MEASURING UNION POWER IN BRITISH MANUFACTURING: A LATENT VARIABLE APPROACH*
| Date | 01 February 1993 |
| Author | Adam Wagstaff Peter Holl,Pierella Paci |
| Published date | 01 February 1993 |
| DOI | http://doi.org/10.1111/j.1468-0084.1993.mp55001004.x |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 55, 1(1993)
0305-9049 S3.00
MEASURING UNION POWER IN BRITISH
MANUFACTURING: A LATENT VARIABLE
APPROACH*
Pierella Paci, Adam Wagstaff and Peter Hou
I. INTRODUCTION
The last 20 years have seen a lively debate amongst labour economists,
industrial relations specialists and policy-makers on the effects of trade union
power on economic performance In the empirical literature, in particular,
attempts have been made to assess the effects of union power on crucial
economic variables, such as wages, employment and, more recently, produc-
tivity and technological and organizational change. Work in this area has,
however, been severely hampered by the problems associated with quantify-
ing union power, which is, by its nature, unobservable. Typically proxy
variables have been used and the debate has then shifted to the issue of which
observable variables best reflect trade union power.1 None of the proxies
used, however, captures the full spectrum of factors which result in power
differentials across unions and workplaces.2
This paper proposes an entirely different approach. It models union power
explicitly as a 'latent' variable - an approach which allows an index of union
power to be computed for each observation. Our data are taken from the
British 1984 Workplace Industrial Relations Survey. This survey is rich in
information on industrial relations issues, but has less extensive information
on economic factors. Most notably the survey lacks information on output
and capital stock. Although WIRS has been used to explore the effects of
unions on employment (cf. Blanchflower, Millward and Oswald, 1991), there
is some controversy surrounding the results of this work (cf. e.g. Machin and
Wadhwani, 1991 ).3 In the present paper, therefore, we focus on union power
over wages.
*We are grateful to an anonymous referee for helpful comments on an earlier version of the
paper. We alone are responsible for any errors.
'For an early survey of this debate see Armstrong, Bowers and Burkitt (1976), and, more
recently, Lewis (1983).
2 It is for this reason that Lewis (1983) suggests the inclusion in micro' wage equations of
more than one proxy for union power.
3i that the coefficient on lagged employment in the employment equation of Blanch-
flower, Millward and Oswald (1991) is not significantly different from one, the equation they
estimated is best interpreted as an employment-change equation. Their results concerning the
65
66 BULLETIN
Our latent variable approach not only allows us to quantify the degree of
union power in the manufacturing sector as a whole, but it also enables us to
explore a number of related issues. We are able to explore variations in union
power across the workplaces in the data set. We are also able to contribute to
the debate over the best proxies for union power, by assessing and comparing
the strength of the correlation between the various proxies and our union
power index. We also explore the validity of widely held beliefs about the
determinants of union power and its distribution across establishments with
different characteristics.
The paper is organized as follows. Section II defines the concept of union
power and outlines both its determinants and its potential effects. Section III
presents our method for measuring union power. Section IV describes the
data and model specifications used. Section V presents the empirical results.
Section VI contains some brief conclusions.
II. THE DEFINITION OF UNION POWER, ITS DETERMINANTS AND ITS
EFFECTS
In the game theoretic approach to bargaining, union power is defined in
terms of the share of any 'rent' captured by the union. Traditionally the
absolute size of the 'cake' is given by some combination of the employer's
profit and the union's utility, while the slice captured by the union is
expressed in terms of its two main 'ingredients': wages and employment in
excess of the competitive level.
In this framework, therefore, union power does not, as is often claimed,
depend upon the absolute size of the rent.4 Any causal relationship is in fact
in the other direction: the existence of a positive rent may, for example, result
in reduced union power if the employer uses part of the rent to build up
above-normal inventories to reduce the impact on sales of any industrial
action by the union (Stewart, 1990). This is not to say, of course, that the
absolute magnitude of any rent captured by the union is invariant with
respect to the absolute size of the rent available for division. Clearly, for a
given level of union power, the amount of rent captured by the union is
greater the larger is the 'cake' to be divided. Thus even a very strong union
can be left empty handed if there is no cake to be divided (i.e. if the rent is
zero), as, for example, is the case when the output market is competitive. The
outcome of the bargaining process depends therefore on both the bargaining
power of the union and the absolute size of the rent to be divided.
The power of the union derives from its ability to present the employer
with a credible and sustainable threat, such as the threat of strike action.
Union power is therefore increasing in the union's ability to convert the
effects of unions on employment therefore say more about the change in the impact of unions
over the period in question than about the level of their impact.
an alternative view see, for example, Mishel (1986).
To continue reading
Request your trial