degree level personnel, evidenced not least by the slow progress of many localisation
programmes and the difficulties of recruitment for industrial concerns and in the public
sector. Second, it may be the case that the problem is at least as much one of pervasive
inefficiency and marginal relevance in higher education investment as it is one of over-
investment. Though there is very little evidence on declining quality in higher
education and few studies have taken place (King 1991) there is widespread disquiet
about the orientation and effectiveness of higher education, especially in Africa (Bown
1992, Mazrui 1992).
Third, it is to be expected that, in those countries which are approaching universal
enrolments at primary levels, the percentage of the budget allocated to higher levels
will grow as a result of increased post-primary provision. Of more concern are those
cases where changes in unit costs disproportionately favour higher education. In Sub-
Saharan Africa recurrent unit costs fell at primary and secondary level in the early
1980's, but appeared to increase for higher education. Median values in constant 1983
USS fell from 67 to 48 at primary level; 362 to 223 at secondary; and rose from 2,462
to 2,710 at tertiary level over the period from 1970-1983. These aggregate figures
require cautious interpretation since the number of countries for which data are
available differ at each educational level (World Bank 1988:141-3). Since 1983 the
financial position of higher education has probably worsened substantially and the trend
may have been reversed. Where structural adjustment has taken place this has often
sought to limit the growth of higher education subsidies though success in achieving
this has been mixed. It is probably a minority of countries that have actually seen higher
education as a whole grow more slowly than other levels. This has usually been where
there has been a strong policy preference in this direction e.g. in Malawi and Tanzania
which have deliberately limited the supply of post primary places and have resisted the
pressures of excess demand, at least in so far as the publicly financed elements of the
system are concerned.
Fourth, data on rates return and the conclusions that can be drawn from it has to be
treated with caution for many well known reasons which have long been discussed in
the literature (Hough 1992). The rates quoted above are historic rates many of which
are based on relatively small samples taken more than a decade ago and which contain
assumptions that may no longer be robust. Distortions in income patterns, related to
structural characteristics of the labour market which may under or over value the
contributions to production of different groups of workers, are also common and make
reliable calculation difficult. There are apparently some consistent patterns that show,
for example that the differences in rates of return between levels do tend to collapse
over time as development takes place. This is almost inevitable. As more individuals
acquire a particular level of education the marginal value is likely to decrease and
"wage compression is likely to occur (Knight and Sabot 1990).
Fifth, it is debatable how significant reducing subsidies to higher education might be.