Life Expectancy Heterogeneity and the Political Support for Collective Annuities

• Author: Helmuth Cremer,Philippe Donder
• DOI: http://doi.org/10.1111/sjoe.12153
• Published date: 01 July 2016
• Date: 01 July 2016

Scand. J. of Economics 118(3), 594–615, 2016

DOI: 10.1111/sjoe.12153

Life Expectancy Heterogeneity and the

Political Support for Collective Annuities∗

Helmuth Cremer†

Toulouse School of Economics, FR-31015 Toulouse, France

helmuth.cremer@tse-fr.eu

Philippe De Donder‡

Toulouse School of Economics, FR-31015 Toulouse, France

philippe.dedonder@tse-fr.eu

Abstract

Individuals, differing in productivity and life expectancy, vote over the size and type of a

collective annuity. Its type is represented by the fraction of the contributive (Bismarckian)

component (based on the worker’s past earnings) as opposed to the non-contributive (Bev-

eridgean) part (based on average contribution). The equilibrium collective annuity is either

a large mostly Bismarckian program, a smaller pure Beveridgean one (in accordance with

empirical evidence), or nil. A larger correlation between longevity and productivity, or a

larger average life expectancy, both make the equilibrium collective annuity program more

Beveridgean, although at the expense of its size.

Keywords: Collective annuity; generosity; Kramer–Shepsle structure-induced equilibrium;

longevity; pay-as-you-go pensions; redistributiveness

JEL classification:D78; H55

I. Introduction

A sizeable body of literature deals with the political determination of the

characteristics of a public pay-as-you-go pension system. The influential

paper by Browning (1975) assumed that the only heterogeneity between

voters is their age. Subsequent papers (such as Casamatta et al., 2000a)

have enriched this approach by assuming that agents also differ in income

or in productivity. This richer set of individual traits has allowed these

papers to study the determination of both the size of the pension system

†IDEI, University of Toulouse and Institut universitaire de France.

‡GREMAQ-CNRS and IDEI.

∗A previous version of this paper has circulated under the title “Longevity, Annuities and

the Political Support for Public Pensions”. We thank P. Belan, B. Heijdra, and L. Meijdam

for their comments. We also thank the editor and the referees for their constructive and

extremely helpful comments and suggestions. Financial support from the Chaire March´

edes

risques et cr´

eation de valeur of the FdR/SCOR and from Netspar through a research grant

is gratefully acknowledged.

CThe editors of The Scandinavian Journal of Economics 2015.

H. Cremer and P. De Donder 595

and of its redistributiveness across income levels. As for the latter, the

literature (surveyed by Galasso and Profeta, 2002) has contrasted so-called

Bismarckian systems, where the pension benefit is proportional to the

individual contribution, with Beveridgean systems, where the benefit is

based on the average contribution.

The main stylized fact in this domain is that Bismarckian systems tend

to be larger (as measured by either the contribution rate or the share of pub-

lic pensions in GDP) than Beveridgean systems; see Table 1 in Casamatta

et al. (2000b), and also Disney (2004) or Conde-Ruiz and Profeta (2007).

Very few papers have tried to explain this correlation. Both Casamatta

et al. (2000b) and Koethenbuerger et al.(2008) study the determination by

the median voter of the size of the pension system as a function of the

system’s type, measured by the (exogenous) relative importance of the con-

tributive component, dubbed the Bismarckian parameter. Casamatta et al.

(2000b) assume exogenous wage income and obtain an ambiguous impact

of the Bismarckian parameter on the equilibrium size of the pension sys-

tem. Koethenbuerger et al. (2008) introduce endogenous labor supply, and

find that larger labor supply distortions, generated by the flat-rate Bev-

eridgean pension system versus the earnings-related Bismarckian system,

can explain why more redistributive systems are smaller. Conde-Ruiz and

Profeta (2007) study the simultaneous determination of the size and type

of pension systems. In their model, agents differ in age and income, and

in ability to invest in the capital market. With only three income groups,

a small Beveridgean system is supported by low-income agents, who gain

from its redistributive feature, and high-income individuals, who seek to

minimize their tax contribution and to invest their resources in a private

scheme. Middle income individuals, instead, favor a large Bismarckian sys-

tem. The degree of inequality in earnings and the level of capital market

returns determine which type of equilibrium emerges.

An important dimension of heterogeneity among voters, that is, longevity,

which might play a critical role in the determination of the pension system,

is absent from these studies. It is well known empirically that people of the

same age differ in life expectancy. Moreover, life expectancy is positively

correlated with income or wealth, as shown by Deaton and Paxon (1999)

for the US, by Attanasio and Emerson (2001) for the UK, and by Reil-Held

(2000) for Germany. Average life expectancy has been increasing in most

countries for at least half a century. However, these increases have not

been shared equally everywhere. For instance, in the US, the average male

life expectancy at 65 has increased from 15 to 16.1 years between 1986

and 2006 for individuals in the bottom half of the earnings distribution,

but from 16.5 to 21.5 years in the top half of the earnings distribution

(Waldron, 2007). It is thus important to assess the impact of such variations

on pension programs.

CThe editors of The Scandinavian Journal of Economics 2015.