آینده امنیت اجتماعی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|24212||2008||22 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Monetary Economics, Volume 55, Issue 2, March 2008, Pages 197–218
We analyze the effect of the projected demographic transition on the political support for social security, and equilibrium outcomes. Embedding a probabilistic-voting setup of electoral competition in the standard OLG model with capital accumulation, we find that intergenerational transfers arise in the absence of altruism, commitment, or trigger strategies. Closed-form solutions predict population ageing to lead to higher social security tax rates, a rising share of pensions in GDP, but eventually lower social security benefits per retiree. The response of equilibrium tax rates to demographic shocks reduces old-age consumption risk. Calibrated to match features of the U.S. economy, the model suggests that, in response to the projected demographic transition, social security tax rates will gradually increase to 16%. Other policies that distort labor supply will become less important; labor supply therefore will rise, in contrast with frequently voiced fears.
Many countries with pay-as-you-go financed social security systems are confronted with a secular decline in population growth rates that puts increasing financial stress on these systems. Projections of social security shortfalls in those countries imply that solvency requires substantial increases in contribution rates or cuts in benefits (or a combination of the two). The questions are: which of these options will be implemented, and what will be the macroeconomic consequences? In this paper, we develop a robust analytical framework to address these questions. Moreover, we apply the framework to generate predictions for the medium-term outlook of the U.S. social security system. Benefit levels and contribution rates are politically determined and may at any time be altered in the legislative process. In order to predict adjustments of social security taxes and benefits, it is therefore essential to model the determinants of the political support for social security, and the effect of the demographic transition on these determinants. To this purpose, we introduce political choice in Diamond's (1965) overlapping generations model. We solve for the politico-economic equilibrium and analyze the effect of the projected demographic transition on policies and the allocation. Households in the model are non-altruistic. As consumers, they take prices and policy as given. As voters, they anticipate the effect of policy on equilibrium outcomes, including future political choices. We assume that voters are not bound by past political decisions. The politico-economic equilibrium therefore features subgame-perfect tax and transfer choices supporting a competitive equilibrium. Agents hold rational expectations. Voters, in particular, are fully aware of the equilibrium relationship between future state variables and policy choices, and this relationship shapes their preferences over contemporaneous policy choices. We posit that only fundamental state variables affect future policy outcomes, excluding artificial state variables of the type sustaining trigger strategy equilibria. Our underlying assumption is that changes in the size of social security programs depend more directly on the economic and demographic environment than through underlying reputation mechanisms, even if the existence of these programs may also owe to reputational forces. Focusing on the Markov perfect equilibrium, we aim at identifying fundamental and robust forces that affect the size of social security, without relying on arbitrary assumptions about the parameters of a trigger-type strategy. 1 In fact, the Markov perfect equilibrium we focus on is the unique equilibrium arising in the limit of the finite-horizon economy. We model electoral competition under the assumption of probabilistic voting. The policy platforms of vote-seeking candidates therefore cater to the interests of all voters in society, reflecting both support for social security benefits by the elderly and opposition against them by young tax-payers. However, young voters oppose social security less emphatically than old voters support it, and the politico-economic equilibrium therefore features a structural “bias” in favor of intergenerational transfers. This bias arises because social security taxes do not only generate a cost for workers, but also indirect benefits: By depressing savings, labor income taxes allow to monopolize the supply of capital, and thus, to manipulate the terms of trade with future, unborn generations.2 This benefit of social security as perceived by tax-paying workers is a robust feature. It persists even if an additional policy instrument is available that depresses savings without transferring resources to the elderly. Voters internalize only those general-equilibrium effects of transfers that materialize during their lifetimes—negative consequences borne by subsequent cohorts (due to lower capital accumulation) remain unaccounted for. The pro-transfer bias mentioned before therefore reflects the fact that the political process partly shifts the cost of social security to the future. In contrast, a Ramsey government with “dynastic” welfare weights (i.e., with welfare weights reflecting households’ time preference and cohort sizes) internalizes all general-equilibrium effects. The social security tax rate implemented by such a government therefore tends to fall short of the tax rate in politico-economic equilibrium. Demographic change alters factor prices and changes the relative weight that the political process attaches to the interests of old and young voters. Under standard functional form assumptions, we are able to characterize the resulting transition dynamics of the economy in closed form. (This stands in sharp contrast to most of the literature which characterizes politico-economic equilibria numerically. When we relax the functional form assumptions and thus, have to resort to numerical solutions, we find our central results to be robust.) The model predicts a slowdown of population growth to be associated with (i) higher social security tax rates, (ii) a rising share of pensions in GDP, (iii) but eventually lower social security benefits per retiree. The endogenous response of tax rates to demographic shocks also affects the allocation of consumption risk in the economy. In fact, (iv) old-age consumption risk is lower in politico-economic equilibrium than in a situation where tax rates are constant. These effects are sizeable. When calibrated to match stylized features of the U.S. economy, the closed-form solutions of the model suggest that (v) social security tax rates will gradually increase to around 16%. Moreover, while the social security system will absorb a growing share of GDP, (vi) the importance of other policy instruments with a distortive effect on labor supply will decline. As a result, (vii) labor supply will continue to rise until leveling off in two decades or so. These findings have important implications for the debate about social security reform. Participants in that debate have identified several policies that would restore solvency of the U.S. social security system. According to the 2006 Annual Report of the Social Security Board of Trustees,3 for example, “the projected infinite horizon shortfall [of social security] could be eliminated with an immediate increase in the combined payroll tax rate from 12.4% to about 16.1%” (p. 55). According to the well-publicized reform proposal by Diamond and Orszag (2005), to name another example, tax rates could gradually be increased to around 15.4% and benefits cut by up to 9%.4 The model developed in the present paper predicts that equilibrium tax rates will gradually increase to a level of around 16%. In light of the two above-mentioned scenarios, this implies that benefits will be cut, but most likely by less than 9%. Participants in the reform debate have also discussed the effect of social security tax rates on labor supply. Some observers have voiced fears that further payroll tax hikes might trigger large deadweight losses, rendering the option of raising social security taxes essentially infeasible.5 The model developed in the present paper partially refutes this view. It shows that the link between social security tax rates and labor supply distortions must not be seen in isolation but in the context of a wider set of policy instruments. More specifically, it predicts that—because of deadweight losses—higher social security tax rates will go hand in hand with fewer other sources of labor market distortions; this will have a positive net effect on labor supply. Our work extends a growing literature on dynamic politico-economic equilibrium, with voters sequentially choosing their preferred policies under rational expectations about the effect on future equilibrium outcomes (see, e.g., Krusell et al., 1997 and Hassler et al., 2003). Moreover, it relates to an extensive literature on the sources of political support for intergenerational transfers. Typical explanations in this literature rely on altruism or commitment (see, e.g., Cukierman and Meltzer, 1989, Hansson and Stuart, 1989, Conesa and Krueger, 1999, Tabellini, 2000 and Persson and Tabellini, 2002). Alternatively, they view the political process as representing the interests of a tax-paying median voter who fears that the provision of future benefits hinges on the provision of current ones. According to this view, the link between current and future benefits arises because successive median-voters coordinate on a sufficiently effective trigger strategy (as in Bohn, 1999; Cooley and Soares, 1999; Boldrin and Rustichini, 2000; or Rangel, 2003), or on the self-fulfilling expectation that higher savings, due to lower current social security contributions, trigger a benefit cut (as in Forni, 2005). Our approach differs from these models. It does not rely on altruism, commitment, expected punishments, or an infinite horizon, nor does it restrict policy choices to be binary or population growth to be sufficiently high (to render the economy dynamically inefficient, as do some previous models). Moreover, our approach generates closed-form solutions; fully characterizes the transition dynamics; and introduces multiple policy instruments, allowing to investigate the robustness of the political support for social security (on which the median-voter literature mostly remains silent). These advantages are largely the consequence of the probabilistic-voting assumption which provides a means to capture gradual changes in the support for social security even in a stark two-period-lived overlapping-generations environment.6 In median-voter models featuring a few overlapping generations of homogeneous agents, in contrast, such gradual changes cannot be captured since ageing of plausible magnitude does not alter the identity of the median-voter.7 As a consequence, demographic change in those models induces policy adjustments only to the extent that general equilibrium effects of ageing alter the median-voter's preferred policy. But these general equilibrium implications typically work in the direction of reducing equilibrium tax rates (Cooley and Soares, 1999 and Forni, 2005), which is counterfactual in light of the historical experience.8 The paper also relates to the literature on quantitative policy analysis in large-scale overlapping generations models as pioneered by Auerbach and Kotlikoff (1987) who analyze the effect of social security on labor supply and capital accumulation. Subsequent contributions to this literature consider the welfare effects of social security arrangements when households face financial constraints and are subject to privately uninsurable risks (e.g., Hubbard and Judd, 1987 and İmrohoroǧlu et al., 1995). Galasso (1999) introduces endogenous policy choice in a large-scale overlapping generations model, relying on the median-voter framework and numerical steady-state comparisons. The large number of overlapping generations in Galasso's (1999) model implies that demographic ageing of plausible magnitude raises the age of the median voter, giving rise to higher equilibrium tax rates. Since we depart from the median-voter assumption, we are able to generate qualitatively similar predictions (as well as novel ones) within a more tractable framework featuring much fewer overlapping cohorts. The remainder of the paper is structured as follows: 2 and 3 present the model and characterize politico-economic equilibrium. For ease of exposition, we first consider a simplified setup with inelastic labor supply and a single policy instrument (Section 2) before solving the full-fledged model in Section 3. Section 4 analyzes the Ramsey benchmark, and Section 5 contains the quantitative results. Section 6 concludes.
نتیجه گیری انگلیسی
We have argued that the political support for intergenerational transfers reflects the interests of all voters rather than just a working median voter. The micro-political foundation for that view—probabilistic voting—is natural and has plausible implications. Introducing the probabilistic-voting assumption in the standard Diamond (1965) model preserves that model's tractability and delivers intuitive and novel results in a strikingly transparent fashion. Predictions of the model accord well with frequently expressed notions in the social security debate. For example, in response to population ageing, social security benefits are predicted to be cut, sooner or later. Such cuts do not herald the dismantling of pay-as-you-go social security systems, however. To the contrary, social security taxes and the GDP-share of the social security system will continue to grow. The model also gives meaning to the notion that social security constitutes a burden for future generations—even if these generations are not committed to honor existing social security promises; for in the model, political competition resolves the conflict between old and young voters by shifting some of the cost of the social security system to future generations. As a consequence, intergenerational transfers are too large, relative to a system balancing the interests of all generations. We have assumed that the government's budget is balanced in each period. Relaxing this assumption introduces two further political choices in each period, one regarding the issuance of new debt, the other regarding the default rate on maturing debt. The debt issuance choice is constrained by the fact that agents investing in government debt foresee the possibility of a (partial) default by future political decision makers. Debt can therefore only be issued to the extent that a political incentive compatibility constraint is satisfied. But this implies that allowing for the issuance of defaultable government debt does not affect the politico-economic equilibrium. Since voters only care about the real allocation, the economic equivalence between social security and debt-plus-tax policies extends to the political sphere.44 Since the model is very tractable, it lends itself to a variety of interesting extensions. We have analyzed one, central extension with endogenous labor supply, tax distortions, and multiple policy instruments. This extension proved the predicted support for social security to be a robust feature, a novel result in the literature. Another extension, due to Song (2005), features intragenerational heterogeneity and analyzes the interaction between social security transfers and wealth inequality. Song (2005) finds higher inequality to be associated with higher equilibrium social security tax rates if social security redistributes within cohorts. Further extensions could address questions relating to the interaction between social security and other policies or household choices. We leave these topics for future research.