THE SIMPLE ANALYTICS OF JOB DISPLACEMENT INSURANCE

• Date: 01 June 2019
• Author: Donald O. Parsons
• Published date: 01 June 2019
• DOI: http://doi.org/10.1111/jori.12216

351

THE SIMPLE ANALYTICS OF JOB DISPLACEMENT INSURANCE

Donald O. Parsons

ABSTRACT

Job displacement is a threat to the earnings of long-tenured workers through

(1) unemployment spells and (2) reduced reemployment wages. Although

full insurance requires both unemployment benefits and wage insurance,

supply difficulties limit actual-loss insurance, and separation packages

include partial unemployment insurance and scheduled (fixed sum)

severance pay. The design of this two-dimensional package requires a

systems approach as well as a generalized replacement ratio measure of

adequacy. Job search moral hazard and layoff moral hazard (firing costs)

introduce serious contracting concerns. Economic theory provides a guide to

the integration of these insurance instruments in this complex planning

environment.

INTRODUCTION

Earnings losses following job displacement are large for long-tenured workers in the

United States and similar economies (Canada and the United Kingdom).

1

These

losses arise from a combination of unemployment spells and lower wages on

the reemployment job. Of the long-tenured (3 or more years of service) workers in the

2012 Displaced Workers Survey displaced between 2009 and 2011, only 56 percent

Donald O. Parsons is in the Department of Economics at the George Washington University,

2115 G Street NW, Monroe Hall 368, Washington, D.C. 20052, and at IZA Bonn. Parsons can be

contacted via e-mail: dopars@gwu.edu. This article had an unusually long gestation time. Early

versions circulated under the titles “Private Job Displacement Insurance: Information

Asymmetries and Separation Pay Design” and “Double-Sided Moral Hazard in Job

Displacement Insurance Contracts.” Two anonymous referees provided valuable suggestions.

Detailed comments on early drafts by Bryan Boulier, Timothy Perri, and Torben Tranæs are

gratefully acknowledged, as are comments by seminar participants at the Economic Policy

Research Unit (EPRU) and the Center for Applied Microeconometrics (CAM), both at the

University of Copenhagen; the Korea Labor Institute (Seoul, Korea); the World Bank/IIASA/

Ludwig Boltzmann Institute International Workshop on Severance Pay Reform (Laxenburg/

Vienna); the U.S. Bureau of Labor Statistics; George Washington University; and Ohio

University.

1

Surveys of the U.S. displacement loss literature include Jacobson, LaLonde, and Sullivan

(1993a, 1993b), Fallick (1996), Kletzer (1998), Farber (2004, 2011), Couch and Placzek (2010),

and von Wachter (2010). See the contributions in Kuhn (2002) for an international perspective.

l  6JG ,QWTPCN QH 4KUM CPF +PUWTCPEG. Vol. 9999, No. 9999, 1–30 (2017).

DOI: 10.1111/jori.12216

Vol. 86, No. 2, 351–380 (2019).

2THE JOURNAL OF RISK AND INSURANCE

352

were employed in January of 2012, with another 26.7 percent unemployed and the

remainder, 17.4 percent, out of the labor force altogether ((Bureau of Labor Statistics

[BLS], 2012). For the reemployed, expected wage losses are substantial and strongly

tied to job tenure. Farber (2004, p. 115), analyzing various Displaced Worker Surveys

(DWS), reported that earnings losses of reemployed workers increase systematically

with job tenure: 3.4 percent at 1–3 years of tenure, 10.1 percent at 4–10, 20.5 percent at

11–20, and 28.1 percent at greater than 20.

2

Consumption studies indicate that these large losses for long-tenured workers are

poorly insured.

3

This is not because the risks are unknown or unattended to. In the

United States, the workforce is covered by public unemployment insurance and often

by employer-provided severance pay. Internationally, mandated severance pay is

ubiquitous as is government supplied unemployment insurance in the industrialized

world (Holzmann et al., 2012). This study stresses the importance of treating

unemployment insurance and severance pay as an integrated job separation benefit

package.

The early employment contracting literature considered severance pay as an

alternative to unemployment insurance in second-best contracts (Baily, 1977), but

severance pay also arises as scheduled wage insurance. The full job displacement

insurance design question does not arise in many optimal unemployment

insurance studies: (1) because the models ignored reemployment altogether,

presumably in order to focus on unemployed status, or (2) because reemployed

workers were assumed to receive wage offers equal to those in the displacing job.

Parsons (2015) lays out informally the argument for severance pay as scheduled

wage insurance as well as scheduled unemployment insurance and the need to

integrate the two programs into a unified system. This article develops this

argument formally.

In order to develop a deeper understanding of the coordination of the

two primary instruments (unemployment benefits and severance pay), a

single-period contracting model is employed.

4

A plausible outline of the first-

best package is straightforward and requires little formal motivation—severance

pay serves as wage insurance, covering the expected earnings losses (reemploy-

ment earnings less prior earnings), while unemployment insurance covers

the loss of TGGORNQ[OGPV earnings during the unemployment spell.

5

In the

model below, with additively separable state utility functions in consumption

2

All values converted to percentage change from differences in natural logs in the original.

3

Parsons (2015) provides a review of relevant consumption studies.

4

Reviews of the employment contracting literature include Parsons (1986), Malcomson (1999),

and Salani

e (2005).

5

Full reemployment wage insurance would require pooling of the earnings of the winners and

the losers, but this has not been seriously proposed. Most policy designers propose a negative

option payment—for those who make losses, some fraction of the losses will be paid through

premiums from all (Baily, Burtless, and Litan, 1993). For additional discussions, see Parsons

(2000), Kling (2006), and LaLonde (2007).