Words: functional literacy, earnings determinants, Ghana, Sub-Sahara Africa Abstract This article analyzes the determinants of literacy and earnings in Ghana. It links literacy and earnings with a variety of factors, including age, gender, family educational background, distance to school, and income. Literacy and age are negatively correlated, suggesting that efforts at strengthening the supply and quality of basic education programs in recent years have been successful in raising literacy rates.
Females are less literate than males, controlling for other actors. Parents’ education is positively associated with literacy.
Distance to the nearest primary school, residency in a rural area, and poverty affect literacy negatively. Functional literacy appears to be a prerequisite for entering the labor market, Which may partly explain the lack Of return to education Other than middle school and technical/professional training. The policy implication of the study is that basic education and literacy programs should target females and poorer households, especially in rural areas.
We would like to thank Rosemary Bellmen’ and Helena Rib for invaluable support.
We are also indebted to Rocco Castro; Ronald Reneging; Nicolai Christensen; our discussant, Inhabitant Data Guppy; and the other participants at the conference on the Economics of Education and Human Capital, held by the Centre for Labor Market and Social Research and the Argus School Of Business, Denmark, in June, 1399, for helpful comments and suggestions. The views expressed here are those Of the authors and should not be attributed to the World Bank or any of its member countries.
Addresses: Department of Economics, The George Washington University, 2201 G street, Washington, DC 20052, USA and world Bank, 1818 H street, NW, Washington, DC 20433, USA E-mail: nblunch@worldbank.
Org, dverner@norldbank. Org. 1. INTRODUCTION Significant and rapid increases in earnings and education have taken place over the past hundred years in industrial economies. In developing economies the picture is different: High illiteracy rates and very low incomes, and thus widespread poverty, are realities for large parts of the world.
Literacy and income are closely linked. Establishing and assessing the nature of these links may help increase both literacy rates and earnings, thereby eradicating poverty. In this article, we analyze the determinants of literacy and earnings in Ghana based on two household surveys. Our results link literacy and earnings with a variety factors, including age, gender, family educational background, distance to school, and income. Literacy and age are found to be inversely correlated, implying that younger generations are more literate than older generations.
This relationship indicates that recent efforts to strengthen the supply and quality of basic education programs have been successful. Females are found to be less literate than males, controlling for Other factors. Parents’ education is positively correlated with their children’s literacy. Distance to the nearest primary school and residency in a rural area, are negatively correlated with literacy rates. Poverty and literacy are also negatively correlated. Our analysis Of the determinants Of earnings reveals no significant returns to education other than middle school and technical/professional training.
This result may indicate that the quality of education in Ghana generally is poor. Alternatively, it could suggest that education is not serving as a signaling device in Ghana. Functional literacy affects selection into the labor market. In evolving countries, jobs are rationed (that is, demand-side determined). We therefore interpret this result to indicate that functional literacy is a prerequisite for entering the labor market. This interpretation may partly explain the lack of returns to education.
The policy implications of these results are that greater efforts should be devoted to developing functional literacy skills and basic education. Policymakers should aim to increase both the supply and the quality of basic education and literacy programs. Basic education and literacy programs should target females and poorer households, especially in rural areas. The article is organized as follows. Section 2 describes changes in and determinants of literacy. Section 3 describes the Ghanaian economy. Section 4 presents the economic model and the econometric methodology underlying the analysis.
Section 5 presents the data and descriptive statistics, and section 6 presents the regression analysis. Section 7 summarizes the article’s findings and draws policy conclusions. 2. CHANGES IN AND DETERMINANTS OF LITERACY opinions differ widely over how best to define literacy. Unesco defines a functionally literate person as any person 15 or older who can “read and write a simple statement on his or her everyday life” (Unesco 1993, p. 24). The World Development Report (1997) also adopts this definition of functional literacy. Others propose a broader and more explicitly political definition.
Paulo Ferrier, the Brazilian educator, sees literacy as a process of “concentrations” that involves “reading the world” rather than merely “reading the word” (Ferrier and Macedon 1987). Widespread literacy is a twentieth century phenomenon. Before the nineteenth century, when public school systems were developed, education was reserved for the few. School systems developed in industrial countries largely in response to increased and more peccadillo industrialization, which in turn lead to increased economic growth and demand for an even more educated labor-force.
Over the past decade, education worldwide has exploded, as a result of the overstraining demand for still more specialized labor. Attitudes toward education have also shifted. Less than 50 years ago, education, especially higher education, was reserved largely for men. Today people in industrial countries believe the entire population has the right to education. 2. 1 Developments in Literacy in the United States Literacy progressed in stages in the united States. Initially, according to Has (1996), literacy spread because radical Protestants wanted to read the Bible.
Their need for literacy lead to agitation for general public education in the nineteenth century. Literacy rates later rose as a result of several societal changes, beginning with Reconstruction (Coy 1988). Immigration during the Industrial Revolution and the Progressive reform movement increased literacy later in the nineteenth and early twentieth centuries. World War and the Depression forced the federal government into a more active and direct role, and literacy rose even farther. The civil rights event of the 1 *ass shifted the focus to minority groups, broadening efforts to fight illiteracy.
Coy believes that the concept of functional literacy developed during this period and that the formalization of that concept helped increase the number of adult literacy programs. 2. 2 Developments in Literacy in Great Britain In Great Britain literacy also progressed in stages, Street (1995) identifies three distinct stages in the modern development of adult literacy programs. First, the recognition of adult illiteracy being a widespread phenomenon – in the sass – lead to increased focus on the issue.
Government grants were provided, a national “Right to Read Campaign” was launched, and local practice and experience Vass developers During the sass and early sass the government-funded agency Adult Literacy and Basic Skills Licit emerged. The unit provided materials and guidelines for good practice and funded small research projects. Since the late sass there has been a shift in policy and focus, aiming at adjusting education toward changing national and economic needs. 2. 3 Rates of Literacy in Developing Countries Substantial regional differences in illiteracy rates exist.
The rate of illiteracy is relatively low in Latin America and the Caribbean (13. 4) and very high in South Asia (50. 6 percent) (table 1). Ghana is in the middle of the spectrum, with an illiteracy rate of 35. 5 percent. Regions with high illiteracy rates also tend to have low per capita GNP and high philanderer ratio. Within Sub-Sahara Africa, substantial differences exist between Anglophone and Francophone countries (table 2). Illiteracy rates in Anglophone countries are 16 percentage points lower than in Francophone countries.
Average per capita GNP in Anglophone countries is more than twice as high as in Francophone countries, school enrollments are higher, and pupil teacher ratios are lower (37. 8 pupils per teacher in Anglophone countries versus 47_3 pupils per teacher in Francophone countries). Table 1. Selected Social and Macroeconomic Indicators, Ghana and Six Regional Groups, 1995 Middle East Latin Europe Cubans America and North South Sahara and the East Asia Central Africa Asia Caribbean Africa Asia Ghana and Pacific Indicator Adult illiteracy rate (percent) 35. 16. 9 -? 13. 4 38. 7 50. 6 44. 0 GNP per capita 350. 0 807. 8 -? 3,419. 8 -? 354. 1 485. 6 Gross primary enrollment ratio 76. 0 1154 99. 6 Ill . 5 96. 5 egg. C 74. 6 (percent) Ratio of primary school pupils to 27. 6 24. 2 20. 0 24. 5 27. 8 62. 7 40. 6 teachers Notes: -? Not available, gross primary enrollment ratio and ratio Of primary school pupils to teachers for Ghana are for 1993. Source: World Bank Detests database. Table 2.
Selected Social and Macroeconomic Indicators, Ghana and Anglophone and Francophone Countries in Substandard Africa, 1995 Indicator Ghana Anglophone countries Francophone countries in Sub-Sahara Africa in Sub- Sahara Africa Adult illiteracy rate (percent) 35. 5 36. 1 51. 9 GNP per capita 350. 0 675. 7 333. 4 Gross enrollment ratio (percent), primary 76. 88. 8 64. 4 Ratio of pupils to teachers, primary 27. 6 37. 8 47. 3 Notes: Gross primary enrollment ratio and ratio of primary school pupils to teachers for Ghana are for 1993, Source: World Bank Detests database. 2. Worldwide Increases in Literacy For the world as a whole, illiteracy rates have declined significantly, falling from almost 40 percent in 19/0 to just AS percent in 1990. Rates of illiteracy fell even more dramatically in developing countries, declining from SO percent in 1970 to AS percent in Gigs (Limb; 1996). Wide differences across gender, geographical region, and age exist within countries, however. Or the most part, males have higher literacy rates than females, urban areas have higher literacy rates than rural areas, and younger generations have higher literacy rates than older generations.
Limb (I egg) also draws attention to the fact that while total illiteracy rates have been falling, the proportion of women in the World’s total illiterate population has been rising. Three reasons for this tendency are suggested: the technologies of goods production, the nature Of human reproduction, and institutionalizing Of violence in the state. 2. 5 Literature on the Determinants of Literacy Little has been written on he determinants Of literacy. Law, Sprat, and Laborer (1995) analyze the determinants of literacy in Morocco.
They find that illiteracy is more widespread among females than among males, higher in rural areas than in urban areas, and inversely correlated with age. The negative relationship between age and literacy may reflect both deteriorating literacy skills over time and improvements in the quality of education. Cavy, Sprat, and Laborer also find that parents’ literacy and household expenditure level positively affect the level of children’s literacy, suggesting that poverty and family background are important determinants to literacy.
Verne (1999) analyzes the determinants of worldwide literacy rates by applying a human capital framework. She finds that enrollment rates, average years of schooling of adults, and life expectancy at birth are the main determinants of literacy, Income affects literacy in a nonlinear fashion, with a negative impact until a threshold of about $2,000 income per year per capita, after which the effect is positive, Institutional and regional variables are not very important in explaining literacy across countries.
Literacy rates differ widely across regions, a finding that can be explained by social and economic notations 3. AN OVERVIEW OF THE GHANAIAN ECONOMY Ghana is a low-income country, With per capita income Of 5406 in 1998 (World Bank 199%). It relies heavily on the agricultural sector, in particular cocoa, which accounts for almost half of GAP (World Bank Bibb). From the mid. CSS to the mid. sass, declining cocoa production and trade restrictions stalled economic growth in Ghana.
The return of more than a million Ghanaian from Nigeria in 1982-83 and a prolonged drought in 1982 caused growth rates to fall to laetrile low levels by about 1984 In conjunction with the MIFF and the World Bank, the Ghanaian government initiated he Economic Recovery Program (ERR) in 1983. The program implemented a number of policy reforms aimed at restoring macroeconomic stability, encouraging savings and investment, providing an enabling environment for the private sector, and improving public sector management, including prevarication of some of the many publicly owned enterprises, The ERR places significant emphasis on education.
The Education Sector Reform Program, established in 1987, improved the efficiency, quality, and relevance of education. The program also increased access to education and shortened the length of pre-university schooling from 17 to 12 years. As a result of the reform program, spending on education rose from 1. 4 percent of GAP in 1983 to 3. 8 percent of GAP in 1934. The government’s plans for additional reforms are outlined in its development starter, “Ghana-vision 2020” (Republic of Ghana 1935). A substantial part of the program’s social agenda is aimed at basic education.
Specific goals include achieving universal basic education and adult literacy, increasing access to secondary and tertiary education, and strengthening laborer skills by increasing technical and vocational training. To achieve these goals, the overspent, With the assistance Of the World Bank and Other donors, launched the Basic Education Sector Improvement Program in 1996. The program plans to increase investment in school facilities and teacher housing in rural areas and to strengthen science and math in the curriculum by raising education expenditures from 3. 8 percent of GAP in 1998 to 4. Percent in 2001. 4. THE ECONOMIC MODEL AND THE ECONOMETRIC FRAMEWORK The framework tort the analysis is standard human capital theory, in which individuals build up knowledge and skills through education and experience specific on-the-job experience as well as general experience (Becker 1975: Minced 1974). According to the theory, individuals who invest in human capital are subsequently rewarded with higher earnings. Formally, the economic model may be derived from the theory of either household or individual demand for schooling, both of which view education as an investment in human capital.
In industrial economies, in which subsidies for education are common, the investment decision may be viewed as an individual decision; in developing economies the relevant decision unit may be the household (Chandler, Lava, and Filmier 1994; Mason and Chandler 1997). Households will invest in education up to the point at which the marginal benefit from an additional year Of schooling equals the marginal cost of an additional year of schooling. In the traditional human capital literature, earnings are determined by education and Other individual, household, and, possibly, community characteristics. Earnings are observed, however. Only for individuals who have positive earnings (that is, who actually supply labor). To take this into account, we specify a labor supply function. Our model then becomes: (1) (2) Ii Is – E(lie, HI, Chi) S(lie, Hair Chi) This implicitly assumes that the household decommissioned possesses perfect information and that capital markets are perfect.
Both assumptions are very restrictive and appear unrealistic in developing economies. Where Ii (earnings of individual i) and Is (the labor supply of individual i) are the dependent variables; is a vector of individual characteristics, such as age and age squared (to capture possible nonlinearities), gender, the individual’s level of education, and the level of education of the individual’s parents; H is a vector of household characteristics, such as the wealth of the household: and C is a vector f community variables, such as urban versus rural location.
Literacy, L, is then determined by the following simple model: (3) Lie = L(lie, Hi, Chi) The explanatory variables are similar to those in the earnings equation, with some differences. In order to investigate the possible link between poverty and literacy, we include earnings and the poverty quintile of the household in H. We also include a measure Of the distance to the nearest primary school in C To analyze the determinants of earnings, we use a Hickman selection model (1976, 1979), which can be briefly described as follows.
Consider the earnings regression: (4) In Wi = Xi; * Ii here In Wi is log-earnings for individual i, Xi is a vector of explanatory variables for individual i, is a vector of parameters, and Ii is an error term capturing unobserved variables, The problem in estimating equation 4 is that we implicitly apply a sample selection rule because we observe only earnings of individuals who work; potential earnings of people who do not supply labor are not incorporated.
If the sample has characteristics that dieter from those of the underlying population in a nonrandom fashion, it will suffer from a selection bias, which, if not taken into account, will lead to biased parameter estimates. Hessian’s solution to this problem is to incorporate the labor supply choice in the earnings equation.
The earnings and labor supply choice equations thus become: In Wiz = Uzi where equation 5 is the earnings equation (equivalent to equation 4), in equation 6 is a latent variable that reflects the excess utility from participating in the labor market, and Uzi is a vector of variables explaining the labor supply decision of individual i. The latent variable l* corresponds to the indicator variable: lie- 1 if > 0, 0 otherwise The model is estimated by first estimating the inverse Mill’s ratio and then including it as an additional regresses in equation 5: (7) In Wi – Kip -t- Ii where ski is an estimate of the inverse Mill’s ratio for individual i. The Hickman model views labor supply as an individual choice. This view may be inappropriate in a development context, where the absence of (public) safety nets means that there is not likely to be much of a choice involved in the labor supply “decision. ” The labor actually supplied to the market is likely to be determined more from the demand side than from the supply side.
This contrasts with industrial economies, in which the labor supply decision is likely to be made in a different Skilled workers are more likely to supply their labor than unskilled workers since they forgo more income than do skilled workers by staying idle (given that there is a social safety net whose benefits are high enough not to “force” them into working). We applied the maximum likelihood version of the Hickman selection model -? rather than the Two-Step version in order to be able to weight the data.
We view the Hickman model as the general model, the validity of which must be tested against the reduced model, The reduced model here is the standard earnings equation, which is nested within the selection model. That is, the standard earnings equation is a special case of the selection model in which the selection correction terms, Ai, are statistically insignificantly different trot zero. S. THE DATA AND DESCRIPTIVE STATISTICS We test the model using data from two household surveys, the 1931192 Ghana Living Standards Survey (GILLS) and the 1997 Core Welfare Indicators Questionnaire (CSCW).
GILLS aims at obtaining measures of the living standard in Ghana on several dimensions, including health and education/literacy_ The survey is very extensive and includes 4,565 schooled. The CSCW aims mainly at providing data applicable for analyzing factors affecting poverty, education, and labor markets issues. It contains a much smaller number of questions (questions about earnings, for example, are not included) but a larger sample of households (14,514) and individuals (60,686). 5. Results on Literacy Investigating literacy and its covariates for the GELS data enables differentiating between several types Of literacy and reveals that being able to read and write in English is associated with higher earnings than is being able to read and write in one or more Ghanaian languages (table 3). The various measures of literacy are highly correlated, however. A problem that is likely to cause collinear in the regression analysis of the next section. To circumvent these problems, we combine the various literacy variables into a single composite measure of functional literacy.