Unemployment burden and its distribution: Theory and evidence from India

• Date: 01 May 2018
• Author: Karthikeya Naraparaju,Sripad Motiram
• DOI: http://doi.org/10.1111/rode.12367
• Published date: 01 May 2018

REGULAR ARTICLE

Unemployment burden and its distribution: Theory

and evidence from India

Sripad Motiram

1

|

Karthikeya Naraparaju

2

1

University of Massachusetts, Boston,

Massachusetts

2

Indian Institute of Management Indore,

Indore, India

Correspondence

Sripad Motiram, Department of

Economics, University of Massachusetts,

Boston, MA 02125-3393.

E-mail: sripad.motiram@umb.edu

Funding Information

This work was supported by Indian

Council of Social Science Research

(ICSSR) (Grant No. RESPRO/31/ICSSR/

2013-14/RPS).

Abstract

We develop a measure of unemployment that takes into

account both the duration and intensity of unemployment.

This measure satisfies several desirable properties, includ-

ing distribution sensitivity, which deals with differences

among the unemployed. It is particularly suited to develop-

ing countries because individuals in these countries display

considerable variation in labor force participation, unem-

ployment duration, and unemployment intensity. It can also

be decomposed into mean and variance components and

contributions to unemployment by various subgroups of

the population. We use this measure and data from

National Sample Surveys on employment and unemploy-

ment to understand unemployment in India during the per-

iod 1993 to 2012. We show that unemployment has

generally fallen, although the distribution of unemployment

has worsened. Moreover, unemployment is driven to a

greater extent by higher educated groups; the unemploy-

ment among these groups is also fairly substantial. We

explain these findings and suggest some policies.

1

|

INTRODUCTION

I suppose there hasn’t been a single month since the war, in any trade you care to name,

in which there weren’t more men than jobs. It’s brought a peculiar, ghastly feeling into

life. It’s like on a sinking ship when there are nineteen survivors and fourteen lifebelts.

George Orwell (1969 [1939]), Coming Up for Air

DOI: 10.1111/rode.12367

Rev Dev Econ. 2018;22:787–807. wileyonlinelibrary.com/journal/rode ©2017 John Wiley & Sons Ltd

|

787

Unemployment imposes severe costs on individuals.

1

These costs increase in the duration of

unemployment, that too at an increasing rate.

2

Individuals unemployed for long durations could

witness atrophy or obsolescence of skills, and can be considered as victims of social exclusion and

capability failure (Sen, 2000). It is therefore necessary to realize that not all of the unemployed are

similar, and that short-term and long-term unemployment are qualitatively different, requiring dif-

ferent kinds of policy responses. However, many commonly used conceptualizations and measures

of unemployment (e.g., the rate of unemployment) and the official and academic discourses in

many countries (e.g., India) do not take this distinction seriously.

3

In this paper, we try to address

this gap by developing a “distribution-sensitive”measure. We use this measure to shed light on

unemployment in India during 1993–2012.

The short-term and long-term/chronic distinction also arises in the context of poverty.

4

The

measurement of poverty has been conceptualized as comprising of two different exercises: identifi-

cation of the poor using a poverty line, and aggregation to arrive at a figure for a country, regio n,

etc., using a poverty line and the distribution of income (Sen 1976). Analogous exercises exist in

the case of unemployment too (Shorrocks, 1992). However, despite recognizing the above consid-

erations, theoretical literature on the measurement of unemployment that has incorporated them

and arrived at newer measures is sparse. Lambert (2009) provides a survey of this literature, and

we can observe that it has been dormant for a while now. Most studies (Sengupta, 1990; Shor-

rocks, 1992, 1993; Paul, 1991, 1992) appeared in the early 1990s, although a few (Borooah, 2002;

Basu & Nolen, 2006) appeared more recently.

5

Moreover, except Paul (1991), the above studies have not been motivated from the perspective

of developing countries and some (e.g., Shorrocks, 1992) have been applied explicitly to devel-

oped countries. As we discuss below, developing and developed countries differ in terms of labor

market institutions and structural features. The nature of data that is available from developing

countries is also different. Hence, applying these models or approaches to developing countries

raises certain problems.

6

For example, Sengupta (1990) assumes uniform labor force participation

and that all individuals have some employment. We discuss the details (concepts, statistics, refer-

ence period, etc.) of labor force participation and unemployment for India below, but both these

assumptions are violated. For example, in 2011/2012, 46 percent (79 percent) of rural males (fe-

males) are absent from the labor force, and some individuals are unemployed throughout the refer-

ence period (i.e., no employment, more on this in Section 3).

Further, a substantial proportion enters and exits the labor force during the reference period: in

2011/2012, about one-third of the female labor force in rural India entered and exited the labor

force. This leads to complex differences across individuals in their labor force participation—some

are in the labor force for the entire period, some are out of the labor force for the entire period,

while the rest are in the labor force for only part of the period. These differences necessitate an

explicit consideration and analysis of unemployment intensity: the number of days a person is

unemployed as a fraction of the time he/she is in the labor force. The analysis of unemployment

intensities is missing in unemployment measures conceptualized for developed country contexts

(such as Shorrocks, 1993).

7

In light of the above, we develop a measure of unemployment that is sensitive to differences

among the unemployed and that takes into account some important features of labor and employ-

ment in developing countries. It can be decomposed into “mean”and “variance”components that

can help us understand how unemployment is being affected by changes in its interpersonal distri-

bution. It can also be decomposed into contributions made by subgroups of the population. At a

theoretical level, we provide an impetus to an important, but sparse literature on unemployment.

788

|

MOTIRAM AND NARAPARAJU