^{(note 1 in page 48)} E. F. Denison, Why growth rates differ (The Brookings Institution, Washington, 1967); chapter 7 (pp. 78-108) is entitled ‘Education of the labour force’, and Appendix F (pp. 373-403) is devoted to the statistical derivation of the estimates. The other countries included in the study were Belgium, Denmark, France, Germany, Italy, Netherlands and Norway. See also his earlier contribution, ‘Measuring the contribution of education to economic growth’ in The residual factor and economic growth (OECD, Paris, 1964), p.22.

^{(note 2 in page 48)} A. O. Krueger, ‘Factor endowments and per capita income differences among countries’, Economic Journal, September 1968, p. 641.

^{(note 3 in page 48)} Central Advisory Council for Education (England), 15 to 18 (The Crowther Report) HMSO, London, 1959, vol. 1, p.124.

^{(note 4 in page 48)} Committee on Higher Education, Higher education (The Robbins Report), HMSO, London, 1963, p.6.

^{(note 5 in page 48)} Department of Education and Science, Education for 16-19 year olds, HMSO, London, 1981, p.1.

^{(note 1 in page 49)} For recent surveys see for example M. Blaug, ‘The empirical status of human capital theory: a slightly jaundiced survey’, Journal of Economic Literature 1976; G. Psacharo poulos, Returns to education, Elsevier Scientific Publishing Company Amsterdam, 1973, and idem, Returns to education: an updated international comparison, World Bank Staff Working Paper no. 402, Washington DC World Bank, 1980.

^{(note 2 in page 49)} Denison, op. cit., p.95.

^{(note 1 in page 50)} Denison op. cit., p.95.

^{(note 2 in page 50)} Data on apprenticeship qualifications are from Bevöl kerung und Erwerbstätigkeit Fachesrie 1, Reihe 4.2.2 Beruf, Ausbildung und Arbeitsbedingungen der Erwerbstätige 1978, p.31. An attempt to broaden Denison's measure of education for the American case was made by A. C. Bolino Career education, contributions to economic growth (Praeger Pub lishers, New York, 1973). The extension goes beyond that described above for Germany, to include adult education, apprenticeships, correspondence schools, military training and on-the-job training. The result of these calculations was to raise the average years of schooling held by the American population to 10.66 years in 1960, an increase of 0.7 years compared with the unadjusted average in table 1 (see Bolino, p.188). These calculations are however, based on generous assumptions.

^{(note 3 in page 50)} Ibid, p.380.

^{(note 4 in page 50)} A comparison by Professor Phelps-Brown of certain occupational groups in Britain and France over a similar period shows much smaller differentials in Britain than in France. Basing himself on a United Nations study of earnings in Europe (which relied, in turn, on national sources) he noted that the more highly qualified occupations (engineers, pro fessionals and higher administrators) earned about 2.7 times the average in France compared with 2.1 times the average in Britain; while unskilled manual workers were relatively worse off in France, earnings 50 per cent of the average compared with 60 per cent in Britain. H. Phelps Brown, The inequality of pay (Oxford University Press, Oxford, 1977), table 2.1, p.32.

^{(note 1 in page 51)} Ibid, pp.91, 105.

^{(note 2 in page 51)} Ibid, p.106.

^{(note 3 in page 51)} A. O. Krueger, op. cit., p.652. A number of similar studies have been critically reviewed by M. Blaug, An intro duction to the economics of education (Penguin, 1970), chapter 3.

^{(note 1 in page 52)} The data were taken from General Register Office, Census 1961, England and Wales, Education Tables (HMSO, London, 1966), table 1; General Register Office, Edinburgh, Census 1961, Scotland, vol. 9, Terminal education age (HMSO, 1966), table 1. Data were not available for Northern Ireland. Our calculation assumed that years of schooling equals terminal age minus five (the legal age for commencing school in Britain). It was not possible to separate urban and rural residents without unduly reducing the number of categories relating to terminal age of education; as the difference in educational attainment between rural and urban regions in Britain was less marked than for the US, and as the rural population is relatively small in Britain, tables for the population as a whole are probably adequate.

^{(note 2 in page 52)} Partly also the difference arises because Krueger's cal culation related to the whole population, whereas Denison's related to the labour force; but this does not explain much of the gap.

^{(note 3 in page 52)} If we adopt Denison's ‘three-fifths assumption’ that only 60 per cent of the income differential between groups with varying amounts of schooling is a result of these educational differences, the estimate of human capital in the British population was 95 per cent of that in America in 1960.

^{(note 4 in page 52)} Cf. the similar conclusion reached by G. Psacharopoulos, based on calculations relating to a sample of seventeen countries at the end of the 1960s, that ‘The contribution of education to growth is generally higher in the less-developed countries’. (Returns to education, Elsevier, 1973, p.117.) His result however depends on his choice of measure of the contribution of education to economic growth and is doubtful even on this basis. A correlation between the logarithms of the variables he examines (per capita income and the con tribution of education to growth) turns out, on our calculation, to be not significant (r = —0.36, P > 10 per cent); and the opposite, and significant, conclusion is yielded by an alter native—‘Denison’ scheme—measure of the contribution of education to growth that he gives on the same page (taking the sixteen countries common to the tables on his p.85 and col. 3 of p.116, we calculated r = +0.52, P ≃ 4 per cent).

^{(note 5 in page 52)} The British data are from the General household survey of 1978, and relate to all persons between 16 and 69 not in full- time education (see table 6.11(b), p.116 of 1978 GHS). The American data are from Money incomes in 1977 of families and persons in the US. Current population reports, Consumer income series, p.60, no. 118 (US Department of the Census, Washington DC). The income weights used here relate to the incomes of all persons over 18 (see table 47 of the Report). As the American data covered all persons between 16 and 69 including those still at school, it was necessary to adjust the British data to include those still in full-time education. This was done by allocating the difference between the total number in the GHS sample in the 16-19 and 20-24 age groups and the numbers who had left school, between different schooling groups. (See table 2.31, 1978 GHS, p.21 for the population figures.)

^{(note 1 in page 53)} Tables of earnings by years of schooling have not been published by the General Household Survey, but cross tabulations for all males over 16 and also over 25 in 1975, were kindly made available to us from a GHS tape held at the Centre for Labour Economics at the London School of Economics.

^{(note 2 in page 53)} R. B. Freeman, The over-educated American, Academic Press, New York, 1976, p.185.

^{(note 1 in page 54)} Qualification levels can be roughly compared with years of schooling as follows: no qualifications, CSE, other grades etc., GCE O level, CSE grade 1—9—11 years of schooling; GCE A level, below degree higher education—13-15 years; and degree or equivalent—16 years.

^{(note 2 in page 54)} The decline in the relative earnings of degree holders cannot be explained by the larger proportion of young people in the 1978 total compared with 1972, since the indexes calculated in table 5 are weighted averages based on a constant age distribution (see footnote (c) to the table).

^{(note 3 in page 54)} Op. cit. pp. 60-79. Freeman argued that lags between the recognition of labour market signals and the ability to act upon them created an unstable situation in the market. If a large number of students opted for a course on the basis of the relatively high salaries of practioners in that field, this tended to lower salaries subsequently and, in turn, deterred the next generation of students from entering that field. The ensuing shortage then led to higher salaries once again, which encouraged students to enter the field and so the ‘cobweb’ cycle recommenced.

^{(note 1 in page 55)} Denison, op. cit., p. 104.

^{(note 2 in page 55)} The percentage difference in per capita incomes explained by human capital differences was calculated by Krueger as (100 — X) ÷ (100 - Y) where X is the amount of human capital per head in the country in question relative to that in America and Y is the per capita GDP of that country as a percentage of US per capita GDP. Britain's per capita income was taken from Denison, op. cit., p. 22, and on American weights was, in 1960, 72 per cent of America's. We estimated earlier that the amount of human capital in the UK population was 92 per cent of that in America (see pp. 9-10).

^{(note 3 in page 55)} Krueger, op. cit., p. 655.

^{(note 4 in page 55)} See, for example, M. Blaug An introduction to the economics of education op. cit., pp. 89-100.