation, that member state is expected to adopt appropriate policies to reduce its deficit, and it can be reprimanded for failing to do so. Reforms of this framework, which were labeled “Six Pack” and “Two Pack,” were adopted in 2011–12 with the intention of strengthening this framework after its failure to avert public debt crises in Cyprus, Greece, Ireland, Italy, Portugal, and Spain. Therefore, there are pressures, especially on euro area members, to reduce budget deficits. This pressure may have affected their tax policies and made tax increases more likely. This yields the following hypotheses:

  • H16: Euro area member states are more likely to raise tax rates.

  • Empirical operationalization: A dummy variable for states in the euro area.

  • H17: EU member states under an “excessive deficit procedure” are more likely to raise tax rates.15

  • Empirical operationalization: A dummy variable for states under an excessive deficit procedure.

  • H18: After 2011, EU rules on debts and deficits resulted in increased pressures to raise tax rates.

  • Empirical operationalization: A dummy variable for the post-2011 period for EU states.

There is an additional wrinkle to the analysis. Baerg and Hallerberg (2016) find that the Council of Ministers is more likely to keep the original text and recommendations proposed by the European Commission when it reprimands the fiscal policies of small EU member states. In contrast, large member states, like France and Germany, frequently seem to be able to achieve a weakening of the recommendations that the European Commission proposed. This difference suggests that small states face a more powerful constraint on fiscal policy from the EU than large states. Population is already included as a measure of size in the analysis under hypothesis H6 (tax competition). In the current context, size can be interpreted as a proxy for the political power of the member states under Economic and Monetary Union. In this case, however, the prediction for EU member states with differential application of the Stability and Growth Pact would be the opposite—large states would not face the same pressure to consolidate as small states, so they would cut taxes more and increase them less than small states.

Empirical Analysis and Results

To test empirically the hypotheses stated above, a binary time-series cross-section ordered logit model is estimated. The dependent variable of this model can take one of three values in each period: 1 for an increase in the tax rate, 0 for no change in the tax rate, and -1 for a cut in the tax rate. Intuitively, the model estimates the likelihood of the occurrence of each of these events given the values of the explanatory variables. The model thus explains the direction but not the intensity of the tax policies in the sample.16 That is, it tells us whether a government is likely to move in a certain direction conditional on the realization of the explanatory variables, but it does not tell us how far it will move in that direction—how much it will raise or cut a tax rate. This limitation is due to the nature of the data.

Models of this kind are estimated for the three types of taxes under consideration, PIT, CIT, and VAT. Apart from the explanatory variables developed in the previous section, the exercise also includes variables indicating the number of previous changes in the tax rate during the sample period as well as the time since the last change. This approach is based on the consideration that governments may want to avoid frequent tax changes. The time period covered is 1990–2015. The overall sample is all OECD countries. There are, however, restrictions on data sources for some countries, so that an unbalanced panel results. Nevertheless, there are at least 24 and up to 34 countries in any given year in the analysis. Table 4.1 provides a summary of the findings, and the statistical results that include marginal effects and standard errors for decreases and increases for the VAT, PIT, and CIT appear in Annex Table 4.1.1 for the full sample and Annex Table 4.1.2 for the EU sample.17

Table 4.1.Summary of Findings
HypothesisEvidence
H1: Governments cut tax rates following low growth; increase rates after high growth.Opposite for VAT: more likely to cut rate with high growth and increase with low growth
H2: Governments are more likely to cut PIT rates following years of high inflation.No support
H3: A low debt burden makes tax rate cuts more likely; high debt burden makes tax rate cuts less likely.No support
H4: Large, successful consolidations are associated with VAT and PIT rate increases; large fiscal expansions are associated with PIT rate cuts.Large, successful consolidations associated with VAT and PIT rate increases
H5: A country changes rates in the same direction as changes in other countries.Supportive evidence, all types of taxes
H6: Small countries are more likely to cut rates.Opposite for CIT: large countries more likely to cut rate. Also, for the EU sample, large states are also more likely to decrease top PIT rate.
H7: Tax cuts are less likely, and tax increases are more likely during banking crises.VAT increase less likely, PIT increase more likely
H8: Tax cuts are more likely during 2008–09.No support
H9: Countries under IMF programs increase tax rates.No support
H10: Reforms in employment lead to tax changes.VAT increase more likely
H11: Right-leaning governments cut CIT and PIT rates and raise VAT rates, whereas left-leaning governments do the reverse.No support
H12: In a financial crisis, right-leaning governments raise VAT, and left-leaning governments raise CIT and PIT rates.Left less likely to raise VAT, more likely to raise PIT in a crisis
H13: The more institutional veto players there are, the fewer changes in tax rates.No support
H14: Tax rate increases are less likely in the years before elections and more likely after elections.CIT rate increases more likely after an election, cuts less likely
H15: Balanced budget rules are more likely to increase tax rates.No support
H16: Euro area states raise tax rates.VAT rate increases more likely in euro area states
H17: Excessive deficit procedure raises rates.No support
H18: There were tax increases after 2011 in EU states.PIT and CIT cuts less likely; PIT and CIT increases more likely
Source: OECD Tax Revenue Statistics (http://www.oecd.org/ctp/tax-policy/tax-database.htm), June 2016; and OECD Consumption Tax Trends 1999, 2001, 2004, 2006.Note: CIT = corporate income tax; EU = European Union; OECD = Organisation for Economic Co-operation and Development; PIT = personal income tax; and VAT = value-added tax.
Source: OECD Tax Revenue Statistics (http://www.oecd.org/ctp/tax-policy/tax-database.htm), June 2016; and OECD Consumption Tax Trends 1999, 2001, 2004, 2006.Note: CIT = corporate income tax; EU = European Union; OECD = Organisation for Economic Co-operation and Development; PIT = personal income tax; and VAT = value-added tax.

Normal Tax Policy

Changes in GDP affect only the VAT rate, with increases in economic growth making a VAT rate cut somewhat more likely (about a 0.3 percentage point increase in the probability for every 1.0 percentage point increase in the growth rate) and a VAT increase less likely (about a 0.8 percentage point decrease in the probability for every 1.0 percentage point increase in the growth rate). Other economic variables, such as the lagged inflation rate or the lagged overall debt burden, do not make it more likely that the three taxes are changed.

There is some clear evidence of tax competition, that is, that the balance of tax rate changes in the OECD affects tax policy in individual countries. This variable is highly significant for all three types of taxes and has the strongest effects for PIT and CIT rates. For example, if half of the OECD countries are increasing their top income tax rate, it is 13.3 percentage points more likely that a country will increase its own rate. If half of the OECD countries cut their top PIT rate, it is 31.0 percentage points more likely that a country will cut its rate. Population size effects only apply for the CIT. Countries with large populations are somewhat more likely to reduce their CIT rates in this period and somewhat less likely to increase this rate.

Tax Policy in Crisis Mode

As a reminder, the main specification includes an interaction term, which means that the marginal effect associated with “banking crisis” is the effect when “Executive Right” is coded zero, that is, when a center or left-leaning government is in power during a banking crisis. The analysis indeed identifies partisan effects during a crisis—a left-leaning or centrist government is about 8.0 percentage points less likely to increase the VAT than a right-leaning government at the p < 0.05 level of significance. A left-leaning government is about 2.0 percentage points more likely to cut the VAT in a banking crisis, and about 7.0 percentage points more likely to increase the top PIT rate (albeit both at p < 0.1 level of significance only). This finding suggests that partisanship affects tax responses in ways anticipated earlier in the chapter for these two types of taxes during banking crises; there are no significant results for changes in CIT rates.

Turning to large fiscal consolidations, large and successful consolidations are found to be associated with tax changes. Increases in both the VAT rate and the top PIT rate are more likely during large fiscal contractions, meaning that it is about 10 percentage points more likely that there will be a VAT rate increase and about 11 percentage points more likely there will be a top PIT rate increase. This finding means that the consolidations in the data set are connected with tax increases.18 Another type of “reform” is also relevant, namely labor market reforms. Reforms of public wage bills and employment, which occur often (though not always) during or after crises, are associated with a 9 percentage point increase in the likelihood of a VAT rate increase.

In contrast, the remaining variables do not seem to have an effect on changes in rates. Being under an IMF program does not spur further changes in these types of taxes. Counter to expectations, the crisis years of 2008 and 2009 also do not have a tangible effect. This outcome suggests that other relevant variables have been identified, such as a banking crisis year and the behavior of other states in the sample, meaning that a simple time dummy is no longer statistically significant.

Political Determinants of Tax Policy

Beyond the clear partisan effect during a crisis, differences between the left and the right in noncrisis years are not found. The results reported in Annex Table 4.1.3 further explore whether partisanship affects the type of tax changes used during a pronounced fiscal consolidation. As found for crisis situations, the left is more likely to raise the top PIT rate and less likely to cut it. Legislative elections in a previous year seem to have an effect only for the CIT, and in the expected direction—governments are about 2.0 percentage points more likely to increase this tax, and about 0.6 percentage point less likely to cut it, after an election. These results, which are replicated for all further specifications in unreported work, also investigate whether there are tax changes the year of an election, and show no evidence that cycles result in changes in tax rates.

The same is the case for the institutional “checks” variable—country-years that have institutions that increase the transaction costs for passing legislation (veto players) do not seem to reduce the chances of changes in tax rates except for one case—more institutional checks make it somewhat more likely that there is a CIT rate cut, and somewhat less likely that there is a CIT rate increase.19

Fiscal Rules

The balanced budget amendment does not seem to spur tax rate changes. The substantive effects in all specifications are close to zero, and in none of them is it statistically significant.

European Union Effects

Annex Table 4.1.2 reports results on specific EU effects. Only EU states are in this sample. Being a member of the euro area affects the VAT especially, making it more likely that a member state increases it and less likely that it decreases it. The effects of banking crisis conditional on partisanship are also more pronounced—the left is 13 percentage points less likely than the right to increase the VAT rate during a banking crisis. At the same time, the political right is significantly more likely to cut the top PIT rate during a banking crisis. Member states under IMF programs are also more likely to reduce the top PIT rate. There are also clear effects from 2012, that is, the period when the “Two Pack” and “Six Pack” economic governance reforms were in place and where notable problems with debt burdens remained. Both PIT and CIT cuts are less likely and PIT and CIT increases are more likely. Curiously, however, over the period, being identified as having an “excessive deficit” does not make tax changes any more likely. As expected, there is a difference between large and small states, with larger states more likely to cut, and less likely to increase, PIT and CIT rates. This finding is consistent with an argument that small states faced more pressure than large states to consolidate in the EU under the Stability and Growth Pact.

Conclusions

This chapter reviews changes over the past 25 years in the three main taxes governments use in OECD countries: CIT, PIT, and VAT. Variation has occurred both across countries and across time.

The analysis identifies political reasons that explain some of this variation. Left-leaning governments are more likely to cut VAT rates and to increase PIT rates during a financial crisis. This propensity means that the costs of adjustment are more likely to be borne by the political opponents of the parties in office. Similarly, electoral cycles clearly have an effect, although they appear only for the CIT—rates are more likely to be increased and less likely to be cut in the year following an election.

There are also relevant associations with significant fiscal adjustments. Governments that successfully increase the cyclical budget balance more than 1.5 percent of GDP in two consecutive years also increase both the VAT rate and the top PIT rate. This suggests that the type of tax used in the consolidation, and not just the divide between expenditure- and revenue-driven consolidations, may be a key policy instrument. Looking further at partisan effects, the left is more likely to use increases in PIT rates as part of a fiscal consolidation.

Many of the OECD countries are also in the EU, and the EU’s rules have some effect on tax policy as well. Euro area member states throughout the period are more likely to raise their VAT rates. In the period after 2011, which is when the overall fiscal framework was strengthened, it became more likely that there would be tax increases in the other two types of taxes. It is noteworthy that the lagged change in GDP is not included in the model. This suggests additional pressure for tax changes during this period.

The limitations of this study must be acknowledged. The analysis measures only whether a given tax rate was increased or decreased and not the scale of the change. Other types of changes to tax systems also affect the base of a given tax. Future work could explore whether the crisis, partisan, and consolidation effects identified here also affect the scale of changes in rates and changes in other dimensions of tax policy.

Annex 4.1. Supplementary Tables
Annex Table 4.1.1.Ordered Logit, Marginal Effects
VATPITCIT
(1)(2)(3)(4)(5)(6)
VariablesDecreaseIncreaseDecreaseIncreaseDecreaseIncrease
Change in GDP (lag)0.002*−0.008**−0.0070.0030.005−0.002
(0.001)(0.003)(0.006)(0.003)(0.008)(0.003)
Inflation (lag)−0.0020.0060.009−0.0040.016−0.005
(0.002)(0.006)(0.006)(0.003)(0.010)(0.003)
Gross Debt (lag)−0.0000.000−0.0000.0000.000−0.000
(0.000)(0.000)(0.001)(0.000)(0.000)(0.000)
Population (natural lag)0.002−0.008−0.0070.0030.019*−0.006*
(0.003)(0.009)(0.019)(0.008)(0.011)(0.004)
Balance of Tax Changes−0.161***0.506***−0.621***0.267***−0.821***0.273***
(0.059)(0.135)(0.126)(0.083)(0.153)(0.096)
Banking Crisis0.029**−0.093***−0.1560.067*0.094−0.031
(0.014)(0.034)(0.103)(0.038)(0.077)(0.026)
2008 or 20090.010−0.0300.008−0.003−0.0540.018
(0.014)(0.045)(0.056)(0.024)(0.058)(0.019)
Fiscal Contraction—Cyclical Balance > 1.5 × GDP−0.030**0.095***−0.259***0.111***−0.1140.038
(0.013)(0.032)(0.082)(0.042)(0.132)(0.046)
IMF Program−0.0290.090**0.051−0.0220.083−0.028
(0.018)(0.042)(0.070)(0.029)(0.077)(0.025)
Reforms Public Wage, Employment
−0.0010.003−0.0250.0110.007−0.002
(0.008)(0.025)(0.043)(0.018)(0.047)(0.015)
Executive Politically Right−0.0160.0520.162*−0.070*−0.1020.034
(0.015)(0.043)(0.095)(0.036)(0.095)(0.032)
Right × Banking Crisis0.000−0.0010.012−0.0050.029*−0.010*
(0.002)(0.007)(0.016)(0.007)(0.015)(0.006)
Checks and Balances0.003−0.010−0.0060.0030.013−0.004
(0.005)(0.017)(0.025)(0.010)(0.033)(0.011)
Legislative Election0.012*−0.039*−0.0310.013−0.055*0.018**
(0.007)(0.022)(0.026)(0.011)(0.030)(0.009)
Legislative Election (lag)−0.0020.005−0.0100.004−0.0280.009
(0.009)(0.029)(0.043)(0.018)(0.039)(0.012)
Budget Balance Rule−0.0000.000−0.025***0.011***−0.029***0.010***
(0.003)(0.008)(0.005)(0.002)(0.006)(0.002)
Number of Previous Tax Changes
−0.002***0.006***0.006−0.002**0.016***−0.005***
(0.001)(0.002)(0.004)(0.001)(0.004)(0.002)
Time Since Last Tax Change0.002*−0.008**−0.0070.0030.005−0.002
(0.001)(0.003)(0.006)(0.003)(0.008)(0.003)
Observations520520576576571571
Source: See the main text under the Hypotheses section.Note: Coefficients displayed are the marginal effects. Standard errors are in parentheses. CIT = corporate income tax; PIT = personal income tax; and VAT = value-added tax.***p < 0.01, **p < 0.05, *p < 0.1.
Source: See the main text under the Hypotheses section.Note: Coefficients displayed are the marginal effects. Standard errors are in parentheses. CIT = corporate income tax; PIT = personal income tax; and VAT = value-added tax.***p < 0.01, **p < 0.05, *p < 0.1.
Annex Table 4.1.2.European Union Member States
VATPITCIT
(1)(2)(3)(4)(5)(6)
VariablesDecreaseIncreaseDecreaseIncreaseDecreaseIncrease
Euro Area Member State−0.022**0.059**−0.0050.002−0.0050.001
(0.010)(0.024)(0.050)(0.019)(0.063)(0.011)
Excessive Deficit Procedure
−0.0060.015−0.0260.010−0.0740.013
(0.016)(0.044)(0.045)(0.017)(0.062)(0.012)
EU Country after 2011−0.0240.064**−0.0810.031−0.1290.023
(0.016)(0.028)(0.091)(0.034)(0.093)(0.018)
Change in GDP (lag)0.002−0.006**0.002−0.0010.003−0.000
(0.001)(0.003)(0.008)(0.003)(0.010)(0.002)
Inflation (lag)−0.007**0.018***0.010−0.0040.021*−0.004*
(0.003)(0.006)(0.010)(0.004)(0.012)(0.002)
Gross Debt (lag)0.000−0.0010.000−0.0000.002−0.000
(0.000)(0.000)(0.001)(0.000)(0.001)(0.000)
Population (natural lag)0.000−0.0000.050**−0.019**0.060**−0.011**
(0.004)(0.011)(0.021)(0.008)(0.024)(0.004)
Balance of Tax Changes−0.208**0.556***−0.679***0.257**−0.961***0.170**
(0.092)(0.144)(0.205)(0.106)(0.235)(0.072)
Banking Crisis0.052**−0.139***−0.1020.0380.154−0.027
(0.027)(0.043)(0.101)(0.043)(0.117)(0.022)
2008 or 2009−0.0100.027−0.0430.016−0.1630.029
(0.028)(0.071)(0.070)(0.025)(0.113)(0.024)
Fiscal Contraction—Cyclical Balance > 1.5 × GDP
−0.044*0.117***−0.367***0.139***−0.1170.021
(0.024)(0.043)(0.085)(0.035)(0.140)(0.025)
IMF Program−0.0330.0890.257***−0.097***−0.0010.000
(0.038)(0.086)(0.059)(0.035)(0.175)(0.031)
Reforms Public Wage, Employment
−0.0340.092**0.012−0.0050.182**−0.032*
(0.026)(0.047)(0.097)(0.036)(0.083)(0.017)
Executive Politically Right−0.0050.014−0.135**0.051*−0.0000.000
(0.011)(0.029)(0.062)(0.028)(0.055)(0.010)
Right × Banking Crisis−0.0250.068*0.255**−0.096*−0.1760.031
(0.018)(0.040)(0.100)(0.051)(0.129)(0.025)
Checks and Balances0.005−0.014**0.027−0.0100.072***−0.013***
(0.003)(0.006)(0.019)(0.008)(0.024)(0.004)
Legislative Election−0.0030.007−0.0190.007−0.0110.002
(0.008)(0.021)(0.026)(0.010)(0.048)(0.009)
Legislative Election (lag)0.004−0.011−0.0470.018−0.145***0.026***
(0.008)(0.023)(0.044)(0.017)(0.047)(0.010)
Budget Balance Rule−0.0050.013−0.0000.0000.069−0.012
(0.014)(0.038)(0.079)(0.030)(0.086)(0.015)
Number of Previous Tax Changes
−0.0010.004−0.043***0.016***−0.052***0.009***
(0.004)(0.009)(0.006)(0.003)(0.011)(0.003)
Time Since Last Tax Change−0.002*0.006***0.014***−0.005***0.033***−0.006***
(0.001)(0.002)(0.003)(0.002)(0.007)(0.002)
Observations300300291291299299
Source: See the main text under the Hypotheses section.Note: The coefficients displayed are the marginal effects. Standard errors are in parentheses. CIT = corporate income tax; PIT = personal income tax; and VAT = value-added tax.*p < .1; **p < .05; ***p < .01.
Source: See the main text under the Hypotheses section.Note: The coefficients displayed are the marginal effects. Standard errors are in parentheses. CIT = corporate income tax; PIT = personal income tax; and VAT = value-added tax.*p < .1; **p < .05; ***p < .01.
Annex Table 4.1.3.Ordered Logit, Additional Interaction
VATPITCIT
(1)(2)(3)(4)(5)(6)
VariablesDecreaseIncreaseDecreaseIncreaseDecreaseIncrease
Executive Politically Right−0.0040.0120.003−0.001−0.0340.011
(0.006)(0.020)(0.033)(0.014)(0.035)(0.013)
Fiscal Contraction—Cyclical Balance > 1.5 × GDP
−0.0230.072−0.308***0.132***−0.2600.086
(0.015)(0.046)(0.084)(0.044)(0.166)(0.058)
Executive Right × Fiscal Contraction
−0.0210.0670.156−0.0670.464**−0.153**
(0.029)(0.085)(0.232)(0.101)(0.204)(0.072)
Change in GDP (lag)0.002*−0.007**−0.0060.0020.003−0.001
(0.001)(0.003)(0.007)(0.003)(0.008)(0.003)
Inflation (lag)−0.0020.0070.010*−0.0040.014−0.005
(0.002)(0.006)(0.006)(0.003)(0.010)(0.003)
Gross Debt (lag)−0.0000.000−0.0000.0000.000−0.000
(0.000)(0.000)(0.001)(0.000)(0.001)(0.000)
Population (natural lag)0.002−0.006−0.0050.0020.019*−0.006*
(0.003)(0.009)(0.019)(0.008)(0.010)(0.004)
Balance of Tax Changes−0.163***0.513***−0.621***0.266***−0.813***0.268***
(0.061)(0.134)(0.127)(0.084)(0.148)(0.093)
Banking Crisis0.021**−0.066**−0.0720.0310.036−0.012
(0.010)(0.028)(0.061)(0.024)(0.052)(0.017)
2008 or 20090.010−0.0320.004−0.002−0.0470.016
(0.014)(0.045)(0.057)(0.025)(0.059)(0.019)
IMF Program−0.0030.0090.098−0.042−0.0360.012
(0.016)(0.050)(0.129)(0.055)(0.110)(0.036)
Reforms Public Wage, Employment
−0.0280.089**0.041−0.0180.114−0.038
(0.019)(0.044)(0.065)(0.027)(0.079)(0.025)
Checks and Balances−0.0000.0000.014−0.0060.028*−0.009*
(0.002)(0.007)(0.016)(0.008)(0.015)(0.006)
Legislative Election0.004−0.012−0.0120.0050.011−0.004
(0.005)(0.017)(0.025)(0.010)(0.031)(0.010)
Legislative Election (lag)0.012*−0.038*−0.0320.014−0.057**0.019**
(0.006)(0.021)(0.027)(0.012)(0.029)(0.009)
Budget Balance Rule−0.0010.003−0.0080.003−0.0300.010
(0.009)(0.029)(0.043)(0.018)(0.038)(0.012)
Number of Previous Tax Changes
−0.0000.001−0.026***0.011***−0.028***0.009***
(0.003)(0.008)(0.005)(0.002)(0.007)(0.002)
Time Since Last Tax Change−0.002***0.006***0.006*−0.002**0.016***−0.005***
(0.001)(0.002)(0.003)(0.001)(0.004)(0.002)
Observations520520576576571571
Source: See the main text under the Hypotheses section.Note: This table adds an interactive term for partisanship and fiscal contractions. It also reports the results when the year of the election is included. The coefficients displayed are the marginal effects. Standard errors are in parentheses. CIT = corporate income tax; PIT = personal income tax; and VAT = value-added tax.*p < .1; **p < .05; ***p < .01.
Source: See the main text under the Hypotheses section.Note: This table adds an interactive term for partisanship and fiscal contractions. It also reports the results when the year of the election is included. The coefficients displayed are the marginal effects. Standard errors are in parentheses. CIT = corporate income tax; PIT = personal income tax; and VAT = value-added tax.*p < .1; **p < .05; ***p < .01.
References

    AlesinaA. and S.Ardagna.2010. “Large Changes in Fiscal Policy. Taxes Versus Spending.” In Tax Policy and the Economy Volume 24 edited by Jeffrey R.Brown3568. Chicago: University of Chicago Press.

    AlesinaA. and A.Drazen.1991. “Why Are Stabilizations Delayed?American Economic Review81 (5): 117088.

    AlesinaA.C.Favero and F.Giavazzi.2015. “The Output Effect of Fiscal Consolidation Plans.Journal of International Economics96 (Suppl 1): S19S42.

    AlesinaA. and R.Perotti.1997. “Fiscal Adjustments in OECD Countries: Composition and Macroeconomic Effects.IMF Staff Papers44 (2): 21048.

    AmaglobeliD.N.EndM.Jarmuzek and G.Palomba.2015. “From Systemic Banking Crises to Fiscal Costs: Risk Factors.Working Paper 15/166International Monetary FundWashington.

    AngelopoulosK.G.Economides and P.Kammas.2012. “Does Cabinet Ideology Matter for the Structure of Tax Policies?European Journal of Political Economy28 (4): 62035.

    Avi-YonahR. S.2009. “The OECD Harmful Tax Competition Report: A Tenth Anniversary Retrospective.Brookings Journal of International Law34 (3): 78395.

    BaergN. and M.Hallerberg.2016. “Explaining Instability in the Stability and Growth Pact: The Contribution of Member State Power and Euroskepticism to the Euro Crisis.Comparative Political Studies49 (7): 9681009.

    BasingerS. and M.Hallerberg.2004. “Remodeling the Competition for Capital: How Domestic Politics Erases the Race-to-the-Bottom.American Political Science Review98 (2): 26176.

    BiglaiserG. and K.DeRouen.2011. “How Soon Is Now? The Effects of the IMF on Economic Reforms in Latin America.Review of International Organizations6 (2): 189213.

    BohnH.1998. “The Behavior of U.S. Public Debt and Deficits.Quarterly Journal of Economics113 (3): 94963.

    ButiM. and P.van den Noord.2003. “Discretionary Fiscal Policy and Elections: The Experience of the Early Years of EMU.Working Paper 351Economics Department, Organisation for Economic Co-operation and DevelopmentParis.

    CalvoE.2014. Legislator Success in Fragmented Congresses in Argentina: Plurality Cartels Minority Presidents and Lawmaking.Cambridge, U.K.: Cambridge University Press.

    CapelloD.2014. “The Politics of Financial Boom and Crisis: Evidence from Latin America.Comparative Political Studies47 (2): 26086.

    CruzC.P.Keefer and C.Scartascini.2016. “Database of Political Institutions Codebook, 2015 Update (DPI 2015).Inter-American Development Bank. Updated version of T.BeckG.ClarkeA.GroffP.Keefer and P.Walsh.2001. “New Tools in Comparative Political Economy: The Database of Political Institutions.World Bank Economic Review15 (1): 16576.

    DevereuxM. and R.Griffith.1998. “Taxes and the Location of Production: Evidence from a Panel of US Multinationals.Journal of Public Economics68 (June): 33567.

    DevereuxM. P.B.Lockwood and M.Redoano.2008. “Do Countries Compete over Corporate Tax Rates?Journal of Public Economics92 (5–6): 121035.

    DreherA.2006. “IMF and Economic Growth: The Effects of Programs, Loans, and Compliance with Conditionality.World Development34 (5): 76588.

    GehlbachS.2008. Representation through Taxation.Cambridge, U.K.: Cambridge University Press.

    GehlbachS. and E. J.Malesky.2010. “The Contribution of Veto Players to Economic Reforms.Journal of Politics72 (4): 95775.

    HallerbergM. and C.Scartascini. Forthcoming. “Explaining Changes in Tax Burdens in Latin America: Do Politics Trump Economics?European Journal of Political Economy. doi:10.1016/j.ejpoleco.2016.07.004.

    HauflerA. and I.Wooton.1999. “Country Size and Tax Competition for Foreign Direct Investment.Journal of Public Economics71 (1): 12139.

    HibbsD.1977. “Political Parties and Macroeconomic Policy.American Political Science Review71 (4): 146787.

    HonohanP. and D.Klingebiel.2000. “Controlling the Fiscal Costs of Banking Crises.Working Paper 2441World BankWashington.

    Hughes-HallettA.R. R.Strauch and J.von Hagen.2002. “Budgetary Consolidation in Europe: Quality, Economic Conditions, and Persistence.Journal of the Japanese and International Economies16 (4): 51235.

    International Monetary Fund. 2016. “Managing Government Compensation and Employment—Institutions, Policies, and Reform Challenges.International Monetary FundWashington, DC.

    KopitsG. and S. A.Symansky.1998. “Fiscal Policy Rules.Occasional Paper 162International Monetary FundWashington.

    LaevenL. and F.Valencia.2013. “Systemic Banking Crises Database.IMF Economic Review61 (2): 22570.

    MahonJ. E.2004. “Causes of Tax Reform in Latin America, 1977–95.Latin American Research Review39 (1): 330.

    NordhausW. D.1975. “The Political Business Cycle.Review of Economic Studies42 (2): 16990.

    Organisation for Economic Co-operation and Development (OECD). 1998. Harmful Tax Competition: An Emerging Global Issue.Paris: OECD.

    OsterlohS. and M.Debus.2012. “Partisan Politics in Corporate Taxation.European Journal of Political Economy28 (2): 192207.

    PerottiR.R. R.Strauch and J.von Hagen.1998. Sustainable Public Finances.London: Centre for Economic Policy Research.

    PoterbaJ. M.1995. “Balanced Budget Rules and Fiscal Policy: Evidence from the States.National Tax Journal48 (3): 32936.

    RogoffK. and A.Sibert.1988. “Elections and Macroeconomic Policy Cycles.Review of Economic Studies55 (1): 116.

    SchaechterA.T.KindaN.Budina and A.Weber.2012. “Fiscal Rules in Response to the Crisis. Towards the Next-Generation Rules. A New Dataset.Working Paper 12/187International Monetary FundWashington.

    SteinE. and L.Caro.2013. “Ideology and Taxation in Latin America.Working Paper 407Inter-American Development BankWashington, DC.

    TsebelisG.2002. Veto Players: How Political Institutions Work.Princeton, New Jersey: Princeton University Press.

    von HagenJ.2006. “Fiscal Rules and Fiscal Performance in the EU and Japan.Monetary and Economic Studies24 (1): 2560.

    von HagenJ. and M.Brückner.2002. “Monetary and Fiscal Policy in the European Monetary Union.Monetary and Economic Studies20: S-1Special Edition on Exchange Rate Regimes in the 21st Century12354.

    WältiS.2016. “Financial Crisis Begets Financial Reform? The Origin of the Crisis Matters.European Journal of Political Economy40: 115.

    WilsonJ. D.1999. “Theories of Tax Competition.National Tax Journal52 (2): 269304.

Mark Hallerberg: Hertie School of Governance, Berlin. Jürgen von Hagen: University of Bonn, Indiana University Kelley School of Business, the Center for Economic and Policy Research, and the Portuguese Fiscal Council. The views expressed in this paper do not represent the position of the Portuguese Fiscal Council. The authors thank Paul Hadji-Lazaro for very able research assistance.

In broad terms, the share of tax revenue as a percentage of GDP was stable, on average, in the OECD, increasing only from 32 percent in 1990 to 34 percent in 2013. These numbers, however, hide considerable variation across countries both in total revenue and in the combination of taxes in place. The most important tax in most countries is the PIT. The OECD average revenue collection as a percentage of GDP declined somewhat, from about 10 percent in 1990 to about 9 percent in 2013. However, the Slovak Republic collected PIT revenues of just 3 percent of GDP in 2013, while Denmark in that same year collected almost 26 percent of GDP. The overall reliance on VAT was proportionately lower, with increases from about 5.0 percent of GDP to 6.6 percent of GDP, on average, in the OECD and a range from 3.5 percent of GDP in Australia to almost 10.0 percent of GDP in Denmark. Finally, for CIT, on profits (and not including capital gains), Australia collected almost 5 percent of GDP in 2013 while Estonia collected less than 1 percent of GDP. Data from https://stats.oecd.org/Index.aspx?DataSetCode=REV, downloaded September 13, 2016.

See OECD (1998). For a review of the results see Avi-Yonah (2009).

These are Australia, Austria, Belgium, Canada, Chile, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, the Republic of Korea, Luxembourg, the Netherlands, New Zealand, Mexico, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, the United Kingdom, and the United States.

Data for all economic variables are taken from the International Monetary Fund’s World Economic Outlook April 2016 database.

Alesina, Favero, and Giavazzi (2015) argue that output losses are lower when fiscal consolidations are based primarily on expenditure cuts rather than on tax increases.

Perotti, Strauch, and von Hagen (1998) also show that, in contrast, the large fiscal expansions preceding fiscal consolidations are often associated with declines in the revenue from taxes on labor as well as increases in government transfers and wages.

Because each country has a weight of only 1/34 in this variable, simultaneity is unlikely to affect the estimates of the relevant coefficients.

Note that the Laeven and Valencia (2013) data set ends in 2013. We assume the banking crisis has ended since they coded a crisis in place in 2013 in Belgium, Denmark, France, Germany, Hungary, Italy, Luxembourg, the Netherlands, Sweden, Switzerland, the United Kingdom, and the United States. We assume a crisis continues through 2015 in Austria, Greece, Iceland, Ireland, Portugal, Slovenia, and Spain.

The data set through 2012 is available at http://www.uni-heidelberg.de/fakultaeten/wiso/awi/professuren/intwipol/datasets_en.html. Downloaded June 18, 2016.

Note that Gehlbach and Malesky (2010) find that increasing the number of veto players can lead to a greater probability of policy change. Their rationale is that having more active players makes it harder for a particular special interest to block the given reform. In terms of operationalizing this concept of “veto players,” Calvo (2014) finds that one-party control of the legislature or the lack of such control is most relevant in this context. To test whether one-party control affects changes in tax rates, we explored whether the “allhouse” variable from the DPI data set (Cruz, Keefer, and Scartascini 2016), which is coded as one when one party controls the relevant institutions, affects the results, but they do not change with this variable included.

See the DPI codebook for further details. We also consider other ways that fragmentation could affect the probability of changes in tax rates. Governments with majorities in parliament should have an easier time passing legislation. Similarly, governments with larger majorities should also be more likely to make changes. We include variables for both concepts from the DPI data set, namely “allhouse” and “govfrac,” as alternative measures of fragmentation. In no cases are these alternatives statistically significant.

See Kopits and Symansky (1998) for a general discussion of fiscal rules and Schaechter and others (2012) for an overview of the design and development of fiscal rules in the past 30 years.

Poterba (1995) shows how budget balance rules at the state level have had adverse effects on the dynamics of state taxes in the United States.

The data set is available at http://www.imf.org/external/datamapper/FiscalRules/map/map.htm. Downloaded June 18, 2016.

This empirical model follows the form of the one reported in Hallerberg and Scartascini (2016) but with a somewhat different set of independent variables.

One should note that with the interaction terms included, the interpretation of the marginal effect of “banking crisis” in Annex Tables 4.1.1 and 4.1.2 is whether there is a change in the tax rate when the interaction term is zero, that is, when there is not a right-leaning government in power.

We also explored a less strict definition of “consolidation” or “expansion,” which counted one year below the given cut-off instead of two. In this case, the results for the VAT are as for the stricter version, but expansions are associated with an 8 percentage point increase in the likelihood of a top PIT rate cut and a 4 percentage point lower chance of a PIT increase. This finding suggests that there is an immediate effect of top PIT rate cuts that does not affect the cyclical budget balance over two consecutive budgets. Moreover, for one-year consolidations only, increases in the top PIT rate are not statistically significant at the p < 0.1 level.

Strictly speaking, the prediction from the “checks” variable is that there should be a greater probability of no change in the three types of taxes. This corresponds to the case in which the dependent variable is coded zero. While we do not report these results here, for the CIT rate with no change, this variable is significant in the expected direction, with the marginal effect about 4 percentage points.

    Other Resources Citing This Publication