Education for all: quasi-experimental estimates of the impacts of compulsory primary education in Hong Kong


This paper examines the causal impacts of Hong Kong’s 1971 policy of free compulsory education on students’ educational attainment. Using a regression discontinuity method and Hong Kong Census data, this study compares children born just before and just after the month in which the compulsory-education law came into effect. The results show that the law reduces approximately 10 percent of the dropout probability by age 12 and 8 percent of the dropout probability by age 15. The effect is substantial considering that approximately 90 percent of primary-school-age children were already in school in Hong Kong when the policy was implemented. The policy has larger impacts for socially and economically disadvantaged children, and the law also increases their probability of obtaining an education beyond middle school. Robustness is tested with several sensitivity checks. The results will help policy makers and stakeholders better understand the potential efficacy of mandated education policies and increasing educational access to targeted populations, especially for developing economies.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3


  1. 1.

    United Nations Development Programme, The Millennium Development Goals: Eight Goals for 2015, available at

  2. 2.

    In 2008, Hong Kong extended its free education to 12 years. Policy makers have begun to debate whether the Hong Kong Government should extend free education to 15 years by including 3-year pre-primary education. But these extensions of free education are not compulsory.

  3. 3.

    Recently, with the availability of data that contain more detailed information on individuals’ birth dates, researchers have utilized the age-eligibility cutoff for public policies in their RDDs. For example, Malamud and Pop-Eleches (2011) use the month and year of birth of individuals in Romania to study the country’s educational reform and its impact on the proportion of disadvantaged students graduating from college. Fitzpatrick (2010) used children’s exact birthdates to study the impact of universal pre-K availability in the US.

  4. 4.

    Please note that because Hong Kong’s legislation was effective for the whole region, for comparability, Table 7 does not include countries that implement compulsory education at the state or municipal level, such as the US, Canada, and Germany.

  5. 5.

    In many countries, having compulsory education does not mean that it is enforced or that the fees are reduced. This is especially true in the initial period of policy implementation due to the costs of monitoring (Murtin and Viarego 2011).

  6. 6.

    For males aged 10 to 14, the labor force participation rate decreases slightly, from 6.0 percent in 1961 to 4.7 percent in 1971.

  7. 7.

    Actually, there were more than enough free school spots for the children of school age by 1971; specifically, there were 862,000 spots, representing 136.6 percent of six- to eleven-year-olds.

  8. 8.

    In 1971, there were 759 government and aided schools, including five English language government schools and 691 private schools (Census and Statistics Department 1981).

  9. 9.

    Similar to the British education system, Form 1 in Hong Kong refers to the first year in secondary school. The period from Form 1 to Form 3 refers to Middle School or lower secondary school. Form 4 to Form 5 refers to High School or upper secondary school. Form 6 to Form 7 refers to matriculation (two years of preparation for college). In Hong Kong, the word grade is only used to refer to grades in primary school (Grades 1 through 6).

  10. 10.

    Therefore, by “left school by age 12” I am estimating whether a person leaves school without finishing primary school and by “left school by age 15” I am estimating whether a person leaves school without finishing middle school.

  11. 11.

    Ideally, we would have household information from 1971 to examine the effects of heterogeneity. However, the 1986 Census data are cross-sectional (longitudinal Census data are not available in Hong Kong).

  12. 12.

    The opportunity costs of sending a child to school would reduce household income, and if schooling is not enforced by law, some parents might opt out of schooling for their children because the savings in direct education costs are still less than the foregone earnings of child labor. Further, my results have shown statistically significant evidence that the 1971 policy reduces the early dropout propensity by age 15 (beyond 1971s compulsory-schooling age) and increases the probability of obtaining an education beyond middle school for children from disadvantaged backgrounds. These results could imply that the positive impact of reduced costs has offset the negative effect of increased opportunity costs of schooling.

  13. 13.

    Other districts include New Town District and rural areas.

  14. 14.

    Similarly, I also run equation (1) by replacing the outcome variables with the predetermined characteristics of individuals. None of the estimates are statistically significant. This ensures that the observed effects are not caused by individual characteristics rather than the treatment itself. Results are not shown but are available upon request.

  15. 15.

    The only reason that effects might differ in my global polynomial and local linear estimates is the mis-specification of the functional form of month-of-birth or noisier estimates in the smaller bandwidths. Lee and Lemieux (2010) emphasize the importance of graphing the data before running regressions and choosing the bandwidths: when the graph shows a more or less linear relationship, local linear estimations using different bandwidths will still yield similar results; however, when the graphs shows substantial curvature, larger bandwidths will not improve efficiency but generate bias. My graphs (Figs. 1, 2, 3) show that for the two dropout probability outcome variables (left school by age twelve and left school by age fifteen), the polynomial estimates and local linear estimates should be identical because Figs. 1 and 2 show a linear relationship of the outcome variables and month-of-birth. It is not surprising to see that for outcome variable education beyond middle school, the local linear estimates are closer to the actual discontinuities because the graphs exhibit more curvature. Therefore, local linear estimates are preferred and have been the focus of my discussion in the Results section (page 13).

  16. 16.

    Even though trying more flexible specification by adding polynomials in month-of-birth can assess the robustness of the RD estimates of the treatment effect, graphing the actual data will avoid mis-specification of the unknown function of month-of-birth (Lee and Lemieux 2010). As I later show in Fig. 3, the relationship between month-of-birth and the probability of obtaining education beyond middle school fits well in a quadratic function. This could explain why the estimates based on cubic form of month-of-birth are not statistically significant. Further, because the actual data shows substantial curvature, larger bandwidths will not improve efficiency but generate bias. So it is not surprising to see some differences among the local linear regression in different bandwidths.

  17. 17.

    Clark and Royer (forthcoming) do not directly report how the policy reduced the dropout probability by age fifteen (caused by the 1947 policy change) or the dropout probability by age sixteen (caused by the 1972 law). Instead, they report that “the 1947 change reduced the fraction of individuals completing nine or fewer years of education by around 0.5. The 1972 change decreased the fraction completing ten or fewer years of education by around 0.25” (Clark and Royer, forthcoming, 17).

  18. 18.

    The school day was split into morning and afternoon sessions, and class sizes were increased to accommodate the increasing number of students (July 1971, Education in Hong Kong, Hong Kong Government Information Service Publication).


  1. Angrist, J. D., & Krueger, A. B. (1991). Does compulsory school attendance affect schooling and earnings? Quarerly Journal of Economics no., 106(4), 979–1014.

    Article  Google Scholar 

  2. Board of Education. (1997). Report on review of 9-year compulsory education. Hong Kong: Government Printer.

    Google Scholar 

  3. Bray, M. (1992). Colonialism, scale, and politics: Development and convergence of educational development in Hong Kong and Macau. Comparative Education Review no., 36(3), 322–342.

    Article  Google Scholar 

  4. Cabus, S. J., & De Witte, K. (2011). Does school time matter? On the impact of compulsory age on school dropout. Economics of Education Review no., 30(6), 1384–1398.

    Article  Google Scholar 

  5. Census and Statistics Department. (1981). Hong Kong social and economic trends: 1970–1980. Hong Kong: Government Printer.

    Google Scholar 

  6. Choi, P.K. (1998). Wan wan liu dian ban: qishi niandai shang ye xiao de nvgong. (晚晚 6 點半:.七十年代上夜校的女工). Hong Kong: Step Forward Multi Media Company Limited.

  7. Clark, D., & Heather Royer. Forthcoming. “The effect of education on adult health and mortality: Evidence from Britain.” American Economic Review.

  8. Dammert, A. C. (2010). Siblings, child labor, and schooling in Nicaragua and Guatemala. Journal of Population Economics no., 23(1), 199–224.

    Article  Google Scholar 

  9. Del Bono, E., & Galindo-Rueda, F., (2007). “The long term impacts of compulsory schooling: evidence from a natural experiment in school leaving dates.” CEE Discussion Papers 0074. Centre for the economics of education, london school of economics.

  10. Devereux, P. J., & Hart, R. A. (2010). Forced to be rich? Returns to compulsory schooling in Britain. Economic Journal no., 120(549), 1345–1364.

    Article  Google Scholar 

  11. Fitzpatrick, M. D. (2010). Preschoolers enrolled and mothers at work? The effects of Universal Prekindergarten. Journal of Labor Economics no., 28(1), 51–85.

    Article  Google Scholar 

  12. Government Secretariat. (1981). Hong Kong education system. Hong Kong: Government Printer.

  13. Grenet, J. (2013). Is extending compulsory schooling alone enough to raise earnings? Evidence from French and British compulsory schooling laws. Scandinavian Journal of Economics no., 115(1), 176–210.

    Google Scholar 

  14. Hanushek, E.A., & Wößmann, L., (2010). “Education and economic growth.” In International Encyclopedia of Education (vol. 2), edited by Penelope Peterson, Eva Baker, and Barry McGaw, 245–52. Oxford: Elsevier.

  15. Harmon, C., & Walker, I. (1995). Estimates of the economic return to schooling for the UK. American Economic Review no., 85(5), 1278–1286.

    Google Scholar 

  16. Hong Kong Government. (1965). White paper: Education policy. Hong Kong: Government Printer.

    Google Scholar 

  17. Kaestner, R. (1997). Are brothers really better? Sibling sex composition and educational achivement revisited. Journal of Human Resources no., 32(2), 250–284.

    Article  Google Scholar 

  18. Kan, K., & Lee, M. J. (2011). The fertility effect of education: Regression discontinuity for counts and exponential models. Institute of Economics Academia Sinica: Working paper.

    Google Scholar 

  19. Lee, D., & Card, D. (2008). Regression discontinuity inference with specification error. Journal of Econometrics no., 142(2), 655–674.

    Article  Google Scholar 

  20. Lee, D. S., & Lemieux, T. (2010). Regression discontinuity designs in economics. Journal of Economic Literature no, 48(2), 281–355.

    Article  Google Scholar 

  21. Llera-Muney, A. (2005). The relationship between education and adult mortality in the United States. Review of Economic Studies no., 72(1), 189–221.

    Article  Google Scholar 

  22. Malamud, O., & Pop-Eleches, C. (2011). School tracking and access to higher education among disadvantaged groups. Journal of Public Economics no., 95(11–12), 1528–1549.

    Google Scholar 

  23. Meghir, C., Mårten P., & Simeonova E.,. (2012). “Education, health and mortality: Evidence from a social experiment.” NBER Working Paper 17932. National Bureau of Economic Research.

  24. Memorandum for Executive Council: Provision on compulsory powers to secure attendance of children at primary school. 20 July, 1971.(XCR(71)171) Hong Kong Government.

  25. Minutes of Education Department (GR) 4/278/68, 8 July 1971. Department of Education, Hong Kong.

  26. Morduch, J. (2000). Sibling rivalry in Africa. American Economic Review no., 90(2), 405–409.

    Article  Google Scholar 

  27. Murtin, F., & Viarengo, M. (2011). The expansion and convergence of compulsory schooling in Western Europe, 1950–2000. Economica no., 78(311), 501–522.

    Article  Google Scholar 

  28. Oreopoulos, P. (2007). Do dropouts drop out too soon? Wealth, health and happiness from compulsory schooling. Journal of Public Economics no., 91(11–12), 2213–2219.

    Article  Google Scholar 

  29. Parish, W. L., & Willis, R. J. (1993). Daughters, education, and family budgets: Taiwan experiences. Journal of Human Resources no., 28(4), 863–898.

    Article  Google Scholar 

  30. Pischke, J.-S., & von Wachter, T. (2008). Zero returns to compulsory schooling in Germany: Evidence and interpretation. The Review of Economics and Statistics no., 90(3), 592–598.

    Article  Google Scholar 

  31. Pong, S.-L., & Post, D. (1991). Trends in gender and family background effects on school attainment: The case of Hong Kong. The British Journal of Sociology no., 42(2), 249–271.

    Article  Google Scholar 

  32. Post, D. (1993). Educational attainment and the role of the state in Hong Kong. Comparative Education Review no., 37(3), 240–262.

    Article  Google Scholar 

  33. Post, D. (1994). Educational stratification, school expansion, and public policy in Hong Kong. Sociology of Education no., 67(2), 121–138.

    Article  Google Scholar 

  34. Powdthavee, N. (2010). Does education reduce the risk of hypertension? Estimating the biomarker effect of compulsory schooling in England. Journal of Human Capital no., 4(2), 173–202.

    Article  Google Scholar 

  35. Salaff, J. W. (1981). Working daughters of Hong Kong: Filial piety or power in the family?. Cambridge: Cambridge University Press.

    Google Scholar 

  36. Smith, J., (2009). “Can regression discontinuity help answer an age-old question in education? The effect of age on elementary and secondary school achievement.” B.E. Journal of Economic Analysis & Policy no. 9 (1): Article 48.

  37. Trochim, W. M. K. (1984). Research design for program evaluation: The regression discontinuity approach. Beverly Hills: Sage.

    Google Scholar 

  38. van der Klaauw, W. (2002). Estimating the effect of financial aid offers on college enrollment: A regression-discontinuity approach. International Economic Review no., 43(4), 1249–1287.

    Article  Google Scholar 

  39. White Paper: Education Policy. (1965). Hong Kong: Government Printer.

  40. Wong, N., (2004). “Educational reform in Japan and Hong Kong: A comparative study of curriculum decentralization.” Unpublished Manuscript, Hong Kong University.

  41. Xianggang nian bao (1972). Xianggang: Tian tian ri bao you xian gong si, 1971–1973 Xianggang nian jian. 1969. Xianggang: Hua qiao ri bao.

Download references

Author information



Corresponding author

Correspondence to Dongshu Ou.



See Tables 7 and 8.

Table 7 Regression discontinuity design studies on the impact of compulsory-schooling laws on educational attainment
Table 8 Panel A Children Aged 6 to 14 not at School in 1971, By Area. Panel B Children Age 10 to 16 not at School and Working in 1971

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Ou, D. Education for all: quasi-experimental estimates of the impacts of compulsory primary education in Hong Kong. Asia Pacific Educ. Rev. 14, 267–283 (2013).

Download citation


  • Compulsory-education policy
  • Dropouts
  • Regression discontinuity design
  • Hong Kong