Spending cuts or tax increases? The composition of fiscal adjustments as a signal

Abstract

This paper shows that the composition of fiscal adjustments, spending cuts versus tax increases, serves as a signal of the government's degree of collusion with special interests. The politico-economic model of fiscal policies, combining retrospective voting with common-agency-type lobbying, presents undominated separating equilibria and intuitive pooling ones, in both of which fiscal adjustments with sufficiently large spending cuts lead to incumbent reappointment whereas those with only tax increases lead to incumbent defeat. These findings are consistent with the recent empirical evidence of voters behaving as fiscal conservatives. The efficiency-enhancing aspects of the signaling mechanism and the effects of imposing a deficit limit are also analyzed.

Introduction

This paper attempts to explain the public choice of the composition of fiscal adjustments, spending cuts versus tax increases, from a signaling point of view. In the literature of recent political economics, while the determination of budget deficits has been analyzed intensively, there is no equilibrium theory, at least to my knowledge, that addresses the issue.¹ Within a politico-economic model that combines retrospective voting with common-agency-type lobbying, this paper will reveal the role of the composition of fiscal adjustments as a signal of the government's hidden degree of collusion with special interests.

A key motivation and inspiration for this research is the range of recent empirical studies on voters’ responses to preelectoral fiscal policy manipulation. Although numerous empirical studies tested for systematic business or fiscal fluctuations coinciding with election periods in major developed countries, they found mixed results at best.² In sharp contrast to such political budget cycle literature, Peltzman (1992), Besley and Case (1995a, b), Alesina et al. (1998), and Brender (2003), among others, present evidence of voters behaving as “fiscal conservatives,” who often tend to remove deficit-producing incumbents from office.³ These studies found that spending cuts are politically beneficial, while the results for tax increases are somewhat mixed. For instance, Peltzman (1992) examined the data on U.S. Presidential and gubernatorial elections and found that a spending increase, not a tax increase, statistically significantly decreases the incumbent party's vote share. Alesina et al. (1998) analyzed recent fiscal data from 19 OECD countries and pointed out that in a period of fiscal adjustment, governments relying on spending cuts may survive longer or enjoy greater popularity than those relying on tax increases. Besley and Case (1995a) found, using data on U.S. gubernatorial elections, that incumbents hold spending low in their first term in office but expand it significantly in the lame-duck term. Besley and Case (1995b) also revealed, using the same data set, that an increase in a state's own taxes produces a higher probability of incumbent defeat. Brender (2003) examined the data on Israeli local elections to find that the excess wages paid in public entities had a significant negative effect on the mayors’ reelection probability.

Fiscal adjustment in Japan during the 1980s, mostly implemented under the Nakasone cabinet (1982–1987), is also an interesting episode in line with the above empirical studies as well as the present analysis. Partly because of recessions after the oil crises, the central government, in 1982, was suffering from an unprecedented huge budget deficit when Prime Minister Yasuhiro Nakasone took office. He embarked on a policy of fiscal consolidation, following the proposals that the Provisional Commission on Administrative Reform (Rincho) recommended under a slogan of “fiscal reconstruction without tax increases.” Expenditure management through setting ceilings on budget requests resulted in spending cuts, particularly in areas such as agriculture, health care, and public works, the areas where Japan's most powerful interest groups press for expenditures often criticized as wasteful by the media. Privatization of the national railway company, whose continual deficits placed a big fiscal burden on the government, was also carried out. Thanks to these expenditure restraint initiatives, the government decreased its deficit drastically with no new taxes introduced and finally succeeded in eliminating a new issue of deficit-financing bonds in the FY 1990 budget.⁴ Those achievements boosted the popularity of Mr. Nakasone, coupled with his presence as a global political figure (such as his close relationship with U.S. President Ronald Reagan). Indeed, exceptionally, he was allowed by the ruling Liberal Democratic Party to be reelected twice as prime minister, holding the office for longer than anyone else over the last three decades.

Recent developments in common-agency lobbying models (for e.g. Bernheim and Whinston (1986), Grossman and Helpman (1994), Grossman and Helpman (1996a), and Dixit et al. (1997), among others) also motivate the research. In those models, inefficient policy outcomes occur when only some sections of society are organized to participate in lobbying. In the context of the political economy of fiscal policies, the inability of unorganized citizens to lobby is the reason why government makes socially wasteful outlays and accumulates deficits in favor of special interests. In other words, those models treat unorganized citizens as a silent majority with no means of affecting political decision-making.

Despite being unable to lobby collectively, they do have a means of disciplining the government, namely voting. If they could identify the degrees of collusion between politicians and special interests, they would be voting to replace more-collusive incumbents with less-collusive ones. Realistically, however, they seem not to have such information, and electoral control is undermined even though they are the majority of voters. If this is the case, the theoretical implication of common-agency lobbying models will not lose power.

Yet, even in a circumstance of asymmetric information, fiscal policies may signal the degrees of collusion. Though being unaware of which candidate is more collusive, uninformed voters can vote retrospectively, based on the past fiscal performance of the government. If loose fiscal policies were implemented, for example, voters might cast ballots against the incumbent in the next election, inferring that he is more collusive with special interests than other candidates.⁵ With the majority of voters behaving in such a way, incumbents may risk future political rents stemming from reappointment when choosing policy in favor of special interests today.

This argument suggests that unorganized citizen-voters can influence the government's fiscal decision in a similar way to lobbying. Just as special interests promise to pay political contributions to policymakers for implementing fiscal policies in their favor, unorganized citizen-voters reward them with reelection rent for implementing fiscal policies that convincingly reveal themselves as less-collusive politicians. By means of retrospective voting, unorganized general citizens no longer remain a silent majority but may implicitly act as another lobby.

Combined with uninformed voters’ retrospective voting, the common-agency lobbying models may produce novel theoretical predictions on the outcome of special-interest politics. The government will make fiscal decisions to build a good reputation among uninformed voters for reelection as well as to please special interests for political contributions. It may then engage in fiscal adjustments, changing fiscal policies from those that special interests want toward what the general public needs.

The information about the degrees of collusion between policymakers and special interests is conveyed through not only the size but also the composition of fiscal adjustments, spending cuts versus tax increases. The reason is that different degrees of collusion result in different political costs of fiscal adjustments, depending on the composition. For those more collusive with special interests, the cost of cutting socially wasteful outlays is higher, relative to that of increasing taxes, since they care more about the well-being of special interests. This implies that tax-based fiscal adjustments are likely to produce beliefs among uninformed voters that incumbents are more collusive with special interests. Conversely, to establish a reputation of being less collusive, they have to carry out fiscal adjustments based primarily on spending cuts. This scenario seems consistent with the recent empirical results cited above and also with the history of fiscal adjustment in Japan during the 1980s.

This paper will formalize the signaling mechanism and investigate the disciplinary power of retrospective voting. It constructs a four-stage game of political competition among three actors: special interests, politicians, and unorganized citizen-voters. Politicians are potentially of two types: more and less collusive with special interests. The type is known only to special interests who engage in lobbying, while unorganized citizens cannot lobby but are the majority of voters.

The game proceeds as follows. In the first stage, nature chooses the type of incumbent politician. In the second stage, special interests offer a menu of political contributions to the incumbent, knowing his type. In the third stage, he chooses a fiscal policy, and in the fourth stage, an election is held, where two candidates, the incumbent and a challenger, run for the next term of office. Unorganized citizen-voters want to appoint only a less-collusive politician. Because of asymmetric information, however, they have to decide which candidate to elect using only the information of the incumbent's past fiscal policy. Employing the concept of perfect Bayesian equilibrium and applying domination-based refinement, the analysis focuses on the undominated separating equilibria as well as the intuitive pooling ones of the game.

The results are as follows. In undominated separating equilibria, a unique policy outcome is realized. A less-collusive incumbent implements fiscal adjustments and is reelected with certainty, while a more-collusive one carries out a pro-lobby fiscal policy, financing socially wasteful outlays with a deficit, and is ousted with certainty. The composition of the equilibrium fiscal adjustment varies with the size of reelection rewards. It is based only on spending cuts in the case of small reelection rewards and tends to rely more on tax increases as they are larger. An incumbent cannot survive in power after implementing fiscal adjustments with only tax increases. Social welfare is also enhanced in the equilibria, compared with the level realized in the absence of retrospective voting. The signaling mechanism not only enables voters to oust a more-collusive incumbent but also induces a less-collusive one to cut wasteful spending. The larger the reelection rewards, the more the wasteful spending that is eliminated. With sufficiently large reelection rewards, government spending is adjusted to the level that maximizes the social surplus.

The game also has pooling equilibria that survive the intuitive criterion. The intuitive pooling equilibria generally produce multiple policy outcomes for given reelection rewards, with both types of incumbent reelected with certainty. For instance, there always exists an intuitive pooling equilibrium in which both types of incumbent engage in fiscal adjustment in the same way as a less-collusive incumbent does in the undominated separating equilibria. However, there also exists one in which both types choose the pro-lobby fiscal policy that a less-collusive incumbent would implement in the absence of retrospective voting. In the latter pooling outcome, a more-collusive incumbent is induced to cut wasteful spending but is allowed to survive longer in power, as compared with the separating equilibria. The possibility of pooling equilibria may undermine the power of retrospective voting to discipline self-interested politicians.

Focusing on the undominated separating equilibria, this paper will also explore the desirability of a deficit limit. Tabellini and Alesina (1990) argued for such a budget rule using the framework of strategic deficit creation, where incumbents strategically accumulate public debt to constrain the policy choices of their successors.⁶ In the model of this paper, running deficits binds no future policies at all, because of the assumed availability of lump-sum taxes. Yet a deficit limit is shown to promote reduction in wasteful spending, because it induces special interests to pay less to policymakers for spending expansion and thereby increases their incentives for reelection.

How the model of this paper differs from those in the related literature should be clarified. The earlier literature on retrospective voting, such as Barro (1973), Ferejohn (1986), Austin-Smith and Banks (1989), and Reed (1994), assumes voters’ collective precommitment to a voting strategy contingent on the government's policy decision. Such precommitment is precluded here, as in Rogoff (1990), Harrington (1993), Coate and Morris (1995), and Dhami (2003). Regarding fiscal policies, the availability of lump-sum taxes is assumed here to analyze the signaling effects in isolation from those related to intertemporal tax smoothing (Barro, 1979, Cukierman and Meltzer, 1989) and strategic deficit creation (Alesina and Tabellini, 1990, Persson and Svensson, 1989, Tabellini and Alesina, 1990, Aghion and Bolton, 1990).

Rogoff (1990) is the seminal paper that analyzed a signaling model of fiscal policy, focusing on the hidden competence of policymakers, though he does not address special-interest politics or budget deficits. Assuming that a more competent politician can cut spending at a smaller cost, we may construct a Rogoff-type politico-economic model that generates spending-based fiscal adjustment as a signaling equilibrium outcome. However, unlike the model of this paper, it will fail to explain fiscal adjustments composed of both spending cuts and tax increases. Also, in this kind of model, every pooling equilibrium is ruled out by the intuitive criterion.

Grossman and Helpman (1996b) combined electoral competition with lobbying in a different manner from this paper, to give a rigorous justification for the reduced-form government objective assumed in their common-agency lobbying models. They incorporated lobbies’ campaign contributions into a spatial model of two-party electoral competition, assuming that the parties can spend that money to attract additional votes. Following this foundation, it is no longer valid to claim that the common-agency lobbying models treat unorganized citizens as a silent majority. Nonetheless, the model of this paper differs from theirs in two important respects. First, it introduces asymmetric information about the type of incumbent and examines the effect of retrospective voting on political outcome, precluding his precommitment to any policy platform in advance of election. Second, while neglecting the effect of political contributions on his chance of reappointment, the model instead takes account of their effect on his incentive for reappointment and also deals with the effect of the voters’ beliefs on electoral outcomes explicitly.

The rest of this paper is organized as follows. The next section presents the model consisting of two periods. For expositional convenience, we will concentrate on the game in period 1, taking the period 2 equilibrium payoffs as given. The derivation of the equilibrium in the period 2 game, as well as its formal description, is left to Appendix A. Section 3 characterizes the fiscal policies realized in the perfect Bayesian equilibria. Section 4 focuses on the unique policy outcome realized in the undominated separating equilibria, and Section 5 analyzes the multiple policy outcomes realized in the intuitive pooling equilibria. Section 6 examines the effects of introducing a deficit limit. Section 7 concludes the paper. The proofs of propositions are collected in Appendix B.