Elsevier

Labour Economics

Volume 15, Issue 1, February 2008, Pages 68-77
Labour Economics

Is there a social security tax wedge?

https://doi.org/10.1016/j.labeco.2006.11.003Get rights and content

Abstract

A Beveridgean pension scheme discourages labour, and is bad for effciency. A Bismarckian one may encourage labour, and be good for effciency. In any case, the same contribution level will discourage labour and reduce effciency less if the scheme is Bismarckian, than if it is Beveridgean.

Introduction

The political discussion on the effects of pension policy appears to take it for granted that a pension contribution is a tax on labour, and will consequently reduce employment. Indeed, a series of empirical studies finds a negative effect of pension contributions on either employment or labour participation; see, for example, Alesina and Perotti (1997), Scarpetta (1996), Tullio (1987). The assumption is justified, and the empirical finding unsurprising, in countries that have given themselves a Beveridgean pension system, because individual pension benefits are then unrelated to individual contributions, and the latter are thus effectively an earmarked tax (the social security tax). Not so, however, in countries where the pension system is essentially Bismarckian, and thus characterized by a close link between benefits and contributions.1 In such countries, pension contributions are a mandatory form of saving. There will be an element of tax only if these contributions are higher than would be required to obtain the same amount of retirement income by other means. If they are lower, labour is being subsidized.

The concept of an implicit pension tax dates back to Lüdeke (1988) and Sinn (1990). More recently, Murphy and Welch (1998) and Orszag and Stiglitz (2000) also have come round to the idea. This theoretical insight has sparked-off a number of empirical studies aimed at measuring the tax element in pension contributions; see, for example, Börsch-Supan and Reil-Held (2001), and Fenge and Werding (2004). Disney (2004) takes the empirical analysis further by estimating the effects of the tax component of pension contributions on labour participation by age and sex. He finds that, if composition is not controlled for, pension contributions reduce participation as in the earlier empirical studies mentioned. If it is, pension contributions have either no significant effect, or a significantly positive one. The tax component has a significantly negative effect on female participation, but little or no effect on male participation.

If the age of retirement is an object of choice, the existence of a public pension system may affect both the length of a person's working life, and the amount of labour that this person will supply over a working life of any given duration. Although there are analogies between the two decisions, the issues involved, and the way of dealing with them, are however quite different.2 The present paper has the limited objective of analytically deriving the labour distortion associated with compulsory participation in a public pension scheme, assuming that the age of retirement is fixed.

We find that a Beveridgean scheme will always impose an implicit tax on labour. By contrast, a Bismarckian scheme may tax some workers, and subsidize other. An implicit pension tax tends to discourage labour, an implicit pension subsidy to encourage it. We also find that credit market imperfection does not affect the size of the labour distortion caused by a Beveridgean pension system. By contrast, if the system is Bismarckian, credit rationing reinforces the distortionary effect of any implicit pension tax, and weakens that of any implicit pension subsidy. If it is actuarially fair, a Bismarckian system can distort labour behaviour only if the worker is credit rationed. These interactions do not appear to have been pointed out before.

These findings have important policy implications. Cutting back a Beveridgean system will definitely encourage labour, and improve efficiency, but will also raise inequality. Cutting back a Bismarckian one may do the very opposite. A country with a Bismarckian pension system should thus be double careful before reaching for the axe.

2 Individual decisions in the absence of a public pension system, 3 Stylized pension systems, 4 Labour implications of alternative pension systems of the paper set out the theory. Section 5 briefly reviews the empirical evidence, and suggests ways of going forward on that front. Section 6 summarizes and discusses the findings.

Section snippets

Individual decisions in the absence of a public pension system

Let li denote the labour, c1i the working-age consumption, and c2i the retirement-age consumption of agent i. His utility is assumed to be given byUi=u1(c1iv(li))+u2(c2i),where v (li) is the money-equivalent of the disutility of labour. The functions ut (.) are assumed increasing and concave, and the function v (.) increasing and convex. The agent chooses (c1i, c2i, li, si) to maximize (1), subject toc1i+si=wili,c2i=sirandsibi,where si denotes i's saving, bi his credit ration (positive or

Stylized pension systems

Let us now introduce a compulsory pension system. This will reduce i's disposable income by the contribution θi while he is of working age, and increase it by the benefit ηi when he is retired. The pension contribution is typically an increasing function of labour income, such that the marginal contribution rate is always less than 100 percent,θi=θ(wili),0<θ(wili)<1.

If the system is of the Beveridgean type, individual benefits may be the same for everyone, or vary with certain personal

Labour implications of alternative pension systems

In the presence of a pension system, the budget constraints (2) and (3) become, respectively,c1i+si=wiliθiandc2i=sir+ηi.

If the system is Beveridgean, the agent supplies labour to the point where the money equivalent of the marginal disutility of labour equals the marginal increase in take-home pay,v(li)=[1θ(wili)]wi.

Comparing (14) with (5), it is clear that the scheme will introduce a wedge between the wage rate and the marginal take-home pay even if the agent happens to be fairly treated.5

Empirical evidence

Börsch-Supan and Reil-Held (2001), and Fenge and Werding (2004) calculate that, except at their very inception,7 all pension systems impose an implicit tax on the average participant. Even so, it may be the case that some categories get an implicit subsidy, paid for by other categories, even

Policy implications

The simple theoretical model outlined in the present paper predicts that public pensions distort labour if the system is Beveridgean, but not necessarily if the system is Bismarckian. In a Beveridgean scheme, individual pension benefits are in fact independent of individual contributions, and the marginal return to labour coincides with the marginal take-home pay. The whole of the pension contribution is consequently a tax on labour. In a Bismarckian system, by contrast, the marginal return to

References (17)

There are more references available in the full text version of this article.

Cited by (17)

View all citing articles on Scopus

The paper has benefited from comments by two anonymous referees, and by the editor.

View full text