Earnings Instability and Tenure

• Published date: 01 April 2016
• Author: Lorenzo Cappellari,Marco Leonardi
• Date: 01 April 2016
• DOI: http://doi.org/10.1111/sjoe.12142

Scand. J. of Economics 118(2), 202–234, 2016

DOI: 10.1111/sjoe.12142

Earnings Instability and Tenure∗

Lorenzo Cappellari

Catholic University of Milan, 20123 Milan, Italy

lorenzo.cappellari@unicatt.it

Marco Leonardi

University of Milan, 20122 Milan, Italy

marco.leonardi@unimi.it

Abstract

We study the effect of tenure on earnings instability in Italy using the reforms of temporary

employment contracts, which affected the average tenure of workers differentially across

cohorts. We develop a model of earnings dynamics, and we exploit the variation of tenure

and instability over time and across birth cohorts to estimate policy-relevant parameters. Our

results indicate that each year of tenure on the job reduces earnings instability by 11 percent;

the drop is faster in the first three years of the match. Workers on a temporary contract have

an earnings instability up to 100 percent higher than workers on a permanent contract.

Keywords: Earnings dynamics; labor reforms; permanent and transitory variance; temporary

contracts

JEL classification:J21; J31

I. Introduction

A large and growing body of literature uses panel data on individual earn-

ings to look at the extent of intertemporal mobility in the distribution of

earnings, distinguishing long-term earnings components (which are related

to changes in the quantity and prices of permanent individual character-

istics) from a transitory component that captures the extent of earnings

instability; see the recent review by Meghir and Pistaferri (2011).1The

distinction between permanent and transitory inequality is important for

various reasons. First, it is useful in evaluating the welfare implications

of changes in inequality. An increase in permanent inequality would cer-

∗We thank Christian Belzil, Chris Flinn, Peter Gottschalk, Luigi Guiso, Fabian Lange, Robert

Moffitt, Magne Mogstad, Chris Taber, and two anonymous referees for helpful comments.

We are also grateful to Alessia Matano and Paolo Naticchioni for their expert advice with

the INPS data.

1An incomplete list of the studies includes the following: Haider (2001) and Moffitt and

Gottschalk (2012) for the US; Dickens (2000), Ramos (2003), and Alessie and Kalwij (2007)

for the UK; Baker and Solon (2003) for Canada; Cappellari (2004) for Italy; and Bingley

et al. (2013) for Denmark.

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

L. Cappellari and M. Leonardi 203

tainly reduce welfare, while increasing transitory inequality would have

a weak effect on welfare measures, unless there are liquidity constraints

that restrict consumption smoothing.2Second, it informs the assessment

of different explanations for the increase in inequality: if rising inequality

reflects an increase in permanent inequality, then a consistent explanation

would be skill-biased technical change; in contrast, an increase in transitory

inequality could reflect greater flexibility among workers in switching jobs,

and therefore higher income mobility.

Tenure is a potentially important covariate of instability and, in this pa-

per, we model their relation. Although we are not the first to model tenure

in earnings variance models, we are the first to model the transitory com-

ponent of earnings with respect to tenure.3Previous studies modeled the

relation between permanent earnings and tenure, and were motivated by

testing between alternative theories of wage determination (Parent, 2002).

We believe that job tenure should also affect earnings instability. Specifi-

cally, we should expect earnings instability to decrease with job duration

if there is employer-learning about the quality of the match over time

(Lange, 2007), or if firms insure earnings against volatile shocks as they

improve their knowledge of the match quality (Guiso et al., 2005). The

earnings model of this paper includes tenure effects on both the transitory

and permanent components of the earnings process.

The recent body of literature has gone beyond the simple permanent–

transitory earnings decomposition, using two different strategies. The first

strategy involves reduced-form models, which examine the correlates of

instability. Cameron and Tracy (1998) and Baker and Solon (2003) have

explored the relation between instability and age. Although some previ-

ous studies have investigated the impact of both quits and layoffs on the

transitory variance of wages in the US, none of these papers has explic-

itly modeled the effect of tenure on instability (see Huff Stevens, 2001;

Hospido, 2012; Leonardi, 2012). A neighboring and growing body of lit-

erature focuses on earnings volatility (i.e., the variance of year-to-year

2Besides the distinction permanent–transitory, the distinction between “predictable” and “un-

predictable” (Cuhna et al., 2005; Cuhna and Heckman, 2007) and “insurable” and “uninsur-

able” (or partially insurable) shocks is particularly common in the macro literature (Heathcote

et al., 2010, 2014). This distinction requires the modeling of the structure of credit and in-

surance markets, and other risk-sharing characteristics of households (labor supply, family

networks, etc.) or of government transfers, which might crowd out private transfers and

self-insurance (Blundell et al., 2015). For our purposes, it is enough to say that the variance

of transitory shocks is very relevant for individual measures of welfare, either because it is

unpredictable or because it is uninsurable.

3Many papers have focused instead on estimating the average returns to tenure: see Altonji

and Williams (2005), Dustmann and Meghir (2005), and references therein.

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

204 Earnings instability and tenure

earnings changes), and finds mixed evidence on the relation between work-

ers’ turnover and volatility (see Dahl et al., 2011; Venn, 2011; Ziliak et al.,

2011; Celik et al., 2012; Cappellari and Jenkins, 2014). However, papers

that study earnings changes rather than levels do not distinguish between

permanent and transitory shocks to earnings (Shin and Solon, 2011; Dynan

et al., 2012). A common problem with this literature on reduced-form mod-

els is, of course, the endogeneity of job-to-job mobility. We try to address

both problems: we estimate the effect of tenure separately on the permanent

and transitory components of earnings, and we identify changes in work-

ers’ tenure with cohort-time differences and with the differential exposure

of birth cohorts to the succession of reforms of temporary employment

contracts in Italy.

The second strategy, pursued in another strand of the literature, has

analyzed the economic forces behind the degree of persistence and of

variability in earnings, building structural models to better characterize

behavior. Low et al. (2010) model labor supply and job mobility in a

search and matching framework. Their approach is explicit about distin-

guishing between shocks and responses to shocks (i.e., job mobility) and

between different types of uncertainty, loosely associated with employment

risk (i.e., rates of arrival of job offers) and productivity risk (i.e., shocks to

the match). Confirming that job-to-job mobility is important, they f ind that

if mobility is ignored, the estimated variance of the permanent innovation

to wages doubles, leading to an impression of much greater risk in the

earnings process. This is because many of the wage fluctuations are due to

individuals’ moving to jobs with better match-specific effects. Flabbi and

Leonardi (2010) and, on a more complex scale, Altonji et al. (2013) esti-

mate a model of wages and transitions between jobs and into unemployment

driven by exogenous shocks, which are the underlying source of fluctua-

tions. In the same line of research but with more focus on wage growth,

Adda et al. (2013) model workers’ career progressions in a framework

in which wages grow because workers learn on the job and through job

shopping. Relative to reduced-form estimates, these models have the advan-

tage that one can construct counterfactual life-cycle profiles, by comparing

profiles with and without returns to experience, tenure, or job mobility.

However, this is obtained at the cost of imposing specific distributional

assumptions.

We use Italian Social Security records between 1986 and 2003, and

illustrate how the variation in tenure is connected with a succession of

labor market reforms that liberalized temporary employment contracts and

had a differential impact across birth cohorts and over time. In the US,

it has been difficult to establish a link between earnings instability and

workers’ tenure because the empirical literature has found little evidence of

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