【正文】 th the size of the university. In contrast, the marginal costs for Research Postgraduate students are substantially higher than both Coursework Graduate and Undergraduate students, and university size does not explain the observed variations. Of the ten largest universities on the list, three have marginal costs over$75,000 for their Research Postgraduate students, but at the same time, the remaining large universities have corresponding marginal costs of under$35, the other end of the scale, some smaller universities have marginal costs for research postgraduate students that are over$60, selection of universities have marginal costs for coursework postgraduate students that are significantly higher than research postgraduate13 students: these include Macquarie University, La Trobe University, RMIT, QUT and Monash University. This variation is likely to arise for a bination of reasons. First, large universities are not necessarily those with large proportions of postgraduate students. Second, there may be large differences in position by discipline with some universities specialising in high cost courses. Third, small sample sizes in some(smaller)universities may lead to misleading results. In general, the marginal cost of higher education is Ushaped, as in Figure 3 below, reflecting increasing returns with economies of scale, then diminishing returns derived from fixed inputs to teaching. At low student numbers, marginal costs are relatively high but this gradually declines as student numbers rise to about 1200 EFTSUs. Thereafter, the marginal costs typical marginal cost curve for Postgraduate Research students is shown in Figure 3, derived using fixed values for postgraduate coursework and undergraduate student loads for UNSW. The cost of providing postgraduate courses is higher than undergraduate courses because of the high level of specialised resources required. The distribution of students across the various disciplines is also important in explaining differences between Universities, with many universities specialising in the provision of certain courses. . The Cost of Overseas Students The cost function can be used to estimate the cost of providing higher education to overseas students. However, information on the costs of higher education in Australia is available only for the total number of students, that is, overseas plus domestic students, at a particular date. No data are available relating to the cost that would be incurred if only domestic students were being educated. This means that estimates of the extra cost involved in educating overseas students can only be obtained by using university cost functions and evaluating the costs under alternative assumptions about the number of students. The quality of such estimates is therefore necessarily constrained by the properties of empirical cost functions. The cost function is applied to domestic student data to predict the average and marginal costs for domestic students. Corresponding values for overseas students are obtained as the residual values of the cost estimates for all students and for domestic students only. The results of this exercise for 1997 are shown in Table 5. Universities across Australia vary widely in the distribution of student load between domestic and feepaying overseas students. In general, the overseas student population prise less than five per cent of the student population in most regional universities, but are much more important in universities located major city centres .In 1997,the top five universities, in terms of numbers of fullfee paying foreign student loads are Monash University, RMIT, the University of Melbourne, UNSW and Curtin University of Technology. However,in relative student shares, RMIT is highest with 29 per cent of its student load from overseas, followed closely by Curtin University with foreign students prising 27 per cent of its total student load. There is a positional issue in generating costs for overseas students. If the cost structure of overseas students is projected from information about all students and domestic students, the implicit assumption is made that the discipline position of overseas students is the same as that of domestic students. However, this is not the case. Data from DETYA,1998 show the discipline position of overseas students. In 1997 about half of all overseas students were undertaking courses in the Business, administration and economics field of study whereas among domestic students the proportion undertaking courses in this discipline was about 20 percent. Since business, administration and econo