Does it Pay to Work for Free? Negative Selection and the Wage Returns to Volunteer Experience
| Author | Guido Cozzi,Robert M. Sauer,Noemi Mantovan |
| DOI | http://doi.org/10.1111/obes.12183 |
| Published date | 01 December 2017 |
| Date | 01 December 2017 |
1018
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
doi: 10.1111/obes.12183
Does it Pay to Work for Free? Negative Selection and
the Wage Returns toVolunteer Experience
Guido Cozzi†, Noemi Mantovan‡,and Robert M. Sauer§
†Institute of Economics, University of St. Gallen, (e-mail: guido.cozzi@unisg.ch.)
‡Bangor Business School, Bangor University, (e-mail: n.mantovan@bangor.ac.uk.)
§Department of Economics, Royal Holloway, University of London, (e-mail: robert.sauer@
rhul.ac.uk.)
Abstract
This paper offers the first instrumental variables estimates of the wage returns to volunteer
experience. The returns are substantial and differ considerably by gender.The results imply
that the unequal valuation of volunteerexperience by gender is more important in explaining
the gender earnings gap than is the unequal valuation of part-time paid work experience.
The results also indicate negative selection into unpaid work. In a simple model of optimal
volunteering, negative selection implies that a lower cost of volunteering would produce
both an expanded and higher-skilled pool of volunteers, and greater societal benefits from
volunteer work.
I. Introduction
Working as a volunteer is a widespread social activity. Many people engage in it in order
to help correct perceived social problems. But volunteers also benefit personally and eco-
nomically from this kind of pro-social behaviour. The personal benefits include the ‘warm
glow’ that one feels when giving to society. The economic benefits of volunteering derive
from the acquisition of human capital, an expanded network of contacts, and the signalling
of potentially productive characteristics to the labour market. Hence, working for free not
only brings a feeling of personal satisfaction, it may also havesubstantial investment value
that leads to higher wage offers in paid work opportunities.
In this paper, wefocus on measuring the investment value of volunteer workand offer the
first instrumental variables (IV) estimates of the wage returns to unpaid work experience.
The data on volunteering decisions and annual earnings that we use are drawn from the
British Household Panel Survey (BHPS) between the years 1996 and 2008. Information
from the BHPS is supplemented with data on rainfall in England, Scotland and Wales
during the same period. We construct an instrumental variable based on the rainfall data
in order to explicitly address the endogeneity problems involved and identify the wage
returns.
JEL Classification numbers: C26, D64, H41, J16, J31, J71.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 79, 6 (2017) 0305–9049
Does it pay to work for free? 1019
Rainfall data is particularly appropriate in this context since it has recently been shown
to be a credible source of exogenousvariation in the cost of participating in different types of
outdoor activities. Examples include attendance at 4th of July celebrations, political rallies
and riots (Collins and Margo, 2007; Madestam and Yanagizawa-Drott, 2011; Madestam
et al., 2013). Taking inspiration from these studies, we conjecture that rainfall also induces
exogenous variation in the cost of engaging in volunteer work. Greater expected rainfall
in a locality may lower the opportunity cost of volunteering, as alternative outdoor leisure
activities become less attractive. This could increase the propensity to engage in unpaid
work which mostly takes place indoors. Greater anticipated rainfall should also have no
direct effect on earnings after controlling for unobserved individual fixed effects and a
detailed set of observables, including paid work hours.
According to ordinary least square (OLS) estimates that do not seriously address the
endogeneity problem, the increase in mean annual earnings due to volunteer experience,
in constant 1987 pounds sterling, is a very modest £394 for men and a negligible -£29
for women. However, specifications which make use of the longitudinal aspect of the data
and include unobserved fixed effects yield higher and more precisely estimated returns of
£1,372 for men and £649 for women. IV estimates that exploit the rainfall instrument in
addition to including fixed effects produce still higher estimated annual returns of £4,859
for men and £3,096 for women. The IV wage returns to volunteer experience are precisely
estimated and substantial in magnitude as sample average incomes are £11,725 and £7,007
for employed men and women respectively.
It is interesting that in all of our specifications, the estimated returns to unpaid work
experience for men exceed those for women. A standard decomposition indicates that the
larger returns for men can account for up to 20.2% of the gender earnings gap. To put this
contribution to the gender earnings gap in perspective, we find that that it is smaller than
the (uncorrected) contribution of full-time paid work experience (25.3%) but greater than
the (uncorrected) contribution of part-time paid work experience (8.4%). The implication
of the decomposition exercise is that an unequal valuation of volunteer experience by
gender may be relatively more important in explaining the gender earnings gap than is
the unequal valuation of part-time paid work experience, and nearly as important as the
unequal valuation of full-time paid work experience.
In order to explore possible mechanisms underlying the substantial wage returns to
volunteer experience for both men and women, we also explore descriptive data from the
UK Citizenship Survey (UKCS). The UKCS does not reveal strong evidence in favour of
a human capital or networking explanation for the wage returns. For example, gaining a
recognized qualification or improving employment prospects is not stated to be a main
motivation for volunteering. Rather, volunteers appear to be seeking a sense of personal
achievement, a feeling of being needed and enjoyment from doing things at which they
excel. Thus, signalling may be a more likely source of the returns if these latter person-
ality characteristics are also productive in the workplace and not immediately observable.
Disconcertingly, the UKCS data do not reveal obvious gender differences in sources of
satisfaction, motivations, types of volunteering organizations and activitiesthat might help
explain the differential returns by gender.
The policy implications of the study are explored by formulating a simple model of
optimal volunteering and linking the theory to the empirical work. The model implies
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial