Literacy and numeracy skills and education sector reform: evidence from Ghana

Abstract

Several African countries instituted education reforms in the 1980s and 1990s. Yet, there is only little evidence on the effectiveness of these programs. Additionally, most previous studies of the determinants of literacy and numeracy have considered the proficiency in only one language and, possibly, numeracy. This paper examines both of these issues for the case of Ghana, analyzing the trends in and determinants of four different literacy skills and numeracy. A comparison of outcomes before and after the 1987 Education Sector Reform indicates that literacy and numeracy skills levels increased following the Reform, and more so for English literacy skills than for Ghanaian literacy skills, thus confirming prior expectations. The results from linear probability models of literacy and numeracy outcomes indicate that, consistent with the objective of the Reform, the productivity of primary education in terms of literacy and numeracy increased, while the productivity of other levels of education remained constant or decreased. Additionally, the results highlight the importance of school quality and other community level factors in the creation of literacy and numeracy and therefore also the importance of incorporating these factors in empirical analyses. The results indicate that the focus of the 1987 Education Sector Reform might have been too narrow, so that future education policy in Ghana may want to focus on strengthening the quality of education above the primary level, also.

JEL classification:

• I210
• J240
• O150

Keywords:

• literacy and numeracy skills
• education sector reform
• human capital
• Ghana

Acknowledgements

I thank Bryan Boulier, Dean Jolliffe, Donald Parsons, David Ribar, participants at the Annual Meetings of the Southern Economic Association, the European Society for Population Economics, the Population Association of America, the IZA/World Bank Conference on Employment and Development, and seminar participants at The George Washington University for helpful comments and suggestions on earlier versions of this paper. The comments and suggestions from two referees helped greatly improve this paper. Remaining errors and omissions are my own. The data were kindly provided by the Ghana Statistical Service. The findings and interpretations, however, are those of the author and should not be attributed to the Ghana Statistical Service.

Notes

1. Ghana has more than 63 different indigenous languages (Herbert and Robinson 2001).

2. Again, in Ghana by the 1980s, the decline in the education system had reached a point where primary school completers frequently were not literate, especially in remote areas where school quality was particularly low (World Bank 1989). There is therefore a case to be made that even basic literacy and numeracy skills production may well continue well into the secondary school system.

3. While this is relevant for Ghana, also (Blunch 2006; Blunch and Pörtner 2011) the timing issues related to adult literacy course participation when specifically examining the possible impact of the 1987 Education Sector Reform on skills outcomes implies that the importance of adult literacy course programs in this context cannot be assessed. To be sure, while adult literacy course participation might be relevant for all adults with low levels of education, the available data only contain information on whether an individual has participated but not when. It is therefore not possible to determine whether an individual participated in an adult literacy program before or after the 1987 Education Sector Reform. As a result of this, the analysis therefore focuses on primary, middle/junior secondary and secondary and above – where also the strongest predictions regarding the possible impact of the Reform can be made.

4. One implication of this for the empirical analysis is that parental education should be included as explanatory variables.

5. A given individual can therefore be proficient in none or one or more (even all) of the five literacy and numeracy skills.

6. Note that while from Table 1 the incidence of English reading and writing skills and of numeracy skills are almost identically 100% – consistent with all secondary and above completers being literate in English and numerate, as one would probably also expect, ‘even in a poor education system’ (as noted by a referee) – the main issue why this is not the case for the Ghanaian literacy skills incidence in Table 1 is mainly due to the increased importance of English as the language of instruction in the Ghanaian education system.

7. Eight individuals report having completed ‘other education.’ These are dropped since it is not clear what ‘other education’ is.

8. One might suggest also including variables for rural–urban location and region of residence. However, these would be ‘washed out’ by the cluster fixed-effects due to perfect multicollinearity. Hence, including cluster fixed effects effectively is the most effective way of controlling for supply and demand factors related to geographical location, including school quality, by controlling for these factors at lowest possible level of (spatial) aggregation.

9. Sensitivity analyses examine the consequences of restricting the age range of the pre-reform cohort in Section 6 to take into account possible human capital depreciation for this older group.

10. As is well known, there may be some concern about using 2SLS/OLS, or, in effect, the linear probability model (LPM), when the dependent variable is binary. For example, predicted probabilities may fall outside the (0,1) range and heteroskedasticity also is present by default. However, it can be argued that the LPM approximates the response probability well, especially if (1) the main purpose is to estimate the partial effect of a given regressor on the response probability, averaged across the distribution of the other regressors, (2) most of the regressors are discrete and take on only a few values and/or (3) heteroskedasticity-robust standard errors are used in place of regular standard errors (Wooldridge 2002). All three factors seem to work in favor of the LPM for the purposes of the application here.

11. Namely, either using as IVs (1) various combinations of time of year, YOB and/or geographical area of birth dummies to capture variation in institutional factors relevant for human capital accumulations such as compulsory schooling laws or expansion of educational programs (Angrist and Krueger 1991; Duflo 2001) or (2) variables for proximity or exposure to educational institutions in the local area (Card 1995, 2001).

12. These results are not shown here but are available upon request.

13. The IV/2SLS results are given in Table A2 in the appendix. They are roughly similar for the first IV strategy but not for the two others, where the results are not statistically significant.

14. There are of course other education outcomes than literacy and numeracy, which are also part and parcel of people's educational decisions. For studies examining the same data set used here in the context of wages and child health, see Blunch (2006).