A. Introduction

1. Since independence, the Namibian economy has failed to generate significant job growth. Estimates from the 2002 Labor Force Survey suggest that 20 percent of the economically active population is unemployed. Broader measures of unemployment yield estimates as high as 54 percent. Estimates from the International Labor Organization (ILO) put total unemployment at 34 percent in 2000.^{2 3} These relatively high unemployment levels are not driven by short-run business cycle fluctuations, but appear highly persistent. Although there are some differences in sampling methodology, unemployment among the economically active was around 19 percent in both the 1993/1994 and 1997 Labor Force Surveys—the ILO estimate for 1997 was 35 percent. By a substantial margin, unemployment is also widely perceived to be the most important problem facing the country (Table I.1).

View Full Size

• Download Figure
• Download figure as PowerPoint slide

2. Using the Afrobarometer Project survey data, this chapter develops some stylized facts about the Namibian labor market, focusing on the link between education, earnings and unemployment. There is strong evidence that, unlike more developed economies where unemployment also affects educated workers, unemployment in Namibia is primarily an unskilled phenomenon—unemployment among those with post secondary education is almost non existent, and decreases sharply with education attainment.

3. The chapter finds that unemployment probabilities depend on the level of education. The completion of secondary schooling is an important threshold in the Namibian labor market, and those who have not completed secondary schooling account for a disproportionately large share of the unemployed. In addition, there is evidence that women face a higher risk of unemployment than men with the same age and education level.

4. Consistent with the differences in unemployment probabilities, there is also evidence that returns to education are large, and become larger as the level of education increases. Specifically, compared to a primary school graduate, a high school graduate has a 38 percent higher chance of being a high-income earner. And a college graduate has a 56 percent higher chance of becoming a high earner compared to a high school graduate.

5. Although the methodology and available data cannot predict how large-scale changes in the education distribution will affect unemployment and the skill premium—this awaits a general equilibrium analysis—it does suggest some tentative policy conclusions. Since unemployment is significantly more likely among unskilled workers, public policy measures that reduce the number of unskilled, such as education reforms that improve access and completion rates for secondary, tertiary and vocational education, can help lower unemployment in the long run, possibly raising average incomes.

6. At the same time, steps that increase the demand for unskilled labor may also have a large impact on unemployment. Public initiatives such as works programs, which could possibly be funded through donor support, could help. But given the magnitude of unemployment and the already large role of the public sector in providing employment,⁴ success in tackling the problem mainly depends on the private sector. More efficient approaches might include modifying the labor legislation to reduce the cost of employing unskilled labor, especially for smaller businesses. In addition, given the possible complementarities between unskilled and skilled labor [Ramcharan (2004)], relaxing visa restrictions for the entry of more educated workers could potentially increase the demand for local unskilled labor.

B. Empirical Framework and Main Results

7. The empirical framework is based on Afrobarometer Survey data, which closely mirror the results of the Namibian labor market surveys (Box 1). Taking the standard definition of unemployment as those not employed but currently searching for a full time job, Table I.2 reports that nearly 28 percent of the respondents are unemployed. Close to 10 percent are employed part time, but still searching for a job. That is, unemployment, including underemployment is around 38 percent—similar in magnitude to the labor market surveys. In fact, only about 33 percent of the survey respondents are employed full time, suggesting that the dependency ratio—the employed to total population—closely mirrors that found in the labor market surveys, and is consistent with the widespread perception of unemployment as the most serious problem facing the country.

View Full Size

• Download Figure
• Download figure as PowerPoint slide

Box I.1.

The Afrobarometer Survey

The Afrobarometer Survey is designed in conjunction with Michigan State University, the Institute of Democracy in South Africa, and the Center for Democratic Development in Ghana. Funding in part is provided by the National Science Foundation (U.S), the World Bank, USAID, the African Development Bank, as well as the UK, Swedish, and Dutch Governments. In addition to academic research, the surveys are also widely used in the local policy-making process. For example, the government of Lesotho adopted the report of the 2000 Afrobarometer survey as an official working document, while the South African Human Rights Commission adopted the survey data for its ongoing monitoring activities.

The data are collected during face-to-face interviews. Participants are chosen by national probability samples that represent an accurate cross section of the voting age population. In the case of Namibia, the survey consists of 1200 individuals. Random selection is used at every stage of sampling and the sample is stratified to ensure that all major demographic segments of the population are covered.

The table below provides some information about the survey participants. The survey is nearly equally split between the sexes, as females account for 49 percent of the respondents. The survey is also focused on the voting-age population, and the average age of those surveyed is about 34—the median age is 30. Survey participants are also linguistically representative of the Namibian population. About 45 percent of the respondents speak Ovambo languages, and the percent of Afrikaans, Herero and Nama/Damara speakers are each around 12 percent.

Respondents’ Demographic Characteristics

Respondents’ Demographic Characteristics

Total Number of Respondents	Gender				Age
	Male		Female		Average Age		Median Age
1200	51.08%		48.92%		34.13		30
Language
English 1%	Portuguese 0.67%	Afrikaans 12.92%	Ovambo Languages 45.17%	Kavango Languages 8.33%	Herero Languages 12.58%	Caprivi Languages 5.25%	Nama/Damara 12.67%

View Table

Respondents’ Demographic Characteristics

Total Number of Respondents	Gender				Age
	Male		Female		Average Age		Median Age
1200	51.08%		48.92%		34.13		30
Language
English 1%	Portuguese 0.67%	Afrikaans 12.92%	Ovambo Languages 45.17%	Kavango Languages 8.33%	Herero Languages 12.58%	Caprivi Languages 5.25%	Nama/Damara 12.67%

View Full Size

• Download Figure
• Download figure as PowerPoint slide

8. Survey data suggest a close relationship between education and unemployment. Table I.3 provides a comparison of the breakdown of the education level between the unemployed and the general population. The evidence suggests that the completion of secondary schooling is an important threshold in the Namibian labor market. While some 71 percent of the unemployed have some secondary education or less, this group only accounts for about 62 percent of the overall sample. That is, those who have not completed secondary schooling account for a disproportionately large share of the unemployed. Conversely, individuals with post-secondary qualifications and beyond make up 15½ percent of the overall sample, but just 2½ percent of the unemployed. Put differently, the odds of unemployment are 5.66 times higher for those without a high school degree, compared to someone with post-secondary qualifications.

View Full Size

• Download Figure
• Download figure as PowerPoint slide

9. A simple logit regression framework supports the notion from tabular evidence that unemployment is mainly concentrated among the less educated. The logit regression framework estimates the link between an individual’s education level and the probability that the individual describes himself as unemployed. The coefficient in Table I.4 suggests a large negative and robust relationship between education and unemployment. Specifically, moving from no formal schooling to informal schooling reduces the probability of unemployment by 9 percent. Likewise, compared to some secondary schooling, completing high school reduces the probability of unemployment by 12 percent. But the steepest reductions are reserved for those that acquire tertiary education. In this case, some university or college education reduces the probability of unemployment by 14 percent compared to those that report only post secondary education, but not university.

View Full Size

• Download Figure
• Download figure as PowerPoint slide

10. Adding demographic factors does not alter these conclusions. While education attainment might be correlated with other demographic factors, such as age, gender, and the region of residence, which might also affect employment outcomes, column 3 of Table I.4 shows that such impact is small, and age itself does not appear to significantly change the odds of unemployment. However, the odds of a woman reporting unemployment is about 1.4 times greater than that of man.⁵ That is, a woman having the same age and education level as a man, and from the same region is about 40 percent more likely to be unemployed.

11. In addition to being less likely to become unemployed, the returns to education increase dramatically for those who have at least completed high school. This is suggested by Table I.5. Unskilled workers such as farm laborers, maids and vendors have the lowest monthly household income. But household income increases steeply with the level of education, although the rate of increase slows after secondary completion. That is, teachers and professional workers generally have post-secondary education and report the highest household income. They earn about 1.67 times as much as clerical workers—who usually have completed high school. But professional workers and teachers earn nearly 20 times as much as farm workers and others who typically have completed only primary schooling. These simple statistics suggest that while the return to a university degree is large, the returns to completing secondary education are huge.

View Full Size

• Download Figure
• Download figure as PowerPoint slide

12. An ordered logit framework is applied to better understand the relationship between education and household income. Ordered logit estimates reveal how changes in education levels affect the probability that an individual reports a particular household income category. Specifically, let X_i denote individual i’s education level from among the 10 categories listed in Table I.5, then the probability that individual i reports a household income j from among the 10 possible income categories—again listed in Table I.5 is:

Pr(Household Income=j)=Pr(k_j-1<BX_i+u_i<k_j)

The parameter B indicates how changes in education, X_i, affect the likelihood that the individual reports household income j. The variable u_i are unobserved features of individual i and are assumed to be logistically distributed. The k_s are the cutoff points that determine the probability that an individual with education X_i reports a household income category j. Table I.6 presents the ordered logit estimates. The parameter B is positive and highly significant (p-value=0.00), suggesting that higher levels of education greatly increase the probability the individuals report higher monthly household income.

View Full Size

• Download Figure
• Download figure as PowerPoint slide

13. Using these estimates, for each of the income categories Figures I.l to 9 depict the relationship between an individual’s level of education attainment and the probability that the individual reports a particular income category. In Figure I.1 for example, individuals with no schooling, informal schooling or some primary schooling have the highest probability of reporting monthly income between N$0-N$500—the median and mode value in the sample. There is also virtually little change in this probability when moving from no schooling to informal schooling or even some primary schooling. However, the probability of household income being in this category falls steeply for those who graduated from high school, and is at its lowest among those with some form of tertiary education.

View Full Size

The Probability of Reporting Household Income Less than N$501

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$501 and N$1,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$1,0001 and N$2,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$2,001 and N$4,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$4,001 and N$6,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$6,001 and N$8,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$8,001 and N$10,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$10,001 and N$20,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Probability of Reporting Household Income between N$20,001 and N$50,000

Citation: IMF Staff Country Reports 2006, 153; 10.5089/9781451828429.002.A001

• Download Figure
• Download figure as PowerPoint slide

14. The average respondent in the sample claimed household income in the range of N$501-N$l,000. Figure 1.2 shows that relationship between education and the probability of reporting income in this category is an inverted u-shaped or non monotonic. Individuals with no formal schooling are the least likely to report income in this category. The N$501-N$1,000 is most commonly reported among those with a high school degree or some post-secondary education. Beyond this education threshold, the likelihood drops, as individuals with higher education levels are likely to report higher incomes.

15. Indeed, beyond the N$501-N$1,000 category, the relationship between education and income is strictly increasing, and becomes convex at higher income categories. Specifically, higher educated individuals have a higher probability of reporting higher incomes. But the rate at which education increases the probability of reporting higher incomes is also increasing in the level of education. For instance, moving from completing primary schooling to completing secondary schooling increases the probability of reporting monthly household income in the range of N$6,000-N$8,000 by 38 percent. However, a change in education status from high school to university graduate increases this probability by 56 percent. And as column 3 of Table I.6 indicates, these results are robust to the inclusion of the standard demographic controls such as gender, age and region.

C. Conclusion

16. These results are consistent with the well-known advantages of education. As noted in Mincer (1993), relative to less educated workers, educated workers have higher wages, greater employment stability and greater upward mobility of income. Likewise, the evidence strongly suggests that unemployment in Namibia is primarily concentrated among those without a high school education. And while our estimates are unable to provide net returns, since we lack data on schooling costs and forgone wages, these results also suggest that the gross returns to education in Namibia are large, and increase dramatically up the schooling ladder.

17. That said, the relative differences between educated and less educated workers in Namibia are much larger than typically observed elsewhere. In the US for example, those with a bachelor’s degree earn about 2.88 times as someone without a college degree. The Namibian data however suggest that the relative difference between the two groups is about 28 times. In addition, US evidence implies that the possession of a college degree increases the probability of being in the labor force by nearly 23 percent over a high school graduate; for the OCED as a whole the increase is around 17 percent. In Namibia, the difference in probabilities between the two groups is around 35 percent. Therefore, unlike OECD countries [Levine (1998)], where unemployment also impacts white collar workers, unemployment in Namibia is mainly concentrated among less educated, becoming increasingly rare as education levels rise. As such, unemployment in Namibia, as in South Africa [Kingdon and Knight (2004)], is primarily an unskilled phenomenon.

18. This suggests that effective public policy measures to reduce unemployment should tackle both the demand and supply sides of the problem among the unskilled. Specifically, the very high returns to secondary and tertiary education suggest that improving education access and completion rates at these levels will reduce the supply of unskilled labor, lowering unemployment and possibly raising average incomes⁷. But in the short term, increasing the demand for unskilled labor in both the public and private sector could also reduce unemployment among the unskilled, although the impact on average incomes might be less. In addition, public initiatives such as works programs, which could possibly be funded through donor support, can help. But given the magnitude of unemployment and the already large role of the public sector in providing employment, success in tackling the current high level of unemployment mainly depends on the private sector.

19. Moreover, exempting small employers from some of the annual leave and other costly provisions in the new Labor Act may increase unskilled employment. Small firms and businesses—auto garages, landscapers, restaurants etc—are more likely to employ unskilled labor. These firms would also be the least able to absorb the increased costs of employment associated with the provisions in the new law. Exemptions from some of the costlier labor provisions may lead to an increase in unskilled employment, or at the very least, avoid decline a in the demand for unskilled labor among the smaller firms.

20. Likewise, given the possible complementarities between unskilled and skilled labor [Ramcharan (2004)], relaxing visa restrictions for the importation of more educated labor can potentially increase the demand for local unskilled labor. For instance, allowing a highly trained foreign mechanic to enter the work force may allow local firms to conduct repairs that they otherwise could not, prompting the hiring of local unskilled labor to facilitate these repairs. That said, this analysis is partial equilibrium in nature and does not provide information about how large scale changes in the education distribution will affect wages and employment.