Fiscal Politics
Chapter

Chapter 6. Do Elections Affect the Wage Bill?

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

Introduction

The electoral cycle can affect fiscal policy when incumbents wish to influence political outcomes. The wage bill—which, at about 25 percent of government expenditure on average, is among the largest items in government budgets—is one important instrument at the disposal of elected officials. Increases in the wage bill generally provide a short-term boost to economic output, which can improve the odds for incumbents at the ballot box. Moreover, the wage bill largely reflects government employment and pay policies, which are crucial to ensuring key government functions are carried out, including education, health, sanitation, and security. Increases in the number of employees can visibly enhance the quantity and quality of these public services, which voters can interpret as a sign of the competence of the incumbent. Finally, in some countries the government is one of the largest employers, and increases in government pay can have a direct impact on the welfare of many households of likely voters.

Politically motivated increases in the wage bill that do not translate into improvements in service delivery could threaten fiscal sustainability, displacing resources from other priority needs or requiring higher taxes. In addition, election-driven changes in employment and compensation practices hinder civil service accountability. Furthermore, political meddling makes institutional development more difficult. In some countries, this interference can encourage the continuation of clientelistic practices that are often a barrier to development, equity, and stability.

Using a new data set on government wages and employment, this chapter analyzes the impact of elections on the government wage bill and makes three important contributions to the literature. First, the focus on the government wage bill allows the impact on an area that is particularly visible to voters to be isolated. Second, by taking advantage of a newly assembled data set, the fiscal and political cycle can be examined in more than 150 countries during 1990–2014, including 49 low-income countries (LICs). To the authors’ knowledge, this is the widest sample of countries used for this type of analysis to date.1 Third, the data on government employment provide an opportunity to examine whether changes in the wage bill in the political cycle correspond to changes in employment or to changes in pay policies—the actual policy levers driving the wage bill. This analysis fills a gap in the literature related to the scarcity of high-quality data on government employment.

This analysis indicates that elections do increase the government wage bill as a share of GDP, particularly in emerging market economies and LICs. The results are both economically and statistically significant: at current levels of the wage bill, elections result in an increase of the ratio of the wage bill to GDP of about 0.2 percentage point in emerging market economies and LICs. In addition, election-year increases in the wage bill tend to be associated more with increases in government employment than with changes in pay, with the exception of emerging economies, where government pay also tends to increase.

The remainder of this chapter is organized as follows. The second section provides a brief literature review. The third section discusses the data used in this chapter and presents descriptive statistics. The fourth section examines the main factors explaining differences in the wage bill and employment across countries and over time. The fifth section analyzes the impact of elections on the wage bill. The sixth section decomposes this impact by employment and pay. The final section concludes.

Literature Review

The interaction between the electoral cycle and macroeconomic policy has been studied extensively.2 The general consensus is that the political cycle is more likely to affect fiscal than monetary policy, with changes in the latter being largely accommodative of fiscal impulses. This finding occurs partly because fiscal policy can have real effects on the economy even when anticipated. In particular, the proximity of elections introduces incentives for incumbents to conduct fiscal expansions to stimulate the economy and improve their chances of winning. Even outgoing governments have incentives for fiscal expansion if it ties the hands of the incoming government, requiring unpopular tax increases or expenditure cuts that can raise the probability of a return to power in the next round of elections (Mulas-Granados 2006). The empirical literature is generally supportive of the electoral-cycle hypothesis on government expenditure (Klomp and de Haan 2013), with the impact being more marked in less developed economies and newer democracies (Shi and Svensson 2006; Brender and Drazen 2005).

This chapter focuses on the impact of elections on the government compensation of employees—the wage bill. This expenditure item can be particularly attractive to policymakers wishing to stimulate the economy just before an election for various reasons. First, the wage bill can have a rapid impact on output. Although theoretical models generally assume that the multiplier for government consumption (the wage bill plus goods and services expenditure) is smaller than that for government investment, no clear evidence verifies that this is the case, at least in advanced economies. In fact, empirical studies often find that the multiplier for consumption can be higher than that for investment (Batini and others 2014), which suggests that, at least in the short term, policymakers can get more bang for their buck by increasing the wage bill. Second, relative to other forms of expenditure—for example, capital spending—increases in the wage bill can be rolled out more easily because they do not require procurement processes commonly required in other forms of government consumption (goods and services purchases) or public investment. Moreover, wage increases and hiring practices are often at the discretion of the executive branch. Third, under an opportunistic political cycle, one could expect public expenditure to increase in areas most visible to voters (Rogoff 1990; Akhmedov and Zhuravskaya 2004). Voters may favorably respond to boosts in government employment in the delivery of services that affect them directly, such as education, health, sanitation, and security. In addition, as the principal employer in some countries, governments can have a direct impact on the welfare of many households through increases in public compensation. Finally, governments may be beholden to powerful constituencies (for example, public sector unions) that hold sway during political campaigns.

A number of studies focus more specifically on the wage bill. Using a sample of 24 developing economies in 1973–92, Schuknecht (2000) finds the political fiscal cycle is largely related to expansion of public expenditure during elections, although the impact on the wage bill is not statistically significant. Eckardt and Mills (2014) find that the government wage bill tends to be more procyclical and responsive to elections than total government expenditure, and the impact appears more pronounced in emerging economies. Cahuc and Carcillo (2012) find that episodes of fiscal drift (when the shares of public wage bills and deficits in GDP rise together) are more frequent during election years using a sample of Organisation for Economic Co-operation and Development countries between 1995 and 2009. In addition, Dahlberg and Mӧrk (2011), using data from Sweden and Finland, find a significant election year effect in local government employment. Drazen and Eslava (2010) also find that the wage bill in local governments in Colombia appears to increase before elections. However, these studies focus on small samples over a limited time horizon. Moreover, little evidence is provided regarding the role of employment in driving the government wage bill.

Data

This chapter uses data for 1990–2014 from different sources. For government compensation of employees and employment, it uses the newly developed IMF Harmonized Government Wage and Employment Data (IMF 2016a).3 For the political cycle, this chapter uses the 2015 Database of Political Institutions (Cruz, Keefer, and Scartascini 2016). Demographic data come from the United Nation’s World Population Prospects (United Nations 2015) and economic growth from the World Economic Outlook Database (IMF 2016b). Descriptive statistics are presented in Table 6.1. The main variables used in the analysis include the following:

Table 6.1.Descriptive Statistics of the Pooled Sample, 1990–2014
Pooled Data (1990–2014)
AllAdvanced EconomiesEmerging Market EconomiesLow-Income Countries
Government Wage Bill (percent of GDP)
Mean8.210.77.96.4
Median7.810.77.85.5
10th Percentile3.86.94.23.1
90th Percentile12.914.211.810.9
Government Employment (percent of working-age population)
Mean9.512.38.34.5
Median8.910.97.34.1
10th Percentile3.97.34.51.0
90th Percentile15.822.513.49.9
Elections
Mean Election Indicator
(1 = elections, 0 = otherwise)0.200.260.190.17
Mean Parliamentary System Indicator
(1 = parliamentary, 0 = otherwise)0.440.860.390.23
Demographics
Population
Mean29.126.438.318.3
Median7.97.56.09.2
10th Percentile0.60.40.31.6
90th Percentile62.762.272.540.3
Percent of Population, Ages 0–14 Years
Mean31.918.530.642.1
Median32.818.230.843.5
10th Percentile16.914.618.834.7
90th Percentile45.522.641.347.4
Percent of Population, Ages 65 Years and Older
Mean7.214.46.63.4
Median4.914.75.33.1
10th Percentile2.710.53.12.5
90th Percentile15.418.113.24.6
Real Economic Growth
Mean4.12.64.64.4
Median4.02.64.54.8
10th Percentile(3.3)(1.5)(3.5)(4.5)
90th Percentile12.57.113.815.0
Nonwage Expenditure (percent of GDP)
Mean23.131.521.918.1
Median22.131.820.616.5
10th Percentile11.923.712.310.1
90th Percentile36.240.333.527.7
Public Debt (percent of GDP)
Mean63.157.151.483.1
Median49.853.042.959.7
10th Percentile17.014.413.327.5
90th Percentile114.3100.490.2150.7
Observations
Government Wage Bill (percent of GDP)3,2357981,508929
Government Employment (percent of working-age population)1,438601671166
Elections and Demographics4,1228741,8731,375
Economic Growth3,7247841,6991,241
Nonwage Expenditure2,9977461,349902
Public Debt3,9938491,8191,325
Countries
Government Wage Bill (percent of GDP)156347349
Government Employment (percent of working-age population)116335429
Elections and Demographics165357555
Economic Growth165357555
Nonwage Expenditure154347149
Public Debt160347353
Average Number of Years
Government Wage Bill (percent of GDP)21232119
Government Employment (percent of working-age population)1218126
Elections and Demographics25252525
Economic Growth23222323
Nonwage Expenditure19221918
Public Debt25252525
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.
  • Wage bill. The wage bill is estimated as the general government compensation of employees as a percentage of GDP. The pooled sample covers 1990–2014 and includes 156 countries (34 advanced, 73 emerging, and 49 LICs) with an average of 21 years of data (Table 6.1). The wage bill ranges between 3.8 percent (10th percentile) and 12.9 percent (90th percentile) of GDP, with a sample mean of 8.2 percent of GDP.

  • Government employment. The indicator for government employment corresponds to general government employment as a share of the population ages 15 to 64 (an estimate of the working-age population). Data are more scarce for government employment: the pooled sample includes 116 countries (33 advanced, 54 emerging, and 29 LICs) with an average of 12 years of data. Government employment is between 3.9 percent (10th percentile) and 15.8 percent (90th percentile) of the working-age population. The sample mean government employment is 9.5 percent of the working-age population.

  • Elections. The indicator variable for election years takes the value of one in the calendar years of executive elections for presidential regimes or legislative elections for parliamentary regimes, and zero in other years. Data are available for 165 countries (35 advanced, 75 emerging, and 55 LICs) in the pooled sample 1990–2014. On average, elections represent 20 percent of the observations, being more frequent in the advanced economies (26 percent of the time) than in the emerging economies (19 percent) and LICs (17 percent).

  • Demographics. In the pooled sample, the population is 29.1 million on average and 7.9 million at the median. The average share of the population younger than age 15 is 31.9 percent and the average share of the population age 65 and older is 7.2 percent. Higher-income economies tend to have older age structures—on average, advanced economies have smaller shares of children and larger shares of elderly in their populations.

  • Economic growth. The average annual real GDP growth is 4.1 percent (median 4 percent) in the pooled sample. On average, growth declines with the level of development, implying some degree of convergence—low-income economies grow on average by 4.4 percent per year compared with 2.6 percent per year in advanced economies.

  • Nonwage government expenditure. On average, the sample government expenditure in items other than the wage bill is 23.1 percent of GDP, ranging from 18.1 percent of GDP in the LICs, to 21.9 percent of GDP in the emerging economies, and 31.5 percent of GDP in the advanced economies.

  • Public debt. The average gross public debt is 63.1 percent of GDP in the pooled sample. On average, government indebtedness is higher in the LICs (83.1 percent of GDP) than in the advanced and emerging economies (57.1 percent and 51.4 percent of GDP, respectively).

Cross-Country Variation in Government Wage Bill: Stylized Facts

Figure 6.1 illustrates the evolution of the share of the government wage bill in GDP using a balanced panel (that is, maintaining a constant number of countries over time) starting in 1990 for advanced and emerging economies and in 2000 for LICs. This figure indicates that government compensation of employees is substantially higher in advanced than in emerging economies and LICs. The average ratio of the wage bill to GDP seems relatively stable or declining in advanced economies, albeit with a noticeable increase in 2009 that coincides with the financial crisis. In emerging economies, the wage bill has been trending up since 2007. In LICs, the wage bill has been steadily rising since the early 2000s.

Figure 6.1.Compensation of General Government Employees

(Percent of GDP)

Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.

A linear regression is useful for describing the factors that explain the variation across countries and over time in the government wage bill. The specification assumes the following form:

where WB is the share of the wage bill in GDP and X is a vector of control variables that includes the level of income, the size and age distribution of the population, the level of government expenditure other than compensation of employees, the level of public debt, and time trends.

In the pooled sample, these variables explain about 39 percent of the variation in the wage bill. The results are summarized in Figure 6.2. Relative to LICs, emerging and advanced economies have higher wage bills (statistically significant for advanced economies). This finding is consistent with Wagner’s law—as economies develop, governments tend to expand to respond to higher demand from the population for the provision of public goods, which tend to be superior goods (Bird 1971; Shiavo-Campo and Sundaram 2001). In addition, as economies develop, the evolving nature of economic activity might facilitate tax collection (for example, the growing share of wages and salaries), providing more resources to finance public activities (Tanzi and Schuknecht 2000).

Figure 6.2.Factors Associated with the Variation in the Ratio of the Government Wage Bill to GDP

(Percentage points of GDP)

Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.

Note: Figure depicts coefficients of an ordinary least squares regression on the share of government compensation of employees in GDP. Sample size is 2,813 country-years and adjusted R-squared is 0.39. For the dependent variables population, share of population younger than 15 years, share of population 65 years and older, nonwage expenditure, and public debt, the figure shows the impact of a one standard deviation increase in these variables. For the time trend, the figure shows the 10-year trend.

**p < .05; ***p < .01.

Demographic factors explain part of the variation in the government compensation of employees. The negative and statistically significant coefficient for population indicates scale effects in the government wage bill—all else equal, a one standard deviation increase in the size of the population is associated with a reduction in the wage bill of 0.9 percentage point of GDP. The age profile of a country also affects the wage bill. A larger share of children in the population can create pressure to increase the government wage bill and employment associated with education (a one standard deviation increase in the share of children increases the wage bill by 0.3 percentage point of GDP). In addition, a one standard deviation increase in the share of the elderly is associated with an increase in the wage bill of 0.5 percentage point of GDP, possibly reflecting pressures related to health care provision.

Fiscal considerations also affect the wage bill. A one standard deviation increase in nonwage expenditure increases the wage bill by 1.4 percentage points. This finding likely reflects country preferences on the size of government. In contrast, increases in public debt are negatively correlated with the wage bill—a one standard deviation increase in public debt reduces the wage bill by 0.2 percentage point of GDP, suggesting that concerns about fiscal sustainability might be associated with wage bill containment.

Finally, the time trend coefficients imply that the wage bill has been declining gradually over time (by about 0.3 percentage point every decade, all else equal).

Impact of Elections on the Government Wage Bill

Incumbents have an interest in using fiscal policy to influence voters’ choices. This influence could be achieved by increasing civil service wages and boosting government hiring before elections, which would increase household incomes for public servants and potentially improve service delivery. Assuming voters put sufficient weight on these outcomes in their political choices, increases in the wage bill could tilt the ballot box in favor of incumbents. Following the literature, the chapter evaluates this hypothesis by examining changes in the ratio of the wage bill to GDP in years with and without elections.

Table 6.2 presents evidence of a link between changes in the wage bill and elections. The data give credence to the presence of a political cycle in the wage bill—the wage bill increases more strongly in election years than in years without elections, particularly in LICs. The average increase during election years is higher than in years without elections by about 0.08 percentage point in the whole sample and 0.19 percentage points in LICs. The results are similar at the median, albeit of smaller magnitude.

Table 6.2.Changes in Government Wage Bill, by Election Year, 1990–2014
Change in Wage Bill to GDP
AllAdvanced EconomiesEmerging Market EconomiesLow-Income Countries
Mean
All0.06–0.020.070.11
No Election0.04–0.030.050.07
Election0.120.010.120.27
Difference (Election − No Election)0.080.050.060.19
Median
All0.00–0.050.000.08
No Election0.00–0.05–0.010.07
Election0.05–0.040.060.20
Difference (Election − No Election)0.050.000.080.13
Observations
All3,0747641,433877
No Election2,4425571,170715
Election632207263162
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.

To assess the impact of elections on the wage bill while controlling for other factors, a regression is estimated that includes macroeconomic and structural controls:

where fi represents country fixed effects, ΔWB is the change in the wage bill as a percentage of GDP, and X is a vector with the key explanatory variable being an indicator variable for election years, including a differential for advanced economies (an interaction between election year and advanced economy indicators). The analysis uses four different specifications. The first three correspond to ordinary least squares regressions using the level of development (an indicator variable for advanced economies), demographics (annual population growth and changes in the shares of children and elderly in the population), and annual GDP growth. The fourth specification is a fixed-effects regression that adds controls for country-specific characteristics and year effects.

Regression analyses confirm the impact of elections on changes in the ratio of the government wage bill to GDP while controlling for other factors (Table 6.3).4 Elections increase the government wage bill in relation to GDP by 0.11–0.19 percentage point (statistically significant under all specifications). At the current level of wages to GDP, this is equivalent to about a 2 percent increase in the wage bill in low-income and emerging economies and a 1 percent increase in advanced economies. The negative coefficient in the interaction between elections and advanced economies indicates that the impact is much less marked in advanced economies.

Table 6.3.Regression Analysis: Impact of Elections on Changes in the Government Wage Bill
OLSFixed Effects
CoefficientP-valueCoefficientP-valueCoefficientP-valueCoefficientP-value
Dependent Variable: Change in Government Wage Bill to GDP
Constant0.082***0.000.061**0.01–0.003*0.060.010***0.00
Advanced Economy–0.104***0.00–0.096***0.00–0.0200.58
Election0.112**0.010.186***0.000.167***0.00
Election × Advanced Economy–0.0640.29–0.174***0.01–0.160***0.01
Growth of Population0.9430.495.033*0.05
Change in Share of Population, Ages 0–14 Years0.0650.260.147**0.03
Change in Share of Population, Ages 65+ Years0.0350.700.1860.21
Growth of Real GDP (t − 2)0.866***0.000.915***0.00
Change in Nonwage Expenditure (t − 2)–0.0060.12–0.0090.11
Change in Public Debt (t − 2)–0.003*0.06–0.0010.71
Year EffectsNoNoNoYes
R-squared0.000.000.030.11
Observations3,0743,0742,4562,456
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.Note: OLS = ordinary least squares.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.Note: OLS = ordinary least squares.*p < .1; **p < .05; ***p < .01.

The growth of the population has a positive effect on the wage bill, and the effect is particularly significant in the fixed-effects specification. Although larger countries tend to have lower wage bills because of scale effects (the analysis in levels presented in the previous sections), short-term positive shocks to population growth increase wage bill expenditures, likely as a result of short-term responses to maintain adequate service delivery.

The positive coefficient on the (lagged) growth of real GDP implies that positive short-term shocks to GDP translate into a higher wage bill as a percentage of GDP. Note that the correlation between the growth rates of these two variables is 0.56, suggesting procyclicality—when real GDP increases, the wage bill tends to increase in real terms.

The coefficients on the fiscal variables suggest that increases in other types of spending are not statistically significant. With regard to (lagged) public debt, the coefficients indicate that a positive shock to public debt tends to reduce the wage bill, although this result is not statistically significant in the fixed-effects specification.

One important question is whether the election effects on the wage bill are persistent—if the impact of elections is offset by reductions in the wage bill in postelection years, then the political cycle would not materially affect the long-term path of the wage bill. Repeating the exercise using years just preceding or following election years as explanatory variables yields coefficients that are not statistically different from zero (Table 6.4). This outcome suggests that the wage bill is not ramped up in anticipation of elections and that the additional spending is not reduced even after the votes have been counted.

Table 6.4.Regression Analysis: Impact of Elections on Changes in the Government Wage Bill, by Year Relative to Election Year
Coefficients of Election Dummy
Fixed Effects
Election Year0.17***
One Year after Election–0.05
Two Years after Election0.04
One Year before Election–0.01
Two Years before Election–0.05
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.Note: Table displays coefficient for year dummies relative to election years, using a fixed-effects specification similar to that in the last column of Table 6.3.**p < .05; ***p < .01.
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.Note: Table displays coefficient for year dummies relative to election years, using a fixed-effects specification similar to that in the last column of Table 6.3.**p < .05; ***p < .01.

The results are robust to different specifications. A regression using the wage bill as a share of total government expenditure as a dependent variable provides comparable results. The coefficient for elections is 0.75 percentage point of expenditure, which implies an increase in the wage bill as a percentage of GDP of about 0.2 percentage point.

Drivers of Wage Bill Increases During Elections

This section examines whether increases in the wage bill associated with elections are due to increases in employment or to increases in pay. The ratio of the wage bill to GDP can be expressed as the product of the number of government workers as a share of the working-age population and average government pay as a share of GDP per working-age population:

where average government pay is defined as governmentwagebillgovernmentemploymemt.

Using this identity, the change in the wage bill is decomposed into the increase due to an expansion of employment and the increase due to a rise in average pay.

To examine the impact of elections, a fixed-effects regression is used for each component. The decomposition in election years is compared with that in non-election years (Table 6.5). The sample is restricted to country-years with nonmissing data for both the ratio of the wage bill to GDP and the ratio of employment to the working-age population. The results for the wage bill are similar to those in the full sample presented in Table 6.3—the wage bill to GDP increases by 0.09 in election years, and this result is statistically significant. The decomposition suggests that about two-thirds of the impact is due to changes in employment and one-third due to increases in wages.

Table 6.5.Decomposition of Changes in the Ratio of the Wage Bill to GDP
AllAdvanced EconomiesEmerging Market EconomiesLow-Income Countries
Impact of Election on Government Wage Bill to GDP0.09**0.050.17**0.05
Due to Government Employment0.060.050.050.17*
Due to Government Pay0.030.000.12–0.11
Observations97045743776
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.Note: Table displays coefficient for year dummies relative to election years, using a fixed-effects specification similar to that in the last column of Table 6.3. *p < .1; **p < .05.
Source: Authors’ calculations using IMF Harmonized Government Wage and Employment Data.Note: Table displays coefficient for year dummies relative to election years, using a fixed-effects specification similar to that in the last column of Table 6.3. *p < .1; **p < .05.

In advanced economies, the annual change in the wage bill in election years is 0.05 percentage point of GDP, but this is not statistically significant. The decomposition indicates that the impact of elections in the wage bill is largely due to increases in the ratio of government employment to working-age population, while government pay does not seem to be affected by elections.

In emerging economies, the average annual change in the wage bill is positive and significant (0.17 percentage point per year). In these countries, only one-third of the impact is due to employment, and the remainder is due to changes in government pay.

In LICs, the wage bill increases by 0.05 percentage point of GDP in election years. In these countries, election year increases seem more associated with increases in employment (associated with a 0.17 percentage point of GDP increase in the wage bill, statistically significant) than with changes in government pay (which seems to decline in election years, but the result is not statistically significant).

Conclusion

Substantial evidence suggests that electoral cycles can affect fiscal policy, but less is known about whether governments use the public wage bill to affect political outcomes. This chapter sheds new light on this question by taking advantage of a newly assembled data set covering 25 years of public wage bill data in more than 150 countries, including 49 LICs. The data set also provides information on public employment, which is used in this analysis to separate the contribution of average pay from that of employment in election-cycle-induced changes in the public wage bill.

The chapter finds evidence that the wage bill is subject to a political cycle. Positive changes in the wage bill are associated with election years, particularly in emerging economies and LICs. The impact of elections is robust as well as economically and statistically significant, increasing the share of public wages in GDP by 0.11–0.19 percentage point. This finding is the equivalent of an approximately 2 percent increase in the wage bill in emerging economies and LICs and a 1 percent increase in advanced economies at current levels of public wages to GDP. The analysis finds some evidence that elections have a positive impact on public employment, though the result is not robust across specifications. Taking advantage of the data on government employment, the analysis finds that overall, election-year increases in the wage bill tend to be associated more with increases in government employment than with changes in pay, with the exception of emerging economies, where government pay also tends to increase.

This work has important implications for public policy. Efforts to ensure fiscal sustainability should emphasize the need to strengthen wage bill oversight and depoliticize government pay and employment policies. One attractive option for delinking pay raises from the election cycle is to rely on independent agencies for advice on wage increases, typically taking into account private sector developments and fiscal affordability (as is done, for example, in Australia and Japan). For employment, it is crucial to control hiring to ensure that it reflects genuine service delivery needs. This effort should be supported by periodic functional reviews to help identify potential overlap in functions across different government entities.

References

    AkhmedovA. and E.Zhuravskaya.2004. “Opportunistic Political Cycles: Test in a Young Democracy Setting.Quarterly Journal of Economics119 (4): 130138. http://qje.oxford-journals.org/content/119/4/1301.short.

    BatiniN.L.EyraudL.Forni and A.Weber.2014. “Fiscal Multipliers: Size, Determinants, and Use in Macroeconomic Projections.Technical Notes and Manuals International Monetary Fund Washington. https://www.imf.org/external/pubs/ft/tnm/2014/tnm1404.pdf.

    BirdR. M.1971. “Wagner’s Law of Expanding State Activity.Public Finance26 (1): 126.

    BrenderA. and A.Drazen.2005. “Political Budget Cycles in New versus Established Democracies.Journal of Monetary Economics52 (7): 127195. http://www.sciencedirect.com/science/article/pii/S0304393205000887.

    CahucP. and S.Carcillo.2012. “Can Public Sector Wage Bills Be Reduced?Working Paper 1788National Bureau of Economic ResearchCambridge, MA. http://bit.ly/2cwZcQr.

    CruzC.P.Keefer and C.Scartascini.2016. “Database of Political Institutions Codebook, 2015 Update (DPI2015).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.

    DahlbergM. and E.Mӧrk.2011. “Is There an Election Cycle in Public Employment? Separating Time Effects from Election Year Effects.CESifo Economic Studies57 (3): 48098. http://bit.ly/2cD4Dwb.

    DrazenA.2001. “The Political Business Cycle after 25 Years.” In NBER Macroeconomics Annual 2000 edited by B.Bernanke and K.Rogoff75117. Cambridge, MA: MIT Press.

    DrazenA. and M.Eslava.2010. “Electoral Manipulation Via Voter-Friendly Spending: Theory and Evidence.Journal of Development Economics92 (1): 3952.

    DuboisE.2016. “Political Business Cycles 40 Years after Nordhaus.Public Choice166 (1): 23559. http://link.springer.com/article/10.1007/s11127-016-0313-z.

    EckardtS. and Z.Mills.2014. “What Goes Up Must Come Down: Cyclicality in Public Wage Bill Spending.Working Paper 6760World BankWashington. http://bit.ly/2cCXvzW.

    International Monetary Fund (IMF). 2016a. “Managing Government Compensation and Employment—Institutions, Policies, and Reform Challenges.Policy Paper IMF Washington DC. https://www.imf.org/external/np/pp/eng/2016/040816a.pdf.

    International Monetary Fund (IMF). 2016b. World Economic Outlook Database. Washington. https://www.imf.org/external/pubs/ft/weo/2016/01/weodata/index.aspx.

    KlompJ. and J.de Haan.2013. “Political Budget Cycles and Election Outcomes.Public Choice157 (1): 24567. http://link.springer.com/article/10.1007/s11127-012-9943-y.

    Mulas-GranadosC.2006. Economics Politics and Budgets: The Political Economy of Fiscal Consolidations in Europe.Basingstoke: Palgrave Macmillan.

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

    RogoffK.1990. “Equilibrium Political Budget Cycles.American Economic Review80 (1): 2136. http://scholar.harvard.edu/files/rogoff/files/51_aer90.pdf.

    Schiavo-CampoS. and P.Sundaram.2001. To Serve and to Preserve: Improving Public Administration in the Competitive World.Manila: Asian Development Bank.

    SchuknechtL.2000. “Fiscal Policy Cycles and Public Expenditure in Developing Countries.Public Choice102 (1): 11530. http://link.springer.com/article/10.1023/A:1005026806998.

    ShiM. and J.Svensson.2006. “Political Budget Cycles: Do They Differ across Countries and Why?Journal of Public Economics90 (8–9): 136789. http://bit.ly/2cvbax6.

    TanziV. and L.Schuknecht. 2000.Public Spending in the 20th Century: A Global Perspective.Cambridge, U.K.: Cambridge University Press.

    United Nations. 2015. “World Population Prospects: The 2015 Revision, Methodology of the United Nations Population Estimates and Projections.Department of Economic and Social Affairs Population Division New York. https://esa.un.org/unpd/wpp/.

Past research has relied on smaller sample sizes that include very few observations of LICs. Klomp and de Haan (2013) include 30 advanced, 28 emerging, and 7 LICs; Brender and Drazen (2005) include 31 advanced, 30 emerging, and 7 LICs; Shi and Svensson (2006) use 27 advanced, 38 emerging, and 20 LICs; Cahuc and Carcillo (2012) include 29 advanced and 5 emerging; Eckardt and Mills (2014) use 6 advanced, 19 emerging, and 3 LICs; and Schuknecht (2000) uses 24 developing economies (a group that includes both emerging and LICs).

Nordhaus (1975) develops a formal model illustrating the interaction between elections and monetary policy. Assuming inflation expectations depend on recent changes in prices (that is, myopic voters), policymakers can maximize their probability of reelection by conducting expansionary monetary policy to stimulate the economy before the election and reversing course after voters have cast their ballots. Research has since extended this work to integrate forward-looking voters, rational expectations, and fiscal policy and to conduct empirical tests (Drazen [2001] and Dubois [2016] provide surveys of this literature).

For compensation of employees, the IMF data set compiles data from sources including the IMF’s World Economic Outlook and Government Finance Statistics, the Organisation for Economic Co-operation and Development’s General Government Accounts, and EUROSTAT’s Annual Government Finance Statistics and AMECO. For general government employment, the IMF data set assembles data from the International Labour Organization’s LABORSTA data (public sector employment and employment of general government sector) and ILOSTAT (employment by institutional sector), and data from individual countries.

Similar regressions using the share of government employment in the working-age population suggest a positive impact of elections of 0.06–0.08 percentage point of the working-age population, although this coefficient is only statistically significant in one specification.

    Other Resources Citing This Publication