Professional Documents
Culture Documents
Daly1986 Education and Productivity
Daly1986 Education and Productivity
NOTES
a. A further breakdown o f these figures for twelve manufacturing industries and ten non-
manufacturing industries is available in Daly (19x3).
h. Members of professional institutions with qualifications of degree standard (Census of
Population ’b’ levcl) and graduate teachers are included in the British figures. For the U . S .
it includes all who have completed four or more years of full-time college.
c. Includes for Britain, non-graduate qualifications such as H . N . C . and City and Guilds Full
Technological Certificates; O.N.C./O.N.D. apprenticeships, G.C.E. ‘A’ level, ‘0’level
and C.S.E. qualifications. For the U.S. it includes those who have completed high school
and those who have completed one to three years of full-time college education. The U.S.
data include those who have completed one or two year courses and those who started but
did not complete, a four year programme.
d. For Britain. includes those with no educational or vocational qualification. For the U . S .
includes those who have not completed high school.
Education and Productivity 253
interpreted as representing an exact match between the education systems
of the two countries but rather to show the broad order of magnitude for
each of three groups; those who have completed university o r a four-year
college degree, those with qualifications beyond the minimum school
leaving qualifications and those without education or vocational qual-
ifications (the notes t o Table 1 give details of the coverage of each category).
The most striking feature of the summary totals is the large proportion of
the U.S. labour force in the university category, over three times the
proportion of the British labour force in the group. Part of the difference
reflects the wider range of courses covered in the U.S. by four or five-year
bachelor degrees; for example, nursing and physiotherapy are degree
courses in the U.S. but training in these occupations falls outside the full-
time education system in Britain. We are, however, unable to correct for
these differences due to the level of aggregation of the data.
Over half the U.S. workforce fell into the intermediate category. Despite
the smaller gap between the two countries at this level, there remained a
large difference in the number with no qualifications. About a quarter of the
U.S. workforce had no qualification compared with over a half of the British
workforce. This finding is in line with the earlier studies by Prais (1981) and
Hollenstein (1983) showing that the proportion of the British workforce
without any educational or vocational qualifications was about twice that of
these successful industrialised countries.
TABLE
2
Average Number of Students Receiving First Degrees from Universities
and Colleges in Britain and the U.S. in 1976-8, by Main Subject of Study.
Sources: Education Statistics for the United Kingdom, 1976 and 1977 (HMSO, 1979, Tables 22,
25 and 35). Digest of Education Statistics, 1980 (National Center for Education Statistics,
Washington, 1980, Tables 110 and 113).
254 British Journal of Industrial Relations
NOTES
Includes 57,049 first degrees from Universities, 12,925 degrees from CNAA. 1,446 London
external degrees and estimates in respect of 6,879 degrees from the Open University. Open
University degrees are multi-disciplinary, and no classification by main subject of study was
available, though a classification is available in respect of its student population. We have
estimated the distribution of degrees by subject in proportion to the distribution of the
student population.
Includes 920,559 bachelor’s degrees requiring four or five years at an institution for higher
education, that is either a university or a four year college. 66,581 first professional degrees
are also included (e.g. law and medicine). These degrees require a total of at least six
academic years of college work and represent the completion of the academic requirements
for practising in a particular profession.
Includes courses not usually taken at the degree level in Britain, for example, nursing and
medical record librarianship. The figure in brackets omits graduates with the following U S .
degrees: general health professions, hospital administration, nursing, occupational ther-
apy, dental hygiene, public health, medical record librarianship, speech pathology, clinical
social work and physical therapy.
Includes law, accountancy, business management, government and public administration,
architecture, library science, agriculture and veterinary science; and for America it also
includes communications (journalism and advertising) and home economics. The figure in
brackets omits graduates with the following U.S. degrees: banking and finance, insurance,
real estate, secretarial studies, communications, home economics, and law enforcement
and corrections.
Includes 46,000 teaching qualifications in Britain.
‘Excludes law, accountancy and business management (see note 5).
The U.S. figures cover a wider range of degrees than are found in Britain,
so we have included, in brackets, U.S. figures covering only those subjects
taught in British universities for the two subject areas where these
differences were concentrated, namely medicine and health, and vocational
studies. This reduced the number of U.S. graduates in medicine and health
by a half and in vocational studies by 22 per cent; but the U.S. superiority in
these areas remained.
In the U.S. there was a wide choice of specialised practical degrees which
contrasted with the more general and academic options available in Britain.
The most striking difference was in the numbers completing courses included
in the category ‘vocational studies’ (such as accountancy, business studies and
public administration) where the output of U.S. graduates per member of the
labour force was ten times greater than the British figure. There were more
U.S. graduates per member of the labour force in the science-based subjects,
but taking a wider comparison, for engineering there were relatively more
graduates produced in Switzerland and Germany than in the U.S. The annual
output ofengineeringgraduatesaccountedfor0.054percentofthe workforce
in the US.,0.060 per cent in Germany and 0.082 per cent in Switzerland (see
Hollenstein (1983) Table 4). They, however, accounted for an even smaller
percentage (0.043) of the British workforce.
TABLE
3
Average Number of Students Receiving Associate Degrees in America
and the Number Receiving Comparable Awards in Britain, 1976-8
Sources: Education Staristicsfor the United Kingdom, 1976 and 1977. Tables 23 and 24. Digest
of Education Statistics. 1980, Table 125.
NOTES:
(a) Includes 11.148 students passing the Higher National Diploma and 16,600 passing the
Higher National Certificate and other awards as described in footnotes c-f.
(b) Includes awards received after completing occupational courses of at least two years but
less than four years duration.
(c) Includes 9,600 students completing City and Guilds technician certificate (Part 111).
(d) Includes the 9,500 people completing the two-year course for pupil nurses (SEN).
(e) Less than 0.001 per cent.
( f ) Includes 570 students completing a two-year teaching diploma.
TABLE
4
Apprentices in Britain and the U.S. 19745
Sources: U.S. Department of Labor Employment and Training Report of the President, 1977
(U.S. Government Printing Office, Washington, 1977) Tables A15 and F16. U.S. Department
of Labour, Employment and Training Administration, Dictionary of Occupational Titles (U.S.
Government Printing Office, Washington, 4th edition, 1977). For Britain, Department of
Employment New Earnings Survey 1974 (HMSO London, 1975), Tables 136 and 139.
NOTES
(a) Apprenticeship programmes registered with the Bureau of Apprenticeship and Training
must comply with certain minimum standards relating to the availability of classroom
instruction, and supervision of on-the-job training.
(b) Manual workers include blue collar, service and farm workers.
(c) The Survey results have been grossed up by a factor of 138 to cover the whole population
(see The New Earnings Survey 1974, Part A , p. A40).
(d) A non-manual occupation under the British classification but included in this table as
draughtsmen were among registered apprentices in America.
(e) Includes cooks, barbers, beauticians, butchers and bakers.
(f) Includes carpenters, bookbinders, compositors, lithographers, pattern-makers and cabinet
makers.
(g) Includes boilermakers, auto and related mechanics, machinists, millwrights, pipe fitters,
plumbers, tool and die makers and optical workers.
(h) Includes bricklayers, plasterers, roofers, cement layers, lathers, glaziers and structural
steel workers.
(k) Taken as roughly a fifth of registered apprentices (assuming that the average apprentice-
ship is for three to four years and about a quarter drop out or fail each vear).
The preceding sections of this paper have compared the ‘stocks of education’
held by the British and U.S. workforces and the recent additions to these
stocks. This section examines the effects of these differences on economic
performance as measured by net output per employee in a sample of sixty
manufacturing industries in 1967-8 (see appendix for data sources). This
cross-section approach can be used to examine the industry specific factors
affecting productivity levels but does not lend itself to the study of general
factors affecting all industries equally (for example, tax and labour laws).
Education and Productivity 259
The equations estimated are broadly speaking of a Cobb Douglas form
but they are not true production functions (for a fuller discussion of the
theoretical justification of the model see Davies and Caves (1983)). The
dependent variable chosen was output per employee in the industry rather
than the total output of the industry, in order to standardise across industries
by taking account of such factors as differences in their absolute size. The
capital input was measured by a stock variable taken as the best approxima-
tion to the flow of capital services. It was not possible to explicitly include all
the labour input variables because of their multicollinearity (as the variables
are defined, they must sum to one hundred). We have, therefore, included
the skilled labour input, representing the stock of labour input over and
above that of the unskilled. The co-efficient on the unskilled labour input
can be calculated implicitly.
We begin by looking at the relationship between the productivity of an
industry and its level of physical and human capital input, taking each
country separately. We have measured human capital input by the
proportion of the workforce with university and intermediate qualifications;
for the latter we have adopted a more restrictive definition than that used in
Table 1 due to limitations in the data available at this level of aggregation.
We found for the U.S.:
This suggests that U.S. industries performed relatively well where the
proportion of highly skilled workers exceeded that in Britain, for example,
in chemicals and radio and television equipment. However, the co-efficient
on the intermediate skill difference shows a depressing effect on relative
productivity. This may reflect the problems already discussed in defining this
variable and until better information is available about this group, it does
not seem worthwhile pursuing this result.
The relationship between education and relative economic performance
is complex and it is perhaps not surprising that this statistical analysis has left
several important questions unanswered. The conclusion that Britain’s
shortfall in ‘high’ skills had a detrimental effect on relative productivity from
Britain’s viewpoint appears to be fairly robust but the role of intermediate
skills could not be estimated here and other more general questions remain.
We have considered how the level of education in a particular industry
affects its relative productivity but not how the generally higher level of
education in the U.S. may contribute to the country’s absolute productivity
advantage over Britain, nor have we considered the influence of education
on productivity growth over time. A more educated workforce may lead to
better investment decisions, improved labour relations and greater adapt-
ability to changing economic circumstances. These effects are not taken into
account in this kind of analysis.
It is important also to consider the efficiency with which the educated
manpower is utilised. In recent years, there has been considerable discus-
sion of over-investment in education in the U.S. and there is some evidence
of a declining rate of return to investment in college education in the 1970s
(post dating the period to which our regressions relate). Many factors were
involved but among those outlined by Freeman (1975) was a lack of jobs in
fields related to the area of study. The skills acquired by those people who
have ‘over-invested’ in education may have minimal relevance to their
current occupation but they will nevertheless be included in the appropriate
educational groups in our tables. These problems also arise in the case of
education undertaken purely for its consumption benefits. Given the
relatively high level of general education in the U.S. we would expect these
problems to be of greater significance there than in Britain and they may
result in some over-estimation of the effective educational input and an
under-estimation of the effect of vocationally relevant education.
262 British Journal of Industrial Relations
APPENDIX
i) ‘high’ skills (Census groups A and B) included those with university and
CNAA degrees and post graduate qualifications, in addition to those with
equivalent professional qualifications.
ii) ‘intermediate’ skills (Census groups C and D) included those with Higher
National Diplomas and Certificates, Ordinary National Diplomas and
Certificates, City and Guilds Advanced Qualifications and G.C.E. ‘A’ level
and their Scottish equivalents.
Physical capital per worker in the U.S. industries in 1967 was taken from
the 1972 Census of Manufactures published by the U.S. Department of
Commerce, Bureau of the Census. The capital stock was measured by the
gross value of fixed assets. In order to have a comparable measure of the
capital stock for Britain, the estimates made by Ian Elliot, Disaggregation of
1968 Gross Fixed Capital Stock Data to M L H Level-Revision and Testing of
Estimates (NEDO, London), for 1968were used. Median plant size, defined
so that half the employees in the industry were in larger plants and half in
plants that were smaller, was calculated from the U.S. Census of Manufac-
264 British Journal of Industrial Relations
tures and the British Census of Production. The proportion of workers who
were part time, those working less than thirty hours a week, were taken
from the U.S. Census of Population Industrial Characteristics Table 38,
and for Britain from the Census of Population, Economic Activity Tables,
Table 26.
Output per worker for the individual country equations was calculated
from the Census of Production. The productivity ratio for the U.S. and
British industries, which necessitated the conversion of national currencies
into a common currency, was taken from Smith, Hitchens and Davies
(1982) Table 10.1. The results reported have as the dependent variable the
ratios taken directly from that table but further trial calculations were
made with adjusted ratios. Our first adjustment was to substitute the
average price-ratio for the industrial order to those Minimum List Head-
ings given a low reliability assessment (3 or 4) by Smith, Hitchens, and
Davies. This tended to reduce the variability in the productivity ratio. Our
results using this adjusted productivity ratio for 54 industries, were of a
similar nature to those reported in the text but the co-efficient on the
difference between the stock of intermediately qualified workers in the two
countries became insignificant. Secondly, using the unadjusted produc-
tivity ratios, we omitted six miscellaneous industries; this made little
difference to the size of the co-efficients and none to their significance.
An alternative set of regression results were estimated using the more
extensive range of independent ‘control’ variables used by Davies and
Caves in their study of relative productivity performance (such as R and D
expenditure, the proportion of females and the strike-proneness of the
industry). We took their sample of 74 industries and replaced their
education variable, average years of schooling of the industry’s workforce,
with the two education variables we had calculated, the difference in ‘high’
skills and the difference in ‘intermediate’ skills. Neither of these education
variables were statistically significant in this wider analysis but there was a
positive correlation between each of these variables and relative produc-
tivity performance. This contrasted with the highly significant and positive
effect of education (measured by average years of schooling) on relative
productivity found by Davies and Caves.
NOTES
REFERENCES