(Reprinted from HKCER Letters, Vol.45, July 1997)
What Can We Learn From Morocco?
Try to name something that Hong Kong and Morocco have in common, and you might come up with a blank. Perhaps you believe that there is not much we can learn from that remote country. This is a grave mistake. Every educator, student, and parent in Hong Kong should draw some lessons from a social experiment that took place in Morocco fourteen years ago, for Hong Kong is on the verge of performing a similar experiment.
The Education Department of Hong Kong has announced plans to force the majority of secondary schools to switch their language of instruction from English to Chinese (Cantonese). This change is to go into effect beginning in the 1998-99 school year and will arguably be the biggest education policy change in the history of Hong Kong. Yet the decision has been made with surprisingly little consultation. Proponents of the change have long argued that students can learn more effectively when school instruction is in their mother tongue. Parents are understandably wary of any change that may have a negative impact on their children s acquisition of the English language. In the past, gentler efforts to switch the language of instruction have largely failed because of resistance from parents.
Will a change in the language of instruction facilitate the learning process? Will it lower the standard of English among students? Theories abound, but evidence is scarce. There are small-scale studies that compare learning outcomes of students who receive their education in their native language and those who are taught in a foreign language. However, most studies look at short-term outcomes (e.g., test scores) instead of long-term effects, and we know virtually nothing about the potential ramifications when an entire school system (as opposed to individual schools or classrooms) makes a transition in the language of instruction.
Fortunately, however, we have not been left completely in the dark, providing that we are willing to learn from the experience of North Africa. As part of the decolonization process, French was replaced by Arabic as the main language of instruction in Moroccan middle schools and secondary schools in the 1980s. Algeria and Tunisia embarked on similar programs more recently. Two researchers, Joshua Angrist of Massachusetts Institute of Technology and Victor Levy of Hebrew University, have recently published a paper in the Journal of Labor Economics (January 1997, Part 2) that examines the effects of the language transition in Morocco on language ability and on labor market outcomes. Educators should take note.
The Angrist-Lavy Study
Morocco gained independence from France in 1956. Until the autumn of 1983, French was the language of instruction for all subjects taught in Moroccan post-primary schools, except for Arabic literature and religion. The language of instruction in primary schools was and remains Arabic.
The replacement of French by Arabic as the language of instruction was first implemented in the 1983-84 school year for all students entering middle school (grade six). The change then rolled ahead one additional grade at the beginning of each new school year. By 1989-90 the transition had been accomplished in all middle and secondary school grades.
Data for the Angrist-Lavy study were drawn from two surveys conducted in 1990-91. Individuals aged twenty-one and younger who were interviewed in 1991 were potentially affected by the Arabization policy: those who had any post-primary education normally would have entered grade six after the language transition policy had gone into effect. On the other hand, individuals older than twenty-one were unlikely to be affected. Among these older individuals, those who had post-primary education can be presumed to have been taught in French. A comparison of these two groups of individuals can therefore be used to assess the effects of the Arabization policy. The main results are summarized in Table 1.
Panel (A) of Table 1 shows a simple comparison of the average wage premium for post-primary education of individuals aged twenty to twenty-one and younger and individuals aged over twenty to twenty-one. Among the younger group, those who had post-primary education were taught in Arabic, and their hourly wage was on average about 14 percent higher than that of individuals who had primary education only. (The premium in the logarithm of the wages is roughly equivalent to the percentage difference in wages.) Among the older group of persons, those who had post-primary education were taught in French, and their average hourly wage was more than 100 percent higher than that of those who had primary education only. In other words, as measured by the wage premium, the value of post-primary education conducted in French was about 93 percentage points higher than post-primary education conducted in Arabic.
Panel (B) improves on the estimates shown in Panel (A) by using a regression model to control for other differences between the younger age group and the older age group. For example, line (2) of the panel estimates that individuals who received their secondary education (grades ten to twelve) in Arabic earned 50 percent more than similarly aged persons who only went to primary school (grades one to five). In contrast, individuals who received their secondary education in French earned 71 percent higher wages than similarly aged persons who only went to primary school. The Arabization policy apparently reduced the value of secondary school education by 21 percentage points.
One convenient way to measure the value of education is to estimate its internal rate of return. This approach considers schooling as a form of investment in human capital and estimates the returns to such investment by using a regression model. Panel (C) of Table 1 shows the relevant comparisons. Line (1) shows that the rate of return to primary education among the younger age group was 3.6 percent, while that for the older age group was 6.2 percent. This difference of 2.6 percent between these two groups is statistically insignificant (i.e., we cannot rule out with great confidence the possibility that the difference is simply the result of sampling error). The lack of a significant difference in the rate of return to primary education is not surprising, because primary schools were not directly affected by the language transition policy. The language of instruction in primary school has always been Arabic. Post-primary education, however, was directly affected by Arabization. Line (2) of panel (C) shows that the difference in rate of return to post-primary education between the younger and older age groups was much larger than 2.6 percent. Moreover, this difference is statistically significant. The inference is that Arabization has reduced the investment returns from post-primary education by more than half (from 11.7 percent to 5.2 percent).
To explore why the change in language of instruction was associated with a fall in the rate of return to schooling, Angrist and Lavy also utilized information from a literacy survey. Individuals were given a battery of tests on a set of basic skills. A test score of 0 indicates little or no ability, a score of 1 indicates some ability, and a score of 2 indicates functional competence. As one would expect, a higher level of education will tend to increase the average test scores--the interesting question is, by how much? A large difference in average test scores between individuals with middle school education and individuals with primary education would suggest that middle schools have large value added in terms of the skills imparted to their students.
Panel (D) of Table 1 displays this measure of value added for various skills and various education levels. For example, line (1a) shows that, among the younger age group, middle school education added on average 0.42 points to the test scores on French writing ability. Among the older population, middle school education was much more valuable: it added an average of 1.05 points to the test scores on French writing ability. This difference is large quantitatively and is significant statistically. It indicates that the Arabization policy in middle schools was associated with a substantial loss of French writing ability, a valuable skill in the Moroccan labor market.
Lines (1b) to (1e) of panel (D) seem to suggest that Arabization increased the value added from middle schools in terms of abilities other than French writing. The difference, however, is illusory. Angrist and Lavy find that the increases shown in column (3) are statistically indistinguishable from zero; they could simply have arisen from sampling variations.
The comparison of the value added from secondary school education shown in lines (2a) to (2e) tells a similar story. The change of the language of instruction in secondary schools was accompanied by a substantial fall in French writing skills. There was virtually no change in other abilities (the differences in test scores were negative and statistically insignificant). In particular, there is no evidence that changing the language of instruction to Arabic brought about an improvement in Arabic writing and reading abilities.
The Importance of English Language Skills
Just as knowledge of the French language is crucial for labor market success in Morocco, English language ability is crucial for success in the Hong Kong labor market. There is a mountain of econometric evidence indicating that English proficiency increases earnings for immigrants in the United States. Closer to home, an unpublished research paper by Samuel Lui of Lingnan College and myself shows that the ability to speak English carries an earnings premium in the labor market of about 26 percent. We also show that the rate of return to education received in China is substantially lower than the rate of return to education received in Hong Kong. Part of the reason for this is possibly related to the fact that the languages of instruction used in the two school systems are different.
Few people in Hong Kong would need to be convinced that graduates from English language schools face better prospects in the labor market than graduates from Chinese language schools do. Many supporters for a change in the language of instruction, however, argue that the difference is due to a selection effect. They believe that English language schools in Hong Kong get, rather than make, better students. The Moroccan experience tells us that the selection effect cannot be the entire story. When the whole country changed its language of instruction to Arabic, the rate of return to education as well as the level of French writing ability fell.
There is one aspect of the Morocco experiment that needs to be noted. Even after independence, teachers in Moroccan schools were often French citizens or natives who had been trained in French schools. It was only in the 1960s that the Moroccan government began to initiate a policy to replace foreign teachers with Moroccan teachers. In contrast, most teachers in Hong Kong are local citizens whose native language is not English. While the French teachers in Morocco might have been effective at using French as the language of instruction, Hong Kong teachers may not be as effective at using English as their teaching medium.
Labor market success is not the only relevant outcome of an education system. Language reform in school has cultural and political implications as well as implications for the personal development of students. The decision by the Education Department to force a change in the language of instruction would have been more palatable if it had been adopted after more careful deliberation on its potential costs and benefits. A social experiment of this magnitude needs to draw inputs from all possible sources. While parents do not necessarily know best, it cannot be presumed that policy makers know better. We all have a lesson to learn from Morocco.
Dr. Wing Suen is senior lecturer in the School of Economics and Finance, the University of Hong Kong.
Table 1 Effects of the Change in Language of Instruction in Morocco
persons persons not affected by affected by policy policy difference (1) (2) (3)
(A) log wage premium for post-primary education 0.14 1.07 -0.93 (B) log wage premium [regression estimate] (1) middle school education 0.06 0.30 -0.24 (2) secondary school education 0.50 0.71 -0.21 (3) post-secondary education 1.09 1.48 -0.39 (C) rate of return to schooling (1) primary education 0.036 0.062 -0.026 (2) post-primary education 0.052 0.117 -0.065 (D) difference in test scores (1) middle school minus primary school (1a) French writing 0.42 1.05 -0.63 (1b) French reading 1.02 0.72 0.30 (1c) Arabic writing 0.94 0.70 0.24 (1d) Arabic reading 1.03 0.73 0.30 (1e) mathematics 0.71 0.57 0.14 (2) secondary school minus primary school (2a) French writing 0.71 1.47 -0.76 (2b) French reading 0.92 0.94 -0.02 (2c) Arabic writing 0.60 0.77 -0.17 (2d) Arabic reading 0.93 0.94 -0.01 (2e) mathematics 0.81 0.94 -0.13
Note: All figures are derived from Angrist and Lavy s paper.(A) comes from the discussion in the text on page S56. (B) and (C) are obtained from columns (1) and (5), respectively, of Table 3. (D) is derived from columns (3)-(7) of Table 4.