اصلاح امنیت اجتماعی با اولویت های کنترل خود
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|24417||2011||14 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Public Economics, Volume 95, Issues 7–8, August 2011, Pages 886–899
This paper analyzes a fully funded social security system under the assumption that agents face temptation issues. Agents are required to save through individually managed Personal Security Accounts without, and with mandatory annuitization. When the analysis is restricted to CRRA preferences our results are congruent with the literature in indicating that the complete elimination of social security is among the reform scenarios that maximize welfare. However, when self control preferences are introduced, and as the intensity of self control becomes progressively more severe the “social security elimination” scenario loses ground very rapidly. In fact, in the case of relatively severe temptation the elimination of social security becomes the least desirable alternative. Under the light of the above findings, any reform proposal regarding the social security system should consider departures from standard preferences to preference specifications suitable for dealing with preference reversals.
There is an abundance of studies related to the importance of social security and its impact on welfare. The primary reason for this is its dramatically growing scale which has triggered a renewed academic debate regarding the optimal allocation of the available resources. This controversy stems from the huge monetary burden that the mere presence and administration of a social security system entails for the society and the associated budget implications: Old age, disability, unemployment and health insurance policies have evolved into the most expensive items on government budgets. Most of the studies that seem to emerge as a direct or implicit offspring of this debate focus on the welfare implications of alternative social security schemes in an economy. In the very core of this debate one can clearly identify the dilemma between an “unfunded” (Pay-As-You-Go) versus a “funded” social security system. In an unfunded system, resources are transferred statically from the working population to the concurrent retirees (inter-generational transfers). In contrast, a funded system prescribes a dynamic allocation of resources within the same generation (inter-temporal within the same generation transfers). While the implementation of both systems relies on an external institution (e.g. government), their different logic and mechanics eventually induce entirely different risk-sharing properties as well as savings incentives. Therefore, their welfare implications may significantly diverge because of this difference. Population aging as a result of the declining population growth rate and decreasing birth rate has challenged enormously the sustainability of a PAYG system and called for a minimization of the fiscal burden through tax reforms and benefits restructuring. As a result, there are numerous studies suggesting alternative institutional arrangements that could be more robust to adverse demographic shocks. However, as much as converting an unfunded system to a fully (or partially) funded one may seem a plausible solution, in most cases the transition costs associated with such a reform make it prohibitively costly ( DeNardi et al., 1999). The welfare implications of social security are well identified in the relevant literature.2 Several studies (e.g. Storesletten et al., 1999) comparing different social security systems typically compare welfare across alternative steady states, each corresponding to a stationary equilibrium with a different social security system. Focusing only on unfunded social security, Imrohoroglu et al. (1995) emphasize the detrimental effects that such an arrangement has to the overall welfare in a country. However, all the above studies ignore alternative preference specifications that may be binding in several cases: Imrohoroglu et al., 2003 and Fehr et al., 2008 use time-inconsistent preferences while Kumru and Thanopoulos, 2008 and Bucciol, 2008 use self-control preferences to highlight that in a context of unfunded social security welfare may be critically affected by the preference specification.3 In a recent paper, Fehr and Kindermann (2009) study the implications of individual retirement accounts (IRA) on economic aggregates and welfare in the context of a general equilibrium model populated by either rational agents or agents with time-inconsistent preferences. In this study, we would like to quantitatively assess the welfare implications of the reform of the current unfunded social security system into a partially funded or fully privatized one, under the assumption that agents face self-control problems. We proceed by assessing in terms of welfare a hybrid (partially funded) and a fully funded social security system under the alternative hypotheses of self-control or CRRA preferences. The apparatus by means of which we model departures from unfunded social security is a Personal Security Account (PSA). Within that class of “funded” models, we investigate two competing scenarios involving PSAs: one without annuitization and an alternative one that prescribes a mandatory annuitization of retirement benefits. 4 More precisely we compare and contrast the implications of five different social security reforms: • The first reform proposal (PSA) postulates the substitution of the current social security system by a two-tier scheme: a universal PAYG-financed basic pension combined with a Personal Security Account that does not require annuitization of benefits at retirement. • The second reform proposal (PSAA) is similar to the first one except that it prescribes a mandatory annuitization of the funds accumulated in PSA accounts. • The third reform proposal (PSAwoFT) postulates the removal of the PAYG-financed basic pension plan existing in the first reform proposal. • Similarly, the fourth reform proposal (PSAAwoFT), suggests the removal of the PAYG-financed basic pension plan existing in the second reform proposal. • Finally the fifth proposal (elimination) suggests the complete removal of the social security system. Moreover, in order to capture our agents' temptation towards current consumption, our model economies make use of the preference structure pioneered by Strotz, 1956 and Phelps and Pollak, 1968 and further elaborated by Gul and Pesendorfer (2004) to model self-control issues. Gul and Pesendorfer (2004) identified a particular class of utility functions that provides a time-consistent model suitable for addressing the preference reversals that motivated the time inconsistency literature. The key theme here is that self-control preferences assume that agents maximize a utility function that is a ‘compromise’ between the standard utility (or ‘commitment’ utility) and a ‘temptation’ utility. The conflicting ways by which agents derive utility in this setting, is the device through which the trade-off between the temptation to consume on the one hand, and the long-run self interest of the agent on the other is captured. The main benefit is that self-control preferences remain perfectly time-consistent and, contrary to time-inconsistent preferences, allow agents in our model to commit. With the exception of the aforementioned difference in the specification of preferences, our model specification follows that of Imrohoroglu et al., 2003 and Fuster et al., 2005. Furthermore, our economic environment features uninsurable individual income shocks, borrowing constraints and missing annuity markets. We aim to contribute to the debate on the reform of social security. Our augmented model allows us to look at the welfare gains or losses due to the reforms from a different angle. In particular, it allows us to assess the welfare-enhancing potential of mandatory savings versus mandatory annuitization of accumulated PSA wealth at retirement. For the sake of comparability of our results, the particular specification of those alternative policies is purposely chosen to match the proposals analyzed in the literature (Storesletten et al., 1999 and Fuster et al., 2005), as well as those featuring in the reform recommendations made by the 1997 Advisory Council on Social Security. It is nevertheless critical to address a question that lies at the very core of our line of research, namely, why models of social security with time inconsistent or “temptation” preferences are relevant in the first place. Several factors weigh-in in favor of the relevance of these models. A first one relates to theoretical completeness: a change in the preference structure enhances our understanding of the mechanics of similar models in the literature by providing an additional channel through which capital accumulation is distorted. An additional factor is the need for comprehensive policy evaluation: an augmented preference structure is essential for providing a comprehensive comparison framework for policy makers in their evaluation of various proposals. Thirdly, empirical relevance: there is sound empirical and experimental evidence that individuals suffer from self-control problems. Frederick et al. (2002) provide an overview of experimental studies documenting that individuals indeed exhibit bias toward immediate gratification. Huang et al., 2007 and Bucciol, forthcoming study the empirical relevance of self-control preferences using household-level data from the Consumer Expenditure Survey. Their estimates support the presence of temptation.5 Moreover, in a recent paper Fang and Silverman (2009) empirically identify the existence of time-inconsistency that stems from self-control problems, through the estimation of the structural parameters of a dynamic labor supply model.6 Finally, the literature documents that the existence of self-control problems affects fundamentally the economic decisions of individuals. Given that, social security programs might play a beneficial role in environments in which individuals suffer from self-control problems: Laibson et al. (1998) show that when individuals have time-inconsistent preferences a fully funded social security system induces higher savings and improves welfare. Furthermore, Imrohoroglu et al., 2003, Kumru and Thanopoulos, 2008 and Fehr et al., 2008 show that the presence of agents that are either slightly short-sighted or prone to current consumption changes markedly the welfare implications of the system. It is worth emphasizing that this paper differs substantially from Kumru and Thanopoulos, 2008 and Bucciol, 2008, in that the latter two confine their analysis in the implications of one and only type of institutional arrangement with regard to social security, namely a PAYG system. While both these studies share the stimulating intuition that the existence of PAYG social security can enhance welfare by reducing individuals' self-control costs, their scope for policy impact analysis remains rather limited in that they do not match alternative social security schemes against each other under the presence of self-control. Instead, in this paper we confront PAYG social security with various hybrid schemes such as funded and partially funded systems, under the assumption that agents have self-control preferences. In this respect, our work not only represents a significant step forward in furthering our understanding of the mechanics of an environment more pertinent to policy-making, 7 but it also extends, complements and offers a different angle of interpretation of the results of previous studies that analyze the aggregate and welfare implications of fully-funded programs, when individuals have standard preferences, time-inconsistent preferences, and altruistic concerns. 8 Our results in this paper highlight the important role that self control preferences play, especially with regard to the kind of reform that could generate the highest welfare. When the analysis is restricted to CRRA preferences our findings initially are congruent with those obtained in other studies (e.g., Fuster et al., 2005): we also find that the complete elimination of any social security ranks highly with regard to welfare maximization, although it turns out that one of our fully funded scenarios (not analyzed in Fuster et al., 2005), namely PSAAwoFT is the reform scenario that generates the highest welfare. When self-control preferences are introduced instead, we come across a drastic alteration of these results: the elimination of social security becomes the least desirable policy reform from a welfare maximization standpoint, i.e. PSAA and PSAAwoFT reforms give rise to larger welfare improvements under self-control preferences. Along the same lines, our research indicates that the clear-cut advantage of the elimination scenario under CRRA preferences (even if it is not the welfare maximizing alternative) fades away as self control becomes gradually more severe. Our robustness tests confirm this finding: in the case of relatively severe temptation the elimination of social security becomes the least desirable scenario. The above results are derived by using a temptation utility function that reflects both temptation toward current consumption and agents' willingness to trade future temptation consumption with current temptation consumption. In order to address a balanced growth (in)consistency issue that this particular functional form entails, we follow Conesa et al. (2009) and assume that aggregate technology is constant. An alternative way to comply with balanced growth would be to use the temptation function introduced by Krusell and Smith (2003) (as pointed out in Huang et al., 2007). However, the Krusell–Smith type of temptation function does not capture the trade-off between current and future temptation consumptions and as a result, it generates a weaker temptation effect. Our results with a Krusell–Smith type of temptation function show that the ranking of different policy reforms is very similar to what we observe when agents have CRRA preferences. Before proceeding with the rest of the paper, it would be useful to sketch the intuition behind our main findings9: temptation preferences imply that agents incur disutility should they refrain from spending their entire cash in hand, something that reflects the cost of self control. Social security “wealth”, on the other hand, cannot be spent without restraint, so theoretically this is a less painful way to ‘save’ for retirement. Social security “wealth” is not part of cash on hand and therefore there is no disutility from not being able to spend it. In other words, it serves as an externally imposed commitment device.10 Consequently, this is the underlying rationale behind the ability of social security to potentially improve welfare, even in a dynamically efficient economy. Under the light of the above findings, any reform proposal regarding the social security system should consider departures from standard preferences to preference specifications suitable for dealing with preference reversals. The remainder of the paper is organized as follows. Section 2 provides a concise introduction to self-control preferences and time inconsistency. We briefly present and compare the two theories and attempt to shed light on the different implications they have for the question at hand. Section 3 presents our model economy. In Section 4, we outline the parameter values. We present the results of our experiments and conduct sensitivity analysis in Section 5. Finally, Section 6 concludes.
نتیجه گیری انگلیسی
Population aging due to low birth and morbidity rates and the resulting expansion of social security benefits have prompted lively debates around the long-term viability of unfunded social security.41 Several reform proposals are being discussed for the US Economy and other industrialized countries, all of which converge to a common resultant: social security must be reformed in the direction of a funded rather than an unfunded system. In this paper we examine departures from a Pay-As-You-Go social security system towards hybrid systems consisting of two parts: A “defined benefits” component and a “defined contributions” one. This is implemented by means of annexing a(n) (individually managed) Personal Security Account to a universal (PAYG-based) pension system. In parallel, we also consider and evaluate two fully-funded reform scenarios that are constructed merely by shutting off the pay-as-you-go tier in each of our two-tier scenarios. We quantitatively assess the attractiveness (i.e. the welfare enhancing potential) of such partially and fully-funded schemes in an economy populated by agents that face self control issues. To this purpose, we use two different benchmarks: an economy featuring PAYG social security populated by self control agents ( Kumru and Thanopoulos, 2008) and an economy featuring funded (as in this paper) social security but populated by CRRA agents. Furthermore, we refine our analysis by investigating the relative appeal of two distinct partially-funded systems, namely, PSAs without annuitization and PSAs with mandatory annuitization. Finally, we assess the relative attractiveness of two fully-funded reform scenarios. Our analysis suggests that the availability of PSAs increases savings. This is principally true for CRRA agents. In addition, self-control agents seem to have their welfare maximized under a hybrid model featuring a pay-as-you-go tier and a PSA with mandatory annuitization (i.e. PSAA). However, it still remains ambiguous as to whether self-control agents are better off with a partially funded system of the analyzed hybrid structure that prescribes mandatory annuitization, or with a fully-funded system. It eventually depends on the intensity of the self-control problem. Our results in this paper highlight the important role that self control preferences play, especially with regard to the kind of reform that could generate the highest welfare. When the analysis is restricted to CRRA preferences our findings initially are congruent with those obtained in other studies (e.g., Fuster et al., 2005): we also find that the complete elimination of any social security ranks highly with regard to welfare maximization, although it turns out that one of our fully funded scenarios (not analyzed in Fuster et al., 2005), namely PSAAwoFT is the reform scenario that generates the highest welfare. When self-control preferences are introduced instead, we come across a drastic alteration of these results: the elimination of social security becomes the least desirable policy reform from a welfare maximization standpoint, i.e., PSAA and PSAAwoFT reforms give rise to larger welfare improvements under self-control preferences. Moreover, our analysis of the two fully-funded settings in addition to the partially funded reform scenarios reveals a very interesting aspect. Contrary to what has been documented in the literature so far, the relative appeal of the elimination scenario to CRRA agents, seems to be driven mainly by the zero payroll tax rather than the unnecessity of social insurance for CRRA agents: our results manifest that CRRA agents do value some form of social insurance. Overall, as we move from an economy populated by CRRA agents to an economy with self-control agents, we observe a complete reversal of the importance of PSAA relative to elimination, a trend that gets even more pronounced as self control problems become more severe. This result is fairly intuitive: Gul and Pesendorfer self-control preferences entail a (ex ante) desire for commitment. PSAA imposes mandatory savings and delays benefits, something that prevents self control agents from over-consuming before retirement; at the same time, the feature of annuitized benefits at retirement prevents the agents from over-consuming at the early stages after retirement. As a result, the plan spares the agents the cost of exerting self control throughout their entire lifetime and this results in higher welfare. This observation combined with the documented appeal of a fully-funded scenario with annuitized benefits to both CRRA and self-control agents hints to the increasing importance of the annuitization of benefits regardless of the assumptions on preferences. Our analysis with a Krusell–Smith type of temptation function may not be pertinent to capture the trade-off between current and future temptation consumptions but manages to capture the observed growth in the data. Our results show that the ranking of different policy reforms is very similar to what we observe when agents have CRRA preferences. An interesting implication for policy making is a “policy equivalence” result that obtains in this part of our analysis: when temptation is sufficiently weak, a lower forced savings rate under the Krusell–Smith temptation can give rise to a similar ranking of reform plans to the one induced by a higher forced savings rate, but in a strong temptation environment (under a power temptation function) instead. In conclusion, our research indicates that the incontestable advantage of the elimination scenario under CRRA preferences rapidly vanishes as self control is introduced and becomes progressively more severe. All of our experiments confirm this finding: in the case of relatively severe temptation the elimination of social security becomes the least desirable policy.