Growth of government and the politics of fiscal policy

Abstract

US government expenditures increased rapidly during the post-war period, then slowed in the 1980s and began falling in 1992. To examine the dynamics of the growth and subsequent reduction in government spending, we present a general equilibrium growth model in which politicians chose government spending to maximize support by their constituents. That is, output and government spending are endogenous and jointly determined. The model predicts that government expenditures will initially mimic Wagner's law—the tendency for government spending to increase with GDP—but eventually diverge from output due to the growth of the welfare state. After government expenditures become large, we identify an endogenous threshold on the economy's growth path where it is optimal for politicians to shrink the welfare state, cut taxes, and stimulate output growth. We show that the policies chosen by politicians are Pareto suboptimal and cause endogenous cycles in output. Such cycles are of several types, and we characterize when the equilibrium growth path will result in a reduction in the size of the welfare state, as well as when the welfare state cycles between small and large.

Introduction

Between 1929 and 2000 US real GDP grew at an average rate of 3.4% a year. During the same time period, real government outlays (federal, state and local) grew at an average of 3.2% a year. Writing in 1893, Adolph Wagner posited that increased political pressures would accompany the development of modern industrial societies, giving rise to a continual expansion of the public sector. When government spending grows faster than output, ‘Wagner's law’ is said to hold. Empirical tests of Wagner's law for developed countries affirm its existence.¹ Modern explanations for this finding range from extensions of the franchise which reduced the income of the median voter and increased transfers (Meltzer and Richard, 1981, Husted and Kenny, 1997) and enfranchised women (Lott and Kenny, 1999), to legislative self-oversight (Miller and Moe, 1983) and increasing centralization (Borcherding, 1985), to more extensive government monitoring required in an increasingly complex economy (Chapell and Keech, 1985) and a complicated tax structure (Becker and Mulligan, 1998), to the role of government agencies and size of bureaucracy (Wilson, 1989), to an aging population which increased transfers (Azariadis and Lambertini, 1997).² The raison d’être of this paper is that the sources of government growth cannot be understood without examining the motivations of those setting policy. Further, policy determination must account for the interdependence between fiscal policy and output growth.

Conducting empirical tests for the period 1929–2000, we show in Section 2 that Wagner's law does not hold for the US. Fig. 1 informally demonstrates this by plotting aggregate real government expenditures—not including defense expenditures—as a percentage of real GDP.³ The figure partitions the data into four distinct regions. Though the data are noisy before WWII, Wagner's law appears to hold from 1945 to 1975. This relationship starts to breakdown after 1975 as government spending slowed markedly while output growth accelerated, invalidating the presumed co-movement between them. More starkly, US aggregate government expenditures began to fall in 1992, with this decline continuing through the present, while output growth during the 1990s was extraordinarily high.

Taking the trend depicted in Fig. 1 as a point of departure, this paper constructs a political theory of the composition of government expenditures within a neoclassical growth model to explain the pattern of aggregate government expenditures. The theory shows that over subperiods, Wagner's law holds as the growth of the ‘welfare state’ exceeds output growth.⁴ We demonstrate that after government expenditures become a large proportion of the economy, a threshold emerges at which politicians optimally reduce the size of the welfare state—at least temporarily—in order to maintain positive output growth. The model thus predicts an endogenous switch in the time trend of government expenditures, just as US data show. The catalyzing factor driving these results is the choice by politicians of both the level and composition of government expenditures which is made to maximize support from their constituents. When politicians set policy, we show that the size of the welfare state oscillates. We characterize the sources and types of oscillations, and show that after the welfare state shrinks, there are strong incentives for its subsequent growth.

Comparing the equilibrium dynamics induced by politically motivated policy-setting to Pareto optimal policies and constant policies, we show that political policies not only are Pareto-suboptimal, but result in distinctly different dynamic paths for the economy. Indeed, Pareto optimal policies produce endogenous growth in which the economy never reaches a steady state, while politically motivated policies and constant policies lead to steady states in per capita income. Perhaps most interestingly, political incentives cause the equilibrium path of the economy to exhibit cycles. Output oscillations arise in an economy that would otherwise have a monotone growth path, and are the direct result of politically motivated policy-setting.

The paper is structured as follows. Section 2 discusses the results of cointegration tests of real US government expenditures and GDP. These tests reject Wagner's law which motivates the model of policy-setting in a dynamic economy presented in Section 3. In Section 3 we derive politically-motivated policy choices and characterizes both the aggregate impact of such policies and the size of the welfare state. Section 4 draws implications from the analysis and concludes.