Understanding the Determination of Severance Pay: Mandates, Bargaining, and Unions*

• Author: Markus Poschke,Samuel Danthine,Stéphane Auray
• DOI: http://doi.org/10.1111/sjoe.12367
• Date: 01 July 2020
• Published date: 01 July 2020

Scand. J. of Economics 122(3), 1073–1111, 2020

DOI: 10.1111/sjoe.12367

Understanding the Determination of

Severance Pay: Mandates, Bargaining,

and Unions*

St´ephane Auray†

CREST-ENSAI,FR-35712 Bruz Cedex, France

stephane.auray@ensai.fr

Samuel Danthine

ENSAI, FR-35712 Bruz Cedex, France

danthine.samuel@gmail.com

Markus Poschke‡

McGill University,Montr ´eal, QC, H3A 2T7, Canada

markus.poschke@mcgill.ca

Abstract

A substantial share of severancepayments derives from private contracts or collectiveagreements.

In this paper, we study the determination of these payments. We analyze joint bargaining

over wages and severance payments in a search-and-matching model with risk-averse workers.

Individualbargaining results in levels of severance pay that providefull insurance, but also depend

on unemployment benefits and job-finding rates. Unions also choose full insurance. Because

their higher wage demands reduce job creation, this requires higher severancepay. Severance pay

observed in eight European countries, to which we calibrate the model, lies between predictions

from the bargaining and union scenarios.

Keywords: Bargaining; severance pay; unemploymentinsurance; unions

JEL classification:E24; J32; J33; J64; J65

I. Introduction

Job termination can have important welfare consequences for workers. As

a result, arrangements for severance pay exist in many countries around the

*We would like to thank Nicolaas Vermandel (Laga) for providing us with some data about

severance pay measures. We also thank the participants at various seminars and conferences

for helpful comments, and we are especially grateful to Bruno Decreuse, our discussant at the

workshop on “Unemployment, Wage Inequality and Labor Market Policy” in Konstanz. SD

gratefully acknowledges financial support from the Ministerio de Ciencia e Innovaci´on under

project ECO2011-29355, and from Junta de Andalucia under project SEJ4941. MP gratefully

acknowledges financial support from de Fonds de recherche sur la soci´et´e et la culture under

project 133431.

†Also affiliated with Universit´e du Littoral Cˆote d’Opale (ULCO).

‡Also affiliated with CIREQ and IZA.

C

The editors of The Scandinavian Journal of Economics 2019.

1074 On unions and severance pay

world. These arrangements differ substantially. In some countries, such as

Canada and Spain, severance pay is government-mandated. In others, such

as Japan and the United States, it is reached through private bargaining.

In yet other countries, it is partly government-mandated and partly reached

through bargaining (e.g., Belgium, France, Germany, and Italy). In countries

where arrangements for severance pay are reached privately, this can

occur through individual contractual arrangements, or through collective

bargaining with unions.1

The existing literature on severance pay has mainly considered

government-mandated severance pay.2However, severance pay reached

through collective or private agreements is quantitatively important. For

instance, using data from Massachusetts in the United States in which

86 percent of workers were covered by a severance pay agreement,

Kodrzycki (1998) shows that a typical arrangement features severance pay

of one week’s wage per year of service. Assuming a replacement rate of

unemployment insurance (UI) of 50 percent, this implies that total severance

pay receipts are higher than maximum potential UI receipts for workers

with more than 13 years of tenure. On average, severance pay amounts to

43 percent of maximum available UI receipts for displaced workers in the

sample of Kodrzycki (1998).

If privately reached severance pay agreements are important, this raises

the question of why publicly mandated regimes exist, and how the levels

of severance pay they impose compare with the ones that would come

out of private arrangements. Therefore, the main aim of this paper is a

theoretical and quantitative analysis of private bargaining over severance

pay. We conduct this analysis in a standard Diamond–Mortensen–Pissarides

(DMP) model with bargaining not only over wages, but also over severance

pay. This set-up then allows us to assess how counterfactual, simulated

estimates of privately bargained severance pay compare with observed,

government-mandated levels.

As already alluded to above, collective bargaining cannot be neglected

when analyzing the determination of severance pay. The evidence presented

in Parsons (2005a,b,c) shows very clearly that severance pay arrangements

differ substantially by unionization status of the employee. There is also

evidence that severance payments are indeed negotiated by unions (Millward

et al., 1992), and that, because of the complexity of employment protection

legislation, unions are able to obtain higher firing costs for their members

(Colonna, 2008a,b). Thus, an additional contribution of this paper is to

1See Holzmann et al. (2012), particularly Annex B, for a classification of countries by type of

severance pay arrangement, and see Laga (2012) for some country specific details.

2An important exception, which we discuss below, is Fella and Tyson (2013).

C

The editors of The Scandinavian Journal of Economics 2019.

S. Auray, S. Danthine, and M. Poschke 1075

analyze union behavior in this context. This turns out to be important for

understanding the arrangements in several countries.

Before delving into the theoretical analysis, we discuss evidence on

severance pay in Europe (Section II). In doing so, we do not limit

ourselves to popular measures, such as those from the World Bank and

the OECD, which only cover legally mandated severance pay. Instead, we

focus particularly on negotiated components of severance pay resulting

from bargaining. Of course, privately bargained severance pay might not be

observed in countries where legally mandated levels exceed those that would

result from private negotiation. For these cases, our model can provide

an indication of which counterfactual arrangements could prevail in the

absence of legal provisions.

To analyze bargaining over severance pay, we build a matching model in

the manner of Mortensen–Pissarides with endogenous job destruction and

risk-averse workers, and we add (i) unions as in Delacroix (2006), and (ii)

bargaining over wages and severance payments as in Booth (1994, 1995). A

first theoretical result shows that a worker and a firm bargaining over both

wages and severance pay opt for a level of severance pay that gives the

worker full insurance. This arises because the risk-neutral firm can insure

the risk-averse worker. The level of severance pay required for this decreases

with the UI replacement rate and the job-finding rate. Expressed relative to

average completed tenure, it also depends on the job-destruction rate.

A second theoretical result shows that a monopolistic union that chooses

wages and severance pay to maximize the value of employment, taking

firms’ reactions as given, also chooses full insurance. How much surplus

the union can extract is limited by the equilibrium reaction of job creation

to higher wages and severance pay.

In order to provide quantitative results, we calibrate the model to a set of

eight continental European economies that feature varying levels and types

of arrangements for severance pay. We then first use the calibrated model

to compare outcomes from bargaining to union behavior, taking severance

pay as given. We then perform a number of quantitative counterfactual

experiments on severance pay using the calibrated model. These can be

grouped into two sets of exercises, one aiming to understand the effect of

varying severance pay, and another aiming to provide an indication of what

levels of severance pay would be bargained privately. This second set can

serve as a benchmark to which legislated levels of severance pay can be

compared.

The first set of exercises indicates that exogenously eliminating

mandated severance pay increases job destruction but also increases job

creation, and thus has an ambiguous effect on unemployment. This is

in line with results from the literature, in particular Blanchard (2000)

and Ljungqvist (2002). In calibrated economies, we find that the change

C

The editors of The Scandinavian Journal of Economics 2019.