Productivity Growth in the UK Regions, 1968–1991
| DOI | http://doi.org/10.1111/1468-0084.00079 |
| Published date | 01 November 1997 |
| Author | Richard I. D. Harris,Mary Trainor |
| Date | 01 November 1997 |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 59, 4 (1997)
0305-9049
PRODUCTIVITY GROWTH IN THE
UK REGIONS, 1968–91
Richard I. D. Harris and Mary Trainor
I. INTRODUCTION
Advances in technology, together with the more efficient use of existing
technology, are vital sources of output growth. That is, technical change
shifts the production frontier so that fewer factor inputs are needed to
produce the same level of output (or equivalently more output is
produced with the same level of inputs), while increases in efficiency
move production towards the current ‘best-practice’ frontier.1The other
major source of supply-side growth is increases in factor inputs, but if
non-increasing returns to scale are assumed then increases in demand
tend to dominate an economy’s growth path. Without entering into a
debate on which factors (supply-side or demand-side) are in fact the most
crucial determinants of growth, it is relatively uncontroversial to argue
that improvements in technology must play an important role.
There are several commonly used measures of productivity growth,
with increases in labour productivity most often cited as evidence of
increased competitiveness. The major problem with measures related to a
single factor input is that they also incorporate the effects of factor
substitution which, during periods when factor price ratios range signifi-
cantly, can be a misleading indicator of technical change. In general, the
factor-price ratio between capital and labour services has been falling in
UK manufacturing for a good deal of the last quarter century, and thus
much of the gain in labour productivity has been achieved through
‘capital deepening’ rather than ‘capital widening’. A better measure of
productivity growth is to measure total factor productivity (TFP). The
usual approach is to calculate the ‘Solow-residual’ based on the growth
accounting technique, whereby the growth of factor inputs (weighted by
their shares in output) are subtracted from output growth to obtain TFP
growth. While there are certain limitations to this approach (e.g. competi-
tive markets are often assumed so that each factor’s share in output
exhausts total revenue — i.e. the ‘adding-up’ problem — and this method
1Technical change can be relabelled as ‘innovation’, while efficiency gains leads to ‘catch-
up’. 485
© Blackwell Publishers Ltd, 1997. Published by Blackwell Publishers, 108 Cowley Road, Oxford
OX4 1JF, UK & 350 Main Street, Malden, MA 02148, USA.
does not distinguish the separate effect of changes in efficiency from
shifts in the frontier), nevertheless it is the standard used in the literature
because it is simple to use and produces results that can be easily inter-
preted. Thus, we have also used this approach, although the assumption
of competitive products markets is relaxed (following Hall, 1986).
This paper computes estimates of total factor productivity growth, for
13 industries in ten UK regions covering the period 1968–91. The next
section discusses various approaches to measuring productivity change
and sets out the preferred model. In Section III, the data is briefly
discussed, including the problem of adjusting estimates of the regional
capital stock for premature scrapping. Section IV presents the major
results while in Section V we attempt to explain the region by industry
estimates of TFP growth for 1979–89 using cross-sectional data on the
human-capital attributes of the workforce (given the arguments found in
endogenous growth theory models), as well as the impact of various other
factors such as changes in minimum efficient scale and trade union
density levels. Finally, a summary and some conclusions are presented.
II. MEASURING PRODUCTIVITY GROWTH
Recently Oulton and O’Mahony (1994) carried out a detailed study to
measure total factor productivity growth for various UK manufacturing
industries for the period 1954–86. They used the standard approach
pioneered by Solow (1956), Denison (1967) and Jorgenson and Griliches
(1967), which amounts to subtracting weighted shares of the growth of
factor inputs from output growth to obtain a ‘residual’ which encapsulates
‘all other factors’ including technical change and efficiency gains. Clearly,
this ‘residual’ also captures any misspecification in the underlying model
(e.g. the imposition of constant returns to scale and/or competitive
markets) as well as errors in the data (e.g. due to failure to take account
of premature scrapping of the capital stock). While Oulton and
O’Mahoney (op. cit.) point out data limitations as and when appropriate,
they nevertheless justify their use of the neo-classical approach as follows:
The neo-classical approach to growth accounting is a discipline, which
enforces consistent choices at every step along the way. Without this
discipline, the researcher is liable to flounder in a swamp of ‘adhocery’.
Moreover, no other approach has been so fully worked out. (p. 23)
Other approaches attempt to test for the validity of underlying assump-
tions using both parametric and non-parametric techniques, while at the
same time distinguishing between shifts in technology as opposed to
movements towards the best-practice frontier. For instance, Boskin and
Lau (1992) use the concept of a meta-production function which,
although assuming a common underlying technology across countries (or
regions or industries), tests for divergences across the sub-groups in the
© Blackwell Publishers 1997
486 BULLETIN
To continue reading
Request your trial