OpenUCT :: Browsing by Author "Wittenberg, Martin"

Browsing by Author "Wittenberg, Martin"

Now showing 1 - 20 of 45

Open Access
Begging the question : permanent income and social mobility
(2007) Muller, Seán Mfundza; Wittenberg, Martin
Includes abstract. Includes bibliographical references (p. 35-37).
Open Access
By how much did the Malawi economic crisis affect the welfare of the people? : evidence from survey data
(2016) Mussa, Anwar; Wittenberg, Martin
The study uses data from the Malawi Integrated Household Panel Sur- vey (IHPS) to investigate the impact of the late 2011 through 2012 economic crisis in Malawi on the well-being of the people. Empirical results show that there was no statistically significant decline in the standard of living across Malawi between 2010 and 2013. The poverty incidence levels across areas in Malawi remained at their 2010 levels except for rural-south which experienced a significant decline in the number of poor people in 2013. Stochastic dominance test results show that irrespective of the choice of poverty measure and poverty line, there was no significant change in poverty between the two years save for rural areas in the northern and southern regions of Malawi. These areas actually experi- enced welfare gains, hence a decline in poverty. Sectoral decomposition results indicate that within area changes in poverty incidence, depth and severity contributed heavily to the overall change in poverty as opposed to population movements between rural and urban areas as well as among rural areas in the three regions of Malawi (north, centre and south).
Open Access
Children, investments in education and poverty in Malawi
(2009) Mussa, Richard; Wittenberg, Martin
This thesis presents an examination of three interrelated themes concerning household investments in the education of children, and how the number of children in a household affects a household's poverty situation in Malawi. These themes cover three chapters. Chapter 2, investigates two issues regarding household expenditure on primary educa- tion of own children using the Second Malawi Integrated Household Survey (IHS2) data. Firstly, we look at factors which influence a household's decision to spend or not (the par- ticipation decision), and by how much (the expenditure decision). This is done for urban and rural households. We found that there are differences in the factors which influence both decision levels for the two groups of households. Secondly, to get a deeper under- standing of these rural-urban spending differences, the study develops the Blinder-Oaxaca decomposition technique for the independent Double Hurdle model. The proposed decom- position is done at the aggregate and disaggregated levels. The aggregated decomposition allows us to isolate the expenditure differences into a part attributable to differences in characteristics and a part which is due to differences in coefficients. The detailed (dis-aggregated) decomposition enables us to pinpoint the major factors behind the spending gap. At the aggregate decomposition level, our results show that at least 66% of the expenditure differential is explained by differences in characteristics between rural and ur- ban households, implying that an equalization of household characteristics would lead to about 66% of the spending gap disappearing. At the disaggregated decomposition level, the rural-urban difference in household income is found to be the largest contributor to the spending gap, followed by quality of access of primary schools. Besides, rural-urban differences in mothers education and employment are found to contribute more to the spending differential relative to the same for fathers. Recognizing that in many African countries parents are not just responsible for the educa- tion of their own children, Chapter 3 examines the relationship between household income and schooling costs in the presence of intrahousehold schooling bias against non-biological children. To this end, we construct a two-period model of intrahousehold schooling bias. The model predicts that there is an asymmetry in the impact of changes in costs and income on schooling in the sense that the impact is larger for the non-biological child. It predicts that the asymmetry increases as the relationship distance between the non- v vi ABSTRACT biological child and the parents gets wider. It also shows that an increase in cost of schooling leads to a bigger reduction in schooling for poor households, and that the dif- ference in the impact of cost changes between the biological and the non-biological child declines as household income increases i.e. there is convergence. The convergence is faster the more distantly related to the parents the non-biological child is. An empirical investi- gation of these predictions using IHS2 data, shows that when current enrolment and grade attainment are used to measure schooling, the price and income elasticities of schooling are larger for non-biological children. The results also indicate that households in the low- est income quintile (the poorest) have the largest price elasticities, and households in the highest income quintile (the wealthiest) have the smallest price elasticities. We also found that the price elasticities for biological and non-biological children converge as we move from the lowest income quintile to the highest income quintile, and that the convergence is faster for non-biological children who are non-relatives. Having looked at among other things the role of household economic status on the edu- cation investments of children in Malawi in Chapters 2 and 3, in Chapter 4 we turn the question on its head and investigate the impact of fertility (number of children) on poverty in rural Malawi. We use two measures of poverty; the objective and the subjective. After accounting for endogeneity of fertility by using son preference as an instrumental variable, we found that fertility increases the probability of being objectively poor. This effect is robust for all poverty lines used. It is also robust to accounting for economies of scale and household composition as well as assuming that poverty is continuous. We also found that when fertility is treated as an exogenous variable its impact is underestimated. When poverty is measured subjectively, the results are opposite to those of objective poverty. We found that fertility lowers the likelihood of feeling poor, and that fertility is exogenous with respect to subjective poverty.
Metadata only
Determinants of black women's labour force participation in post-apartheid South Africa
(Journal of African Economies, 2015-05-28) Ntuli, Miracle; Wittenberg, Martin
Open Access
Distributional changes in the gender wage gap in the post-apartheid South African labour market
(2018) Mosomi, Jacqueline Nyamokami; Wittenberg, Martin
Gender inequality in the labour market is real and it matters. This thesis examines gender wage inequality in the post-apartheid South African labour market from 1993 to 2015 and finds that there is a substantial median wage gap of between 35% and 23% since the end of apartheid. This gap is unexplained by differences in human capital characteristics and it is not declining over time. There has however, been a substantial decline in the gender wage gap at the bottom of the wage distribution from 0.47 log points (about 60%) in 1997 to 0.18 log points in 2010 (about 19%). Similarly, the wage gap at the mean narrowed during the period studied, from 0.34 log points (about 40%) in 1993 to 0.15 log points (about 16%) in 2014. An examination of the gender wage gap is important because in the wake of a new democratic government in 1994 and the introduction of anti-discrimination and affirmative action policies, we expect that the gender wage gap should have narrowed over time. Interestingly, at the median where we would expect the legislation to have been binding, the gender wage gap has stagnated. We find however, that minimum wage legislation in low wage industries (agriculture and domestic work), has had some effect of narrowing the wage gap at the bottom of the distribution. This is especially interesting because this type of wage legislation was not specifically targeted at narrowing the gender wage gap. For this analysis we utilize the Post-Apartheid Labour Market Series (PALMS) dataset from 1993-2015. To do this, we need to address several data quality issues involving how household survey data has been constructed over time. One problem we need to address is that measurement has changed over time. In chapter two of this thesis we show how measurement issues affecting data on domestic work in 1994 and 1995 led to mixed results on the gender wage gap with some researchers finding a rise in the wage gap between 1995 and 2006 and others reporting a drop depending on the choice of the base period. This analysis goes beyond pointing out the classification issues in 1994 and 1995 and takes steps to update the data and address the “missing” raw gender wage gap in these two years. In chapter three, we analyse the trends in the gender wage gap in detail first starting with a mean decomposition using the Oaxaca decomposition method. We then extend the analysis to the entire wage distribution using the Dinardo Fortin and Lemieux (DFL) re-weighting method (DiNardo et al. 1996) for the aggregate decomposition and the Re-centred Influence Functions (RIF) method (Firpo et al. 2009) for the detailed decomposition. Results show that the changes in the gender wage gap are heterogeneous across the wage distribution. There has been a substantial narrowing of the gender wage gap at the bottom of the wage distribution which we attribute to improved female human capital characteristics and minimum wage legislation. On the contrary the median wage gap which is greater than the mean wage gap has been stagnant and displays very little movement in the period studied. There was some decline in the gap at the 90th percentile in the period between 1993 and 2005 but the gap seems to be expanding in recent years due to a continually expanding unexplained component of the gender wage gap. Given the snap shot nature of cross-sectional data, results from cross sectional analysis confound life cycle, generational and period effects on the gender wage gap. The fourth chapter of this thesis involves constructing synthetic cohort data from repeated cross-sections which we use to examine dynamic aspects of the gender wage gap. Life cycle trends show that younger cohorts of men and women have experienced a rise in wages over time which we attribute to improved human capital characteristics. Younger cohorts of women have on average more education than the cohorts before them and have more education than men from the same cohort. These cohorts of women are more likely to be in a skilled profession compared to women born 30 years before them and who joined the labour market during the apartheid era. Marriage rates and trade union rates have greatly declined for recent cohorts. These cohort changes mean that more recent cohorts of women have similar labour market or better labour market characteristics than men from the same cohort. This has led to the narrowing of the gender wage gap at the mean evident in the cross-sectional analysis. The general conclusion from this analysis is that the gender wage gap persists in the South African labour market and that the experience of women at the top end of the wage distribution is different from the experience of women at the bottom end or at the median. Therefore, policies towards narrowing the gender wage gap will need to be unique to challenges women face in different parts of the wage distribution. For example, while raising the minimum wage at the bottom of the wage distribution will narrow the gender wage gap in this part of the wage distribution, narrowing the gender wage gap at the top end of the wage distribution should focus on increasing the number of women in top paying occupations. This will require among other things alleviating the disproportionate burden of care work (for example availing child care and providing creche facilities) shouldered by women to enable them to commit more time to the labour market. That non-gender specific wage legislation worked to narrow the gender wage gap at the bottom of the wage distribution is an indication that addressing the gender wage gap may require more than labour market legislation aimed at reducing gender discrimination. It may require a broader look at society and social norms that may have or may not have shifted over time.
Open Access
Do Perceptions of Inequality Negatively Affect Interpersonal Trust?
(2022) Jacobs, Chandré; Wittenberg, Martin
Most studies that investigate the relationship between interpersonal trust and inequality make use of objective or societal measures such as the Gini coefficient, and the effect of individual assessments of inequality has thus far received much less academic attention. Using data from the Round 5 Afrobarometer survey, I estimate the effect of perceived inequality on interpersonal trust in South Africa. Though the Round 5 Afrobarometer survey does not include an income variable, I construct an asset index using Principal Component Analysis and find that individual perceptions of socio-economic rank deviate from this measure of objective wealth. In line with the general conclusions in the literature, I find that perceived inequality has a negative effect on interpersonal trust and that individuals who believe that their living conditions are either better or worse compared to those of other South Africans' are less likely to express trust in others. This effect is more prominent in rural areas, which suggests that individuals have biased reference groups influenced by their area of residence.
Metadata only
Does tenure insecurity explain the variations in land-related investment decisions in rural Ethiopia?
(2017-06-06) Shifa, Muna; Leibbrandt, Murray; Wittenberg, Martin
Open Access
An economic analysis of declining marriages in post-apartheid South Africa, 1995-2006
(2010) Kumchulesi, Grace; Wittenberg, Martin
Survey by survey comparisons of marriage rates in nationally representative cross-sectional datasets suggest that marriages are declining in South Africa. For African South African women of working age (between 15 and 59 years) marriage rates declined from 38.7 percent in 1995 to 31.4 percent in 2004. This change in marriage patterns motivated the current research and we asked whether the drop in marriage rates indicates a real generational shift in marital behaviour, or if this can be explained by changes in sampling frames in the independent surveys. The broad objective of the study is to investigate declining marriages in post-apartheid South Africa. The specific objectives are threefold. First, we construct a synthetic panel dataset from the 1995 to 1999 annual October Household Surveys and from the 2000 to 2006 September wave of the biannual Labour Force Surveys. Using the pseudo panel, we make use of the Age-Period-Cohort Model to disentangle marriage trends into age, period, and cohort effects in order to establish whether the change in marital patterns observed in post-apartheid South Africa reflects a real decline in marriages. Having established that the change in marriage rates indeed reflects a generational change in marital behaviour, the second objective focuses on the determinants of women's marriage decisions. To this end, we attempt to account for the interdependence between female labour force participation and marriage decisions by estimating simultaneous equation models for each cross-sectional year from 1995 to 2006. Availability of women's jobs in a District Council locality is used as an exogenous shock in the labour force participation equation to identify the marriage equation. The analysis finds that age, education, labour market status, availability of potential partners in the local marriage market, and location are all important factors in a woman's marriage decision. While the results show that labour force participation and a high level of education lower the probability of marriage, age and availability of potential partners are found to increase it. The third objective is an explanation of the trend towards fewer marriages by relating the changing effects of the variables to the marriage decline. Using the Blinder-Oaxaca decomposition technique, we establish that marriage decline is predominantly explained by a change in the marital behaviour of African South African women, rather than change in the distribution of the characteristics that determine marriage decisions. A detailed decomposition of the characteristic portion of the marriage decline analysis indicates that a rise in education levels and in labour force participation contributes to increasing the marriage decline. On the other hand, distribution in age of women, how they are geographically distributed, and the [unequal] distribution of men and women contribute to narrowing the marriage decline.
Open Access
Errors in the October household survey 1994 available from the South African Data Archive
(2006) Wittenberg, Martin
The on-line electronic documentation supplied with the 1994 October Household Survey by the South African Data Archive (SADA) appears to be incorrect. In particular, the electronic version of the questionnaire does not correspond to the hard copy in the possession of the author. The most serious error is that the race classification in the electronic copy is different from the classification on the hard copy. Researchers relying on the electronic copy will erroneously interchange the categories “Coloured”, “White” and “Black”. This could lead to seriously misleading analyses. The reason for this mistake can probably be attributed to a retyping of the questionnaire using the 1993 OHS as a template.
Open Access
Essays on the cause and consequences of market distortions
(2010) Pimhidzai, Obert; Wittenberg, Martin
In contrast to past literature on competitive bribery, the thesis demonstrates that competitive bribery for rents generated from market distortions is inefficient. However, self enriching policy makers have incentives for implementing these distortions in order to maximize the appropriation of rents. Their success in raising distortions is higher when the electoral process is easily manipulated. Thus distortions are higher and last longer in the absence of electoral accountability. Such distortions worsen the deprivation of human capabilities and functionings, as evidenced by worsening health outcomes among children in Zimbabwe after the policy changes since 2000. Market distortions are sub-optimal and attract bribery. Given arguments in the literature that competitive bribery is efficient, this thesis investigates whether this 'efficiency outcome' extends to bribery for rents generated from distortions. This analysis questions the generalization of the 'efficiency outcome' from models of competitive bribery. The thesis applies auction theory to demonstrate that a symmetric model of competitive bribery can produce an inefficient outcome. In particular, it shows that efficiency in competitive bribery models is sensitive to the source of rents. If the valuation of rents is misaligned to society's preferences, then competitive bribery produces an inefficient outcome. The generalization that competitive bribery is efficient is therefore misleading. This is illustrated using rents generated from market distortions hence the outcome confirms the inefficiency of bribery for rents generated from market distortions. Bribes for rents generated by market distortions accrue to office bearers. Therefore, self enriching policy makers find rent generating distortions attractive despite their negative impact on efficiency and social outcomes. These policy makers seek to extract maximum rents from distortions subject to constraints imposed by considerations for political survival. The political considerations are determined by the quality of electoral institutions. When electoral institutions are weak, incumbents can use electoral fraud to circumvent political pressure generated by the negative effects of high distortions. They maximize on financial rents by raising the level of distortions even when large welfare losses arise. Uni democratic regimes will therefore choose higher levels of distortions. They are less likely to adjust high distortions too. The thesis demonstrates this argument using a model of electoral accountability to compare distortions chosen by an incumbent with and without the possibility of electoral manipulation. The model shows that an incumbent chooses higher distortions when electoral manipulation is feasible. This theory is confirmed on exchange rate distortions in Sub-Saharan Africa. Dictatorships overvalue by between 22.4 and 26.6 percentage points more than complete democracies. These estimates are obtained from dynamic panel GMM estimation using data for non CFA Sub-Saharan countries. The quality of democracy is measured by the polity 2 score from the Polity IV data and exchange rate overvaluation by the black market to official exchange rate ratio. Further confirmation of the hypothesis comes from conditional logit estimation results. They show that undemocratic regimes are more likely to excessively overvalue the exchange rate than democratic regimes. This explains the persistence of excessive distortions in countries with weak institutions. Such distortions can not be wholly attributed to lobbying by special interest groups or to the political use of economic inefficiency as generally argued in the literature. Zimbabwe offers a practical example of distortions which are implemented or sustained for maximizing the appropriation of rents by the incumbent. The government stepped up its intervention in markets since 2000. The resulting distortions shifted command over resources and placed them in the hands of the elite while their intensity and persistence had devastating effects on the economy. Elections over this period have been accompanied by electoral intimidation and manipulation. Thus the incumbent government relied on electoral manipulation to continue the appropriation of rents from market distortions. Zimbabwe provides a good case study of the consequences of market distortions that are sustained to benefit the incumbent in the presence of weak democratic institutions. The thesis uses the 1999 and 2005/06 DHS data to make a comparative analysis of changes in social outcomes after the policy shift in Zimbabwe. This analysis sheds light on the social costs of market distortions that are sustained to benefit the incumbent. The average number of items consumed by under 5 year old children declined by 34%, mean height for age by 19% and mean weight for age by 16%. The biggest declines were on outcomes for children living in poor and middle class households. Children in richer households were the least affected. The significant negative coefficient of food consumption in the height and weight for age regressions show that food consumptions worsened child health outcomes. The decline in food consumption contributed to 37% and 57% decline in means of height and weight for age respectively. ii Wealth inequality in 2005/06 was 16% relatively higher than inequality in 1999. Wealth is measured by an asset index derived from principle component analysis and its inequality by the McKenzie 2005 index. Inequality in food consumption increased as evidenced by a change in the Kakwani concentration index for food consumption by 48% of the 1999 value. A decomposition of changes in health inequality shows that increased inequality in food consumption contributed to increased nutrition inequality by a magnitude of 11% and 6% of stunting and underweight concentration indices in 1999. These results indicate that market distortions worsened social outcomes in Zimbabwe. Yet the distortions persisted for almost a decade due to weak electoral accountability in that period. This shows that a weak intuitional environment which erodes electoral accountability promotes policies that are costly to society. Distortions that arise from such policies have an impact that extends beyond efficiency in allocation. They affect outcomes which are intrinsically important aspects of well-being. iii
Metadata only
Estimating expenditure impacts without expenditure data using asset proxies
(Southern Africa Labour and Development Research Unit, 2015-05-28) Wittenberg, Martin
Open Access
Evaluation of the Impact of the Provision of Triple Combination Antiretroviral Therapy to Employees Through the In-House Health Programme at a Large South African Mining Company
(2018) Brink, Jonathan Edward; Wittenberg, Martin
HIV/AIDS poses a unique challenge to businesses, particularly those operating in Southern Africa. The region is home to one third of the worldwide HIV positive population (as measured by those aged 15 to 49) yet as a whole contributes a diminutive proportion of the total worldwide population by the same measure (UNAIDS, 2016; World Bank, 2016). Relative to its size, the epidemic has introduced disparately large economic strain due to the fact that the highest HIV prevalence rates coincide with the most productive years of people’s lives, with prevalence rates peaking around the 30 to 34-year-old stratum (Shisana, et al., 2012). Loss of business productivity as a result of HIV related illness through a combination of absenteeism and so called ‘presenteeism’1 as well as death due to AIDS and the resultant increase in employee turnover, has motivated companies to implement workplace HIV education, prevention and treatment programmes over and above governments’ efforts to curb the effect of the disease on business operations and the associated economic costs (Granich et al., 2012; Meyer-Rath et al., 2012, 2015).
Open Access
Exploring the status of African Women in the South African Labour Market 1995-2004
(2009) Ntuli, Miracle; Wittenberg, Martin
For so long, black South African women have suffered from cultural and legalised discrimination. This gradually marginalised them from mainstream economic activities. Since the demise of Apartheid in 1994, the new government introduced corrective measures to improve the status of women in the labour market. For example, new legal provisions were enacted while international laws were also embraced. This demonstrated the government's commitment to achieve equality between men and women in the labour market and society at large. Given that more than ten years have passed since the inception of such enabling policies, it is reasonable to assume that remarkable strides were made in improving the status of African women in the job market. Therefore, this thesis aims to investigate whether the position of African women in the labour market has improved or not over the period 1995-2004. This audit is important for poverty eradication initiatives. Enhancing the status of women in the labour market is one vehicle through which poverty can be eradicated in the economy. The research focuses on three central areas. Firstly, the study explores the determinants of African women's labour force participation in 1995, 1999 and 2004. It uses logit models and the Even and Macpherson (1990 1993) decomposition. On the basis of this methodology, the study finds that for each of the three cross sections, education was the major correlate followed by non-labour income, marital status, geographical location and fertility. Furthermore, the increase in female labour force participation between 1995 and 2004 was mainly due to differences in coefficients/behavioural response than to a change in characteristics. The latter was especially due to behavioural response to education and age. However, these changes did not go far enough to make an improvement on the status of African women in the labour market. Specifically, the increase in female labour force participation was weighed down by some labour market inequalities associated with the trend like gender pay gaps. Secondly, gender wage differentials are scrutinized across the entire wage distribution in 1995, 1999 and 2004. The analysis utilises quantile regression estimation and counterfactual decomposition methods. This framework provides different estimates of the 'discrimination' coefficient across the wage distribution. In particular, it reveals that the gender gaps are wider at the bottom than at the top of the wage distributions. To add on, the research finds that the unexplained components of the gender pay gaps 'discrimination' did not substantially decline across the wage distributions between 1995 and 2004. Instead, the unexplained gaps slightly declined in the lower quantiles while increasing at the top end of the distributions. This probably indicates the persistence of substantial discrimination in the South African labour market, and that the incidence is more severe at the bottom than at the top of the wage distributions. However, rather than being a causal factor, this discrimination could be a symptom of the underlying problems. One of the possible causes is the low membership and hence representation of women in v decision making bodies such as trade unions. This invited a consideration of the gender differences in union membership. Finally, the research seeks to establish the nature and extent of the gender differences in trade union membership. It hypothesises that the gaps are either due to family loyalty, differences in union related characteristics or to discrimination. The analysis makes recourse to the Even and Macpherson (1990 1993) decomposition. The study finds that the gender gaps for 1995 and 2004 were mostly due to the unexplained components of the gender gaps/behavioural response, especially, differences in responses to family attachment related variables: marriage, occupations and industries. These outcomes sometimes show that most women spend most of their time carrying out domestic chores when compared to men. This suggests the persistence of patriarchal attitudes in society. Overall, our findings suggest that the changes in the status of African women in the post-Apartheid labour market were mainly due to responses to the constitution induced transformation rather than to a change in labour market characteristics. Nonetheless, this raises a question as to why, on the one hand, there were considerable shifts in labour force participation and on the other, there were negligible changes in unionism and pay gaps, yet all are explained by differences in coefficients. Therefore, we have suggested that the paradox resulted from massive changes in women's expectations about their involvement in paid work which are in concurrence with slowly changing social expectations about the role and place of women in the home and in the greater society. Clearly, women suffer from the work-family conflicts which compromise their advancement in the labour market. Also, it seems employers have not yet changed their discriminatory perceptions about women despite the presence of anti-discrimination legislation. The negative effect of this is to some extent an artefact of the persistence of patriarchal attitudes which continue to give women less voice in the labour market and in the society at large. Thus, we conclude that African women's de facto situation at the bottom of the hierarchy in the South African labour market is not mitigated by their de jure equality status. From these findings we speculate that the retention of patriarchy underlies the virtual restriction of an improvement in African women's labour market status. Therefore, we suggest that there is an urgent need for reforming gender roles at the societal level so that they exist on a more equal foundation and provide the basis for free and fair development of African women in the labour market.
Open Access
The external validity of treatment effects: an investigation of educational production
(2014) Muller, Sean Mfundza; Wittenberg, Martin
The thesis begins, in chapter 1, with an overview of recent debates concerning the merits of randomised programme evaluations and a detailed review of the literature on the extrapolation of treatment effects ('external validity'). Building on the insights of Cook and Campbell (1979) and a result by Hotz, Imbens, and Mortimer (2005), I then argue that the fundamental challenge to external validity may be interactive relationships between the treatment variable and other covariates. The empirical relevance of this claim is developed through two contributions to the economics of education literature, using data from the Tennessee class size experiment known as 'Project STAR'. Chapter 2 contributes to the literature on teacher quality, describing and implementing a novel method for constructing a value-added quality measure that uses a single cross-section of data in which students and teachers are randomly assigned to different-sized classes. The core insight is that constructing the value-added measure within treatment categories creates a plausible measure of quality that is simultaneously independent of treatment. The analysis of chapter 3 concerns the literature on class size effects. I argue that the effect of class size on educational achievement may be dependent on other class-level factors and that this should be considered when estimating educational production functions. Using the variable constructed in chapter 2, I estimate interaction effects between class size and teacher quality and find a number of statistically and economically significant effects. Specifically, higher quality teachers are associated with more beneficial effects of smaller classes. Those results suggest a possible unification of the class size and teacher quality literatures, with the policy problem being one of finding an optimal combination of these two factors. The broader contribution, further to the analysis of chapter 1, is to illustrate an obstacle to external validity: class size effects are unlikely to be the same across contexts where the teacher quality distribution differs. The experimental estimation of class size effects therefore serves as an empirical case study of the challenges to external validity that arise from interaction.
Open Access
Firm Size, Age and Growth in South Africa
(2018) Allen, Caitlin Shannon; Wittenberg, Martin; Kerr, Andrew
The relationship between a firm’s size, age and proportional growth rate is examined using multiple samples of South African firm-level data from the early to mid-2000s. The foundation of this study is Gibrat’s Law of Proportionate Effect (Gibrat, 1931), which states that a firm’s proportional growth rate is independent of its absolute size at the start of a given period. It is assumed that firm growth follows a random walk and, therefore, should not be affected by firm size. An implication of Gibrat’s Law of Proportionate Effect is that the firm size distribution is lognormal. However, based on both empirical and theoretical literature, this theory of firm growth has fallen out of favour and been replaced by the proposal that there is an inverse relationship between a firm’s proportional growth rate and both its size and age. Two questions are evaluated in this research using the samples of South African firms. The first is whether the firm size distribution is lognormal. If this is not the case then Gibrat’s Law of Proportionate Effect can be rejected. However, this approach cannot confirm that Gibrat’s theory is valid and will, therefore, be referred to in this paper as a partial test. It was shown that the log firm size distribution was not normal, but rather right-skewed with a Pareto distribution characterising the upper tail. Consequently, Gibrat’s Law of Proportionate Effect was rejected for the datasets of South African firms. This evidence is largely observational and does not explicitly assess the relationship between proportional growth rates and firm size. Therefore, the second question is whether Gibrat’s Law of Proportionate Effect holds. This was investigated by testing conditions derived from Gibrat’s Law of Proportionate Effect, the results of which can lead to either the rejection or acceptance of this proposition. This study extends Gibrat’s research in order to determine the relationship between firm age and proportional growth. Statistical methods, such as Ordinary Least Squares regressions, considering only firms that survived the period under consideration, were used. The results revealed that Gibrat’s Law of Proportionate Effect was invalid and there was a systematic tendency for the smaller, younger South African firms in the datasets to grow proportionally faster than the larger, older firms. This finding supports the view that firm growth is not entirely random.
Open Access
Household electricity access and households dynamics : insights into the links between electricity access and household dynamics in South Africa between 2008 and 2012
(2016) Harris, Thomas; Wittenberg, Martin
This paper investigates the details behind aggregate shifts in household electricity access in South Africa. More specifically, when viewed from a cross-sectional perspective, we note a significant (and surprising) decline in electricity access between 2008 and 2010, followed by a substantial improvement in access between 2010 and 2012. In order to further investigate these interesting dynamics and move beyond a limited cross-sectional analysis, we then set up the National Income Dynamics Study (NIDS) in a novel form that allows one to track household units in a longitudinal fashion. Using this data, we identify the initial drop in electricity access to have come as a result of a large number of household disconnections, as well as a significant degree of "misdirected" household formation (with people leaving household with access and setting up households in locations without access). We also identify the subsequent improvement in aggregate access to have come primarily as a result of a significant fall in the number of households that lose access over the period, an increase in the number of households that gain access, and favourable household formation processes (with people leaving households without access and moving into households with access). It is therefore vital that those involved in coordinating service delivery take into account that, if one's aim is to improve aggregate electricity access, preventing loss of access is just as important as expanding access. Policy makers should also take note of household formation and dissolution processes when considering service delivery expansion - to prevent government from needlessly chasing a moving target.
Open Access
Household formation in post-apartheid South Africa, 1995-2011: Measurement and Trends
(2021) Thornton, Amy Julia; Wittenberg, Martin
Who is in your household, what they have to share with you, and who does what, has profound implications for your welfare and well-being. Child welfare and progress through school; continued female disadvantage in the labour market and gender-based violence; livelihood strategies and physical health - these are just a few social outcomes that are fundamentally structured by the household. Change in most outcomes, therefore, cannot be abstracted from change in households. In South Africa, households have been changing in systematic ways. Between the first census of the post-apartheid period in 1996, and the latest in 2011, the number of households increased by 62% from 9 million to 14.5 million. By contrast, the population only grew by 28%, so that the average household shrank by almost a whole person. A notable aspect of this change is an extensive increase in the rate at which people live alone: the number of households that were single-person increased by 160% over the same time period. In other words, over time the South African population has been spread more thinly over fewer households, with direct implications for a sweeping set of social outcomes of importance to the post-apartheid project. A large literature on the topic of household composition exists for South Africa, and many researchers have noted the decline in average household size. Relatively fewer studies, however, aim to uncover the process by which this is happening, and those that do, mainly rely on highly localised data. Indeed, the question of why South Africans would form more and smaller households is particularly provoking given that economic circumstances have arguably been challenging in the post-apartheid period, characterised by high and persistent levels of unemployment and extreme wage inequality. As such, this thesis set out to investigate the process of household formation in the post-apartheid period, and how this aligned with known drivers of household formation in the economic literature, being aging, employment, and marriage. To do this, we construct a harmonised series of nationally-representative cross-sectional household surveys collected (almost) annually by Statistics South Africa, covering the period 1995-2011. Our first contribution is to make progress towards overcoming serious data quality issues in these surveys that undermine not only our study, but most studies seeking to relate people to households in data. We then turn towards analysing household formation, which raises the question of how to measure this process. We approach household formation from two angles: firstly, the aggregate household count by studying household heads, and secondly, the process of leaving the parental household by studying young adults. By tracing trends in household headship, we are able to describe which groups have become more or less likely to form households, and how this changed over the period between 1995 and 2011. Thanks to our attention to data quality, we are able to use the surveys to do the same for single-person households, a sub-group of special interest, in a way that is reliable for the first time. Our third contribution is to describe macro level trends in leaving the parental household, also for the first time for South Africa. Our key finding is that household formation in post-apartheid South Africa has been profoundly impacted by the steady decline of marriage over the same period. As such, household formation patterns are highly gendered, and modulated by men and women's differential access to labour market and grant income, over the life-cycle.
Open Access
The income distribution with multiple sources of survey error
(2013) Daniels, Reza Che; Leibbrandt, Murray; Wittenberg, Martin
Estimating parameters of the income distribution in public-use micro datasets is frequently complicated by multiple sources of survey error. This dissertation consists of three main chapters that, taken together, provide insight into several important econometric concerns that arise when analysing income from household surveys. The country of interest is South Africa, but despite this geographical specificity, the discussion in each chapter is generalisable to any household survey concerned with measuring any component of income.
Metadata only
Job creation and destruction in South Africa
(South African Journal of Economics, 2015-05-28) Kerr, Andrew; Wittenberg, Martin; Arrow, Jairo
Metadata only
Job creation and destruction in South Africa
(2015-05-28) Kerr, Andrew; Wittenberg, Martin; Arrow, Jairo

Browsing by Author "Wittenberg, Martin"

Results Per Page

Sort Options