Empirical results Directory UMM :Data Elmu:jurnal:E:Economics of Education Review:Vol20.Issue1.2001:

20 D. Rochat, J.-L. Demeulemeester Economics of Education Review 20 2001 15–26 investments. Proxying lifetime earnings prospects into the educational choice equation does not mean that our model is at odds with the human capital framework. Indeed age-earning profiles show that starting salaries at least partially reflect future earning differentials see Woodhall, 1987; Demeulemeester, 1995 for evidence on the Belgian labour market. We also introduced a measure of the easiness of inser- tion into the labour market for young graduates by orien- tation. 13 These statistics were taken from Demeulemees- ter and Rochat 1995. This measure might be seen as a complement to our measure of expected future benefits besides wage. Ceteris paribus, we expect that students will prefer orientations whose graduates are perceived to insert more easily on the labour market. Finally, we also introduced the legal minimum length of studies to get the final degree. This will allow us to take into account the risk linked with the increased length of studies, as well as the direct and indirect costs of studies. We expect the latter to be deterrent.

4. Empirical results

The first part of the analysis is a multinomial logit estimation of discipline choice, where explanatory vari- ables are only of socio-demographic and ascriptive nat- ure. The results are presented in Table 3 below. They have to be interpreted with reference to the first orien- tation, i.e. “short cycle economic and social vocational non universitary higher education”. The latter encompasses various types of studies, as accountancy, computer sciences, social assistant and is quite represen- tative of the basic, average clerk worker in Belgium the lower stratum of the middle class. Our findings clearly highlight the role of variables such as age negative and significative impact in all orientations but the second, suggesting that the older the student, the less likely he will choose lengthier orientations as university education and gender men typically choose less fre- quently short cycle orientations of a non-economic nat- ure, and prefer engineering and technical subjects. These results are perfectly in line with the theoretical results obtained by the investment in human capital literature in a life-cycle perspective Ben-Porath, 1967. One can also note the influence of father holding an “e´lite” occupation 14 as the professions, high ranked civil 13 These statistics are based on the number of graduates rela- tive the number of unemployed in the same field. Both the entry salary and easiness of insertion are calculated per discipline from the aforementioned sources whereas probabilities of suc- cess by discipline are measured at individual levels. 14 This variable can also proxy the social capital available to the student. servants or top manager in the private sector on students choosing engineering or business-related fields law, economics, business... rather than short cycle economic curricula. Similarly, higher parental incomes possibly linked with positions in the business or legal spheres favour the choice of Economics and Business as well as Legal orientations. The two latter results might suggest a possible link between specific parental occupation and students orientation choices. Finally, parental education proxied by mother or father holding higher education degrees seems also to favour the choice of long-cycle social, economic, legal and literary orientations over short cycle economic ones. All these effects suggest the pertinence of the rolemodels approach developed by sociologists and development psychologists, as well as of human capital investment arguments parental edu- cation or profession proxying for a higher income and less severe budget constraints enabling the students to choose longer or riskier studies, see Becker, 1967. The type of curriculum followed while in high school also influences the choice orientation, in the sense that graduates from “math-intensive” sections typically prefer scientific, medical, engineering or economic subjects at university over short-run economic cycles, while the same is true for “classics-intensive” graduates in literary orientations of the university. 15 Shortly stated, these results, besides the role of gender linked with the role models approach and the assignment of tasks between men and women and age, highlight the influence of prior experiences and models type of curriculum fol- lowed while in high school, influence of parental sphere of activity on the choice of disciplines made when entering university. The second part of the analysis aims at ascertaining the main contributive factors of success in first year of post-compulsory education in each of the seven retained orientations, after a correction for the potential selection bias. The results are presented in Table 4 below. Some variables are significant in the long-cycle and university education orientations only. This is the case for the vari- able accounting for age. Being more aged reduces the probability of success in Natural and Medical Sciences, Engineering and Economics and Business orientations. This suggests the role of the depreciation of human capi- tal in disciplines which heavily rely on the mathematical 15 There are no entry requirements for the technical studies in Belgium, except for the narrow segment of engineering at universities. The students have to pass an entrance examination based upon prior knowledge of mathematics. Students coming from “math-intensive” sections in high school have higher chances of success, but nothing impedes the students from other sections to take the examination. Very often, however, these students take a complementary year devoted especially to math- ematics. 21 D. Rochat, J.-L. Demeulemeester Economics of Education Review 20 2001 15–26 Table 3 The determinants of educational choices: MNL estimations n = 641 Orientation a Variable 21 31 41 51 61 71 Coeff t-stat Coeff t-stat Coeff t-stat Coeff t-stat Coeff t-stat Coeff t-stat Constant 20.353 20.62 0.260 0.45 20.227 20.39 21.880 22.57 0.276 0.57 0.740 1.48 Gender 21.440 23.94 20.703 21.96 20.042 20.13 2.390 4.49 0.007 0.03 20.344 21.07 Age 0.637 1.65 20.976 22.64 21.530 24.59 21.730 24.45 21.240 24.03 21.080 23.19 Nationality 20.515 21.16 20.206 20.42 0.441 0.89 20.197 20.38 20.170 20.42 21.046 22.60 Latin 0.591 1.33 0.035 0.07 0.006 0.02 20.892 21.57 0.550 1.43 0.828 2.06 Mathematics 0.559 1.27 0.534 1.19 1.022 2.67 1.627 3.93 0.980 2.68 0.345 0.83 Single parent family 0.351 0.95 20.291 20.70 0.059 0.17 20.390 20.89 0.028 0.08 20.129 20.35 Father’s education 0.394 1.04 20.027 20.07 0.457 1.28 0.551 1.30 0.436 1.31 0.710 1.95 Mother’s education 0.347 0.88 0.239 0.59 0.467 1.29 0.277 0.64 0.558 1.66 0.651 1.77 Father with “Elite” occupation 20.279 20.45 1.027 1.95 0.572 1.17 1.107 2.13 0.930 2.08 0.736 1.54 Both parents work 0.138 0.38 0.601 1.59 20.159 20.47 20.193 20.49 20.340 21.09 20.244 20.72 Household’s income 20.177 20.44 20.528 21.24 0.311 0.85 0.146 0.34 0.695 2.05 0.131 0.35 Log likelihood at convergence 21068.6 a significant at 1 level; significant at 5 level; significant at 10 level. All results should be interpreted with respect to the orientation “1”. D. Rochat, J.-L. Demeulemeester Economics of Education Review 20 2001 15–26 Table 4 Determinants of success by orientation: empirical results a Orientation 1 Orientation 2 Orientation 3 Orientation 4 Orientation 5 Orientation 6 Orientation 7 Variable n = 90 n = 70 n = 60 n = 102 n = 65 n = 157 n = 97 Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Constant 22.02 22.48 1.701 2.03 22.206 21.30 21.37 20.92 5.537 1.64 0.25 0.27 22.796 23.19 Age 0.72 1.31 - – – – 22.51 22.17 22.174 22.61 21.55 23.99 – – Gender 20.007 20.02 0.875 1.51 21.539 21.95 20.141 20.41 23.962 21.45 20.337 21.19 20.393 20.91 Repetition during high school 20.324 20.81 20.202 20.53 – – – – 0.937 0.76 – – 1.751 1.45 Prior studies 20.694 20.70 0.618 0.59 1.796 1.45 – – – – 0.61 1.11 20.206 20.27 Nationality 1.557 3.23 20.428 20.74 2.741 1.86 20.31 20.39 22.33 21.81 20.43 21.10 2.056 3.72 Number of Siblings 20.066 20.45 20.31 22.60 20.615 22.16 0.048 0.39 20.095 20.32 20.055 20.56 0.027 0.16 Change in living arrangements 20.5 21.36 20.43 21.01 2.748 1.61 0.098 0.29 20.539 20.66 0.247 0.86 0.534 1.42 Scholarship 1.061 2.53 0.539 1.10 2 2 0.059 0.14 2.47 2.33 0.367 1.18 1.389 2.79 Job while studying 20.23 20.56 21.163 22.37 1.021 1.25 20.59 21.22 20.53 20.64 20.276 20.74 20.745 21.51 Single parent family 20.6 21.43 20.213 20.53 2 2 20.089 20.22 20.3 20.32 20.59 21.87 20.712 21.43 Father’s education university 0.56 0.87 0.890 1.51 2 2 1.33 3.68 2.67 3.18 0.105 0.35 20.085 20.20 Mother’s education university – – 20.729 20.71 1.481 1.23 – – – – 1.172 4.07 2.367 4.47 Father with “Elite” occupation 20.53 20.66 0.304 0.31 20.444 20.32 20.53 21.07 0.754 0.83 0.396 1.23 0.211 0.49 Both parents work 0.637 1.80 20.278 20.68 0.362 0.53 20.194 20.54 1.145 1.31 0.193 0.75 20.5 20.97 More aged due to prior studies 0.954 0.91 20.263 20.24 20.962 20.71 1.139 1.22 – – 0.064 0.09 0.021 0.03 More aged due to repetitions – – – – – – 0.561 0.63 21.49 20.90 – – 22.743 22.04 Selection variable 20.786 20.95 20.042 20.06 20.889 20.56 22.23 21.08 0.924 0.92 20.445 20.24 21.732 21.55 Log likelihood at convergence 245.81 235.30 213.13 241.41 213.59 274.10 235.54 Percentage correct predictions 74.44 74.28 88.33 82.35 90.76 74.52 81.44 a significant at 1 level, at 5 level and at 10 level. Some variables were dropped of the regressions for convergence purpose. 23 D. Rochat, J.-L. Demeulemeester Economics of Education Review 20 2001 15–26 and scientific prior knowledge acquired while in high school Ben-Porath, 1967. Parental education seems also to foster academic suc- cess in long-cycle and university orientations. Typically, having father holding university degree promotes success in scientific orientations as Medical and Natural Sciences or Engineering while having mother with university degree seems to foster the chances of success in more literary subjects Law, Social sciences and Humanities where a good command of the language is essential. Lei- bowitz 1974 showed indeed the essential role of the mother in transmitting verbal skills and literacy. The lat- ter effect relating to the importance of mastering the mother tongue, may also partially explain why being Belgian increases the probability of success in Econ- omic, Social, Pedagogical and Artistic short cycles as well as in the Humanities orientation of the university. The negative impact of being more aged due to rep- etitions while in high school on success in Humanities might also be partially explained by a common factor, namely a lack of a good command of his own language see for instance, Ribar, 1993. Our results also highlight the effect of some variables in some peculiar orientations. For instance, the family size as well as the fact of working while studying are both found to exert a negative effect on the probability of success in two orientations for family size 16 and one — at least at the 5 level — for a job market partici- pation during studies. The sign of the coefficient found in some orientations for the variable accounting for the matrimonial status of the student’s parents seems to be in line with the idea that having two parents in a family foster normal personality development Seltzer, 1994. 17 The sign of the coefficient found in some orientations for the variable accounting for the fact that both parent work can be interpreted in a “working mother perspec- tive” where increased parental income might outweigh the reduction in students care time Hetherington et al., 1983. Finally, coming from a poorer background — and for this reason obtaining a scholarship — seems to play a rather positive role on success in at least three orien- tations. This result might be explained through the motivational lines of reasoning of Tinto 1975, 1987 if one considers the concession of a scholarship as a form of contract. 18 16 This result illustrates the idea of a reduced transfer of human capital on each children Hanushek, 1992 due to time constraints see, Becker, 1965. 17 The presence of two parents might strengthen parental con- trol and monitoring, and weaken thereby the potential adverse influence of other role models. 18 The variables accounting for the professional position of student’s father, change in student’s living arrangements or pre- vious higher education studies made before starting current cur- riculum were found to play a significant role in explaining aca- demic success. We now turn to the last step of the analysis which should enable us to ascertain the validity of Mingat and Eicher 1982 thesis concerning the relative weights given by students to risk and return components in the orientation choice process. The results of the estimated conditional logit model are presented in Table 5 below. We test for the significativity of a priori probability of success on the orientation, i.e. whether prospective stu- dents tend to choose disciplines where they have the highest probabilities of success given their socio-demo- graphic and ascriptive characteristics — while con- trolling for length of studies and expected economic returns wage and insertion. For the complete sample of 641 students, it seems indeed that youngsters pay atten- tion not only to expected economic benefits, but also to the length of studies and the mere probability of suc- ceeding in the chosen orientation. In this sense, the mod- elling approach of Mingat and Eicher 1982 seems parti- cularly suited to the actual behaviour of the prospective students. When one concentrates on the poorest students i.e. the 171 students holding a scholarship, one is indeed confronted with a decisional structure where they are more risk-averse as they have less financial wealth and are therefore less inclined to take risk. Indeed expected chances of success receives a significant weight, while this is not the case for wage or length of study. The only economic reward which seems to interest those stu- dents is the expected greater ease of entry on the labour market insertion. These results perfectly match our theoretical a priori expectations and are even reinforced by the results found for the wealthiest students of the sample. Indeed, the richest students defined as those coming from household with high level of income and with father holding university degree have a very orig- inal choice behaviour, in the sense that they seem to fol- low just their own preferences. They do not appear to be sensitive to either the expected chances of success or the economic benefits linked with their orientation choice. 19 We also try to test whether ability could be the source of the risk-averse behaviour of the students. We therefore turn to the case of the brightest students i.e. those with the ex ante highest chances of success in all the disciplines. One is struck by their keen interest in the financial benefits of their educational investment: they give heavy weight to wage. However, as expected, they do not seem to pay much attention to their expected chances of success. This is quite intuitively appealing as such event — failure — is quite improbable for them, whathever their choice. And, ceteris paribus, they are also attracted in shorter studies. Finally, when one con- centrates on the “dullest” students, i.e. those with the 19 This sub-sample of students is only made of university stu- dents. D. Rochat, J.-L. Demeulemeester Economics of Education Review 20 2001 15–26 Table 5 Effect of expected success rate a Full sample Top 10 of Top 15 of Bottom 10 of Bottom 15 of Scholarship Wealthier students n = 614 students b students b students c students c students n = 171 d n = 113 e Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Coeff. t-stat Expected success rate 0.494 3.86 20.146 20.257 0.025 0.06 20.050 20.08 0.191 0.412 0.626 2.52 20.073 20.191 Wage in BFR 0.36E 25 3.69 0.812E-5 2.81 0.748E 253.13 0.206E 244.75 0.228E 24 5.33 0.214E-5 1.11 20.75E-7 20.039 Graduates insertion on the labour 0.0084 1.636 0.0172 0.96 0.00432 0.30 0.056 2.20 0.041 1.99 0.0162 1.74 0.00203 0.159 market Length of studies in Years 20.751 23.687 21.723 22.13 21.388 22.69 25.647 26.03 25.826 26.88 20.563 21.38 20.308 21.04 Log likelihood at convergence 21227.49 2120.29 2175.70 285.86 2128.57 2326.37 2155.76 a significant at 1 level, at 5 level and at 10 level. b The 10 or 15 of students who have an expected probability of success above a common threshold across all disciplines. c The 10 and 15 of students who have the lowest expected probabilities of success in each orientation. d Students whose household’s revenue is such low that they benefit from a governmental financial aid. e Students whose parents’ monthly net income is above 100,000 BEF and whose father holds a university degree. 25 D. Rochat, J.-L. Demeulemeester Economics of Education Review 20 2001 15–26 lowest probabilities of success in every discipline, one is also struck by their interest in the expected economic benefits as wage and ease of insertion. However, they do not seem to pay much attention to their expected chances of success which are low in any cases. Those results seem to indicate that ability may be less important than the socio-economic background in explaining a risk- averse behaviour, with the related avoidance of more dif- ficult but more remunerative and demanded orien- tations. Our results tend to confirm those of Cannings et al. 1993 although ours are obtained within a generalized structure allowing us to jointly test for the importance of ex ante probability of success and expected economic benefits. Cannings et al. 1993 found that the choice of college concentration depends on the perceived prob- ability of success in a particular concentration for all of the students. Similarly to our own findings they put for- wards that brightest students do not pay attention to their perceived probability of success. However, contrarily to our results and quite astonishingly, they find that stu- dents coming from high-income families appear to be more risk-averse in their choice of major than students from low-income family. Globally, the overall picture seems to support quite well Mingat and Eicher 1982 theoretical contentions, namely the fact that students do take into account two dimensions of their prospective educational choice economic returns and a priori chances of success, and that poorest students give a heavier weight to the risk component. However, it might also be true that less priv- ilegiated students tend to avoid disciplines where the post-studies performance requires more social capital Coleman, 1988. But we concentrate on the perform- ance during higher education studies, and in such a con- text we have already tried to assess the impact of those variables on the probabilities of success across orien- tations.

5. Concluding remarks