THE CHANGING RELATIONSHIP BETWEEN PRODUCTIVITY, WAGES AND UNEMPLOYMENT IN THE UK
| DOI | http://doi.org/10.1111/j.1468-0084.1993.mp55001005.x |
| Date | 01 February 1993 |
| Author | C. O. Alexander |
| Published date | 01 February 1993 |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 55,1(1993)
0305-9049 S3.00
THE CHANGING RELATIONSHIP BETWEEN
PRODUCTIVIT WAGES ANT)
UNEMPLOYMENT IN THE UK
C. O. Alexander*
I. INTRODUCTION
This paper uses time series methods to analyse the changing relationship
between unemployment, real wages and productivity. Most economic models
of the labour market, such as bargaining, search, contract and efficiency wage
theories, justify arelationship between these three variables. Only
insider-outsider models cast doubt on any relationship at all between un-
employment and wages.
There is also some empirical evidence that real wages, productivity and the
unemployment rate may be cointegrated in the UK. For example, Hall (1986)
uses quarterly data from 1963 to 1984 to show that wages, prices, hours
worked, productivity and unemployment form a cointegrated system, using
the two-step procedure of Engle and Granger (1987). Cointegration implies
at least one Granger causal ordering in the system (Granger, 1988) and it is
these causal orderings that the present paper investigates.
If X and Y are two jointly covariance stationary processes, then X is said
to 'Granger cause' Y if past Y and past X better predicts current Y than past
Y alone. Hence 'causality' in this paper refers to time lags between changes in
endogenous variables, not the exogenous determination of an endogenous
variable.
Economic theory justifies several possible 'causalities' in our tn-variate
system (although in applications the variables concerned may be jointly
determined endogenous variables). Theories of wage bargaining imply that
the unemployment rate affects wages, either through changes in bargaining
power or by interpreting the status quo pay-off to the union as the competi-
tive wage times the probability of finding a new job (see Layard et al., 1991).
A basic assumption of efficiency wage models is that higher wages will
increase productivity (Yellen, 1984), either through the threat of increased
unemployment or otherwise. A causal flow from wages to productivity is also
justified by the inclusion of relative factor prices in employment equations.
The reverse causality from productivity to wages is also possible: perform-
*1 wouid like to thank Andrew Newell and Mike Sumner for heJpfu comments on the first
draft.
87
88 BULLETIN
ance related pay-schemes are designed to effect a positive causality from
productivity (or, more specifically, worker's effort, which is only one aspect of
productivity as measured by the average product of labour) to wages. There
may also be a direct causality from productivity to unemployment which does
not involve wages, but with an ambiguous sign. The output effect of increased
productivity could decrease unemployment, but greater efficiency in the
workforce could also increase unemployment.
Some of these theoretical relationships have been sufficiently well
accepted to underpin some of the major macroeconomic models of the UK
economy. The Treasury model first included productivity as a shift variable in
the real wage equation in the early 1980's, and in 1986 the Bank of England
and National Institute models followed (see Wallis et al., 1987). But before
then the only recognized link between productivity and wages was a role for
relative factor prices in the determination of employment in all except the
non-manufacturing sector of the National Institute models. Following the
pioneering work by Layard and Nickell (1985, 1986) unemployment has
appeared in wage equations which are based on bargaining models of wage
determination, and these now replace the standard labour supply functions.
The available data are not necessarily entirely accurate (particularly given
the number of times the unemployment count has been changed by the
Conservative Government), but nevertheless the data have been used to test
these economic theories. Most econometric testing assumes a substantial
framework of economic theory at the outset. In contrast, this paper takes a
purely statistical approach and asks which, if any, labour market theories can
be supported by the apparent causal flows between unemployment, wages
and productivity. If the direction and sign of these agrees with established
theoretical models, this also helps to validate the data.
The results show that there has been a marked change in the nature of
causal flows following a structural break in the late 1970's. Before 1979
unemployment is the central variable, being caused by both wages and
productivity. There is no evidence of a direct relationship between wages and
productivity, as one would expect from the incomes policies which were in
force during this period. During the Thatcher period 1979-91 the Govern-
ment's attitude to inflation and unemployment changed, and this is reflected
by a strong bi-vanate causality between wages and productivity, unemploy-
ment becoming almost divorced from the system.
All data (obtained from Economic Trends and Department of Employment
Gazette) are quarterly and seasonally unadjusted, from 195501 to 199103.
The following notation is used: LPROD = log( Y/L), LRW= log( W( 1 - t)/P)
and UR = U/WP where Y= real GDP at factor cost; L= employees in
employment; W'- gross average weekly earnings; t = rate of income tax;
P consumer price deflator; U= total unemployment; WP= working popula-
tion. In an earlier draft of this paper it was found that using the real consumer
wage rather than the real producer wage makes no qualitative difference to
the analysis, and inclusion of the tax wedge is not necessary.
To continue reading
Request your trial