Redistributive Effects of Pension Schemes If Individuals Differ By Life Expectancy

Miguel Sanchez-Romero, Wittgenstein Centre for Demography and Global Human Capital (IIASA, VID/OEAW, WU)
Alexia Prskawetz, Wittgenstein Centre for Demography and Global Human Capital (IIASA, VID/OEAW, WU) and Vienna University of Technology (TU Wien)
Ronald D. Lee, University of California, Berkeley

We set up a general equilibrium model in which heterogeneous individuals by level of learning ability and life expectancy optimally decide about their consumption/saving path, the length of schooling, and their labor supply, both at the intensive margin (hours worked) and at the extensive margin (retirement age). To account for the marginal effects that mortality and health may have on the length of schooling and retirement age, we follow Sánchez-Romero et al. (2016) and Bloom et al. (2014). Moreover, the model is implemented so as to simultaneously analyze the effects that any pension system (Funded/PAYG with Defined Contribution or Defined Benefit) might have on the optimal decisions, even though the model will be calibrated to the US case.

Motivation

Many countries have experiencedan impressive increase in life expectancy accompanied by a rise in the numberof healthy life years. The improvements in morbidity and mortality have beenaccompanied by decreases in fertility. These demographic developments will leadto an increase in the proportion of retirees per worker, thereby putting pressureon current pension systems. Reforms of the social security system, such aspostponing the retirement age or reducing the generosity of the system, are inevitable.However, within countries, the difference in life expectancy between the highand low socioeconomic groups has also widened in recent decades. For instance,in the US, this difference may be as large as 10-14 years between those of highand low socioeconomic status. Given that pension systems have a strong agecomponent, since contributions are paid during working years while benefits arereceived at retirement, ignoring this heterogeneity might jeopardize any reformas the “ex ante” actuarial fairness and the progressivity of the system becomes“ex ante” highly regressive.

The proposed reforms of thepension system need to take into account that individual ageing isheterogeneous across socioeconomic groups. Moreover, it is necessary toinvestigate how alternative pension reforms impact   the decisions ofheterogeneous individuals by socioeconomic status and whether the reforms arecapable of restoring the “ex ante” actuarial fairness and progressivity of thesystem across socioeconomic groups. For such studies we need models of thepension system that account for the behavioral response of individuals thatdiffer by their life expectancy. Sketch of the model

We set up a general equilibriummodel in which heterogeneous individuals by level of learning ability and lifeexpectancy optimally decide about their consumption/saving path, the length ofschooling, and their labor supply, both at the intensive margin (hours worked)and at the extensive margin (retirement age). To account for the marginaleffects that mortality and health may have on the length of schooling andretirement age, we follow Sánchez-Romero et al. (2016) and Bloom et al. (2014).Moreover, the model is implemented so as to simultaneously analyze the effects thatany pension system (Funded/PAYG with Defined Contribution or Defined Benefit) mighthave on the optimal decisions, even though the model will be calibrated to theUS case. Simulations and discussion

To evaluate the impact of thepension reforms on old-age inequality, we calculate the present, survivalweighted value of lifetime pension benefits across different socioeconomicgroups and birth cohorts. Moreover, we investigate the effect that alternativepension reforms have on the length of schooling, retirement, and labor supply.To understand the role of pension systems for labor force, we provide estimatesof the implicit tax/subsidy generated by each pension system at the intensivemargin and also at the extensive margin (a la Gruber and Wise).

First simulation results of the presentvalue of lifetime pension benefits at age 50 by income quintile are presentedin Figure 1 for four different PAYG pension systems: (i) Notional DefinedContribution (NDC) with an average cohort-life table (DC-AM), (ii) NDC with acohort-life table for each income quintile (DC-DM), (iii) PAYG-Defined Benefitsystem with a constant replacement rate (DB-NP), and (iv) the US pension system(DB-P). Figure 1 shows that although the US pension system is the mostprogressive of the four pension systems for the 1930 birth cohort, theincreasing longevity gap across income quintiles (c.f., cohorts 1930 and 1960) leadsthe US pension system to not redistribute across income quintiles as well as anotional defined contribution system that uses specific cohort-life tables byincome quintiles. This result suggests that reforms of the US pension system shouldtake into account the increasing longevity gap across socioeconomic groups.

Figure 1. Present Value of Lifetime Pension Benefitsby Income Quintile and PAYG Pension System, Under Mortality Regimes of the 1930and 1960 US Birth Cohorts for Males in 1000''s of US$

Source: Lee and Sanchez-Romero (2017). Note: Risk-free marketdiscount rate = 3%. Results for cohort 1960 are detrended of labor productivitygrowth. Pay-as-you-go pension systems: Defined Contribution system using theaverage cohort-specific life table (DC-AM), Defined Contribution system usingthe cohort-specific life table by income-quintile (DC-DM), Defined Benefitsystem with a constant replacement rate (DB-NP), and Defined Benefit systemwith an “ex ante” progressive replacement rate (DB-P).

 

Presented in Session 1149: Economics, Human Capital, and Labour Markets