A CRITICAL COMPARISON OF MODELS OF STRIKE ACTIVITY
| DOI | http://doi.org/10.1111/j.1468-0084.1993.mp55003002.x |
| Published date | 01 August 1993 |
| Author | Karen Mumford |
| Date | 01 August 1993 |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 55,3(1993)
0305-9049 $3.00
A CRITICAL COMPARISON OF MODELS OF
STRIKE ACTIVITY
Karen Mumford
I. INTRODUCTION
Despite the attention that strike activity receives, strikes have proven elusive
to model and, whilst many theories of strikes have been developed, no single
theory has gained wide acceptance. Indeed, there is a vast array of models
explaining strike activity in the literature: Kennan (1986) presents an
exhaustive survey of strike models and empirical results, Card (1 990b)
provides a survey of more recent findings. Such an expanse places the analyst
in the difficult position of deciding which model is the most preferable to use.
This paper provides a theoretical and empirical comparison of the major,
and more influential, of the models that attempt to explain strike activity:
incomplete information (Ashenfelter and Johnson, 1969), strike activity and
associated costs (Reder and Neumann, 1980), asymmetric information
(Hayes, 1984), and the world-wise approach (Tracy, 1986). Discussion of the
theoretical implications, and results, of each of these models are presented on
an individual basis in Section II of the paper. The empirical findings are
compared and contrasted in Section III, where an attempt to rank the models
using nested and non-nested testing criteria is used. Finally, the common
themes and predictions of the models are explored in terms of the
relationships which are found to be important in the data set, and an eclectic
model is presented in Section IV.
The empirical application of these models employs annual data from one
of the most strike-torn industries in the world, the coal industry in New South
Wales (NSW), Australia,' from 1 952_87.2 The main sources of this data are
the Yearbooks of the NSW Joint Coal Board, the Mining Census of NSW,
and the holdings of the Australian Bureau of Statistics. The data set is
discussed further in an Appendix which is available from the author.
'O'Dea(1965, p. 78) and Beaumont(1975, p.48).
2A further reason for using the NSW coal industry in an investigation of strike activity is that
several government agencies have been directed to collect information on the industry and data
are therefore available for an extensive range of variables since 1952. The resultant data set is
very rich, enabling the theoretical models to be considered with very similar (indeed, usually the
same) variable definitions as those advocated by the original authors (sometimes for the first
time). All nominal variables have been deflated and are expressed in $1952.
285
286 BULLETIN
II CONSIDERING THE MODELS ON AN INDIVIDUAL BASIS
Ashenfelter-Johnson: incomplete information
Ashenfelter and Johnson (1969) provide one of the earliest models of strikes
which explicitly postulates that the labour market is typified by incomplete
information.3 They argue that there are three parties involved in labour
market bargaining processes: the trade union rank and file membership (who
are the only party without complete information of the bargaining surplus),
the trade union leadership, and the employers. The trade union membership
is represented by the union leadership. The union leadership has two major
objectives: the survival and growth of the union as an institution, and the
personal political objectives of the leaders themselves. To satisfy these
objectives the leaders need to represent the wishes of the rank and file
members. The negotiated wage increase that is acceptable to the union
membership is Ya
Ya ( TI'; - W1_1)/T4';_1. (1)
Where W, is the previous contract wage, and W, - -1 is the acceptable
absolute wage increase. Furthermore, this acceptable absolute wage increase
will be a function of the length of the strike, S. Thus:
Yaf( (2)
Kennan (1986, pp. 101-3) provides a simple presentation of the Ashen-
felter-Johnson model, whereby a profit maximizing firm sells output at a
fixed rate, Q, and a fixed price, P. The firm employs a fixed number of
workers, M, to whom it is originally paying a wage rate of W The union
leadership presents a new wage claim which is an increase of Y over the
previous wage. The minimal proportional increase acceptable to the trade
union is Yo' where Yo =f(0), if the union does not receive this wage increase it
will strike. The wage increase that the union will never accept (regardless of
the length of the strike) is Y* where Y* =f( co). As a strike progresses the
union modifies its claim along an exponential schedule:
y =y(S)=Y*+(Y (3)
Where S is the duration of the strike and r measures how quickly the union
concedes (adapts its expectations).
The firm wishes to maximize its profits, PQ - WM, in this simple model.
The highest conceivable wage increase that the firm will offer per worker will
be Yb = (PQ - WM)/M If the wage increase actually settled on is Yb' then the
3There have since been many applications of this model. 'The bulk of the empirical work on
U.S. strike activity has used aggregate time-series data These studies typically estimate some
variant of Orley Ashenfelter and George Johnson's model...' Tracy (1986, p. 423). See
Pencavel (1970), Turkington (1975), Farber (1978), Mauro (1982), and Abbott (1984). A
survey is provided in Kennan (1986).
To continue reading
Request your trial