Poverty, Inequality and employment in Uganda

Transcription

1 Poverty, Inequality and employment in Uganda Abstract We analyzed the relationship between poverty, employment, and inequality in Uganda. We show that at a macro level, Uganda has managed to maintain positive growth in national output but with stagnant and in some instances declining growth in agriculture. Our results also show that real earnings of households in Uganda have fallen with the recent increase in food and commodity prices. In addition, private investments in agriculture relative to other sectors have declined over the past 15 years. Using the most recent national household survey, for rural areas, we find that poverty is a key constraint to non-farm employment especially non-farm causal wage employment and self employment. However, most recent public initiatives focus more on the employment challenges of urban areas to the detriment of rural areas. Nonetheless, the initiative of cluster development as proposed in the National Development Plan has the potential to create new employment opportunities in the rural areas.

2 1.0 Introduction Due to inequality of opportunity, there is an increasing focus on poverty and employment issues in developing countries (World Bank, 2006). The renewed focus is partly driven by the desire to reduce poverty as part of the millennium development goals (MDGs) and also by the unprecedented reduction in poverty in China over a period of 20 years however with rising income inequality. A number of recent studies show that China was able to reduce the population of people living in poverty by 500 million between 1981 and 2004 by expanding employment in both rural areas and major cities (Ravallion, 2009; Ravallion & Chen, 2007). At the same time, income inequality increased in China as the country expanded employment and reduced poverty. On the other hand, most countries in sub-saharan Africa (SSA) have not achieved similarly large reductions in poverty and in some cases, the welfare status of Africans has worsened in the recent past. For example, Fox & Sekkel (2008) show that the headcount poverty index for SSA increased from 45 percent in 1990 to 47 percent by This is partly attributed to failure to expand employment especially rural non-farm employment. Consequently, questions remains on what factors can influence non-farm employment especially in rural areas and how does the expansion of employment affect poverty and the distribution of incomes. In Uganda, the rampant unemployment especially of the youth and its associated social consequences is continuing to receive more attention from the policymakers (see for example, MoFPED, 2009, Sender, 2009). Indeed, the recent riots in Central Uganda during September 2009 are attributed partly to the high level of joblessness especially by the youth in urban areas. In the recent past, the Government of Uganda (GoU) has intensified efforts to boost employment through promoting industrial 1

3 expansion and skills employment for the youth. For example, the 2009/10 national budget allocated an additional Ushs 3 billion to the directorate of industrial training to offer vocational skills to the youth. However, there are mixed views on this approach. Previous studies in East Africa show that vocational training is used more as means of acquiring an additional certificate rather than acquiring skills (Hann, 2002). Another example targeting expansion of skilled employment in Uganda includes the promotion of industrial jobs through the Kampala Industrial and Business Park (KIBP) project by the Uganda Investment Authority (UIA). But even with the above efforts, non-farm employment opportunities remain relatively few in comparison to either the population of school leavers or the overall national population growth rate. Nonetheless, employment creation remains centre to Uganda s development endeavours. The recently launched National Development Plan ( ) proposes a number of strategies to increase employment in Uganda. In particular, over the next five years, the government will seek to promote high quality employment through youth employment and the setting up of a labour market information system that links employers and job seekers. In addition, the NDP proposes to develop a pool of non formal employable skills through institutional support for entrepreneurship; promotion of enterprise start ups; and the value chains through cluster development. It is hoped that the above strategies will boost both formal and informal employment especially in the rural areas. Nonetheless, questions remain as to why rural households have failed to register significant growth in non-farm employment despite significant changes to the structure of the Ugandan economy. This paper is an attempt to provide some answers as to whether poverty is a constraint to achieving non-farm employment. 2

4 There have been previous attempts to examine the nature, extent, and the key drivers of Uganda s labour market. Examples of empirical studies in the recent past include: Fox & Sekkel (2008), Okurut & Ssewanyana (2007), Sebbagala (2007), Kigima et al. (2006), and Canagarajah et al. (2001). Using the integrated household survey of 1992/93 and 1996 monitoring survey, Canagarajah et al. (2001) show that non-farm employment contributes to widening income inequality in Uganda. Ssebagala (2007) examines male-female wage gaps and find that discrimination is the main explanation for the gender wage gaps. One of the most comprehensive empirical work examining household labour response to shocks in Uganda by Kijima et al. (2006) focused on whether non-farm employment could help mitigate agricultural shocks in rural Uganda. Using panel data of 890 households collected in 2003 and 2005 by the Research on Poverty, Environment and Agricultural Technologies (REPEAT) project in Uganda, the authors examine how agricultural households cope with shocks emanating from rainfall failure and crop diseases. They find that households increase their supply of non-farm labour once faced by shocks but the income from such new employment does not compensate the income loses due to the above agricultural shocks. According to the authors, although rural non-farm employment is good for mitigating the impacts of shocks, it is nonetheless not a substitute for regular rural incomes from agriculture. This paper extends recent works on employment in the following ways: We focus on employment of both individuals and household heads in rural Uganda. In particular, we are interested in the employment choices of rural individuals as these show the largest levels of deprivation. Fox and Sekkel (2008) is the works most closely related to this study, but we do undertake a number of improvements. First, unlike the above study, we include the most recent household survey the 2005/06 Uganda National 3

5 Household Survey. The study by Fox and Sekkel (2008) only considered the period 1992/93 to 2002/03. Second, we examine the correlates of non-farm employment unlike the study by Fox & Sekkel (2008) which focussed on the description of the various employment states in Uganda. In addition, we examine constraints to various employment choices especially non-farm employment. Finally, we examine the policy implications of the changing macroeconomic environment on the employment and earning prospects of rural households. The broad objective of this study is to examine the relationship between poverty, employment, and income inequality in Uganda. First, we examine how the recent macroeconomic performance has impacted on poverty and employment. Second, we examine how changes in the macroeconomic fundamentals especially inflation, have impacted on poverty and inequality among households in Uganda. Third, for the econometric analysis, we use the 2005/06 UNHS to examine the constraints to rural non-farm employment. The rest of the paper is organised as follows. In the next section, we provide an analysis of the impacts of macroeconomic performance on poverty and inequality. The methods, data, and the variables used in the econometric analysis are presented in section 3. The discussion of the results is the subject of section 4. Section 5 concludes with implications for policy. 2. Macroeconomic performance and impacts on poverty and employment Uganda has made significant progress in reducing income poverty, especially during the implementation of economic reforms. The national headcount poverty index reduced from 56 percent in 1992/93 to 31 percent by 2005/06; it is only during the period 1999/00 to 2002/03 that the country witnessed a reversal in poverty reduction 4

6 (Table 1). Despite the reduction in income poverty, growth in non-agricultural employment in Uganda has been limited. According to an earlier study by Okurut & Ssewanyana (2007) examining the Uganda s labour market response to economic reforms during 1992/93 to2002/03- whereas private sector employment increased during this period, real wages declined especially during the period 1999/00 to 2002/03. The latest national figures indicate that the structure of employment has remained unchanged with more than two thirds of the Ugandan population employed in agriculture (EPRC, 2009). As such, the overall sources of income for the majority of Ugandans remain the same. Furthermore, the proportion of household heads receiving some form of wage employment has held steady at about 15 percent. Consequently, it is important to understand why the structure of employment has not changed greatly despite changes in the macro economy as result of implementing reforms. [TABLE 1: HERE] Due to the large subsistence sector, poverty in Uganda still retains an agricultural face. Table 1 also shows the headcount poverty index for Uganda based on the sector of employment of the household head. Heads whose main economic activity is crop agriculture have highest rates of poverty head count across the three survey rounds. Furthermore, the head count poverty rates for heads involved in manufacturing have remained nearly the same. On the other hand, household heads employed in non-crop agriculture have exhibited the most consistent reduction in poverty. The headcount index for this particular group reduced from 41.5 percent in 1999/2000 to 33.8 percent in 2002/03 to 28 percent in 2005/06. The main sub-sectors within non crop agriculture include livestock, fishing and forestry. Among these sub sectors, fisheries 5

7 have registered the largest growth since the 1990s. Incomes from fishing as measured by the value of annual exports peaked at US$ 145 million in 2005 (GoU, 2009). Although the figures for poverty measures are not captured on annually basis like GDP growth, for example, the rate of poverty reduction has lagged behind the rate of GDP growth. As argued by authors examining other developing regions notably Asia e.g. Khan (2007), such a scenario is caused by low overall employment intensity of growth that is the inability to make growth sufficiently employment intensive. As result of the dismal performance of agriculture, the sector s contribution to overall GDP has declined from about 49 percent of GDP to about 24 percent of GDP by 2008/09 (Table 2). As highlighted by Sender and Uexkull (2009), the table shows that the large numbers of agricultural employees are receiving a declining share of earnings. Without addressing such sectoral imbalances in the levels of growth, the levels of deprivation and inequality may rise in Uganda as earlier mentioned. The table also that the structure of the Ugandan economy currently favours services. However, most of the new growth sub sectors such as transport and communication are mainly urban based and as such will not contribute significantly to employment growth in rural areas. On the other hand, there is limited growth in sectors that could provide backward linkages with agriculture as most of the manufacturing is focused on household goods. Overall, although the structure of the Ugandan economy (in terms of contributions to GDP) has changed in the recent past, the structure of employment has not followed a similar path. [TABLE 2: HERE] 3.0 Methodology and Data 6

8 3.1 Model Following the study by Cook (1998), which examines the who get what kinds of jobs in China s countryside, we adopt the multinomial logit model, in order to examine the extent to which poverty is limiting factor in acquiring non-farm employment in rural Uganda. Consider an individual faced with m employment choices. Let jk determine the probability that individual j chooses employment alternative k e.g. non-farm wage employment, let X j represent the characteristics of individual age gender and education attainment and let Z jk be the characteristics of the th k employment choice for individual j. For the multinomial logit, for each of the m employment choices, the probability that individual j chooses alternative k is given by (2) jk = exp( β ' X ) = m m exp( β l = l ' X 1 j ) l = j exp[( β β )' X )] 1 1 l k j Where β 1,...,βm are m vectors of the regression parameters. It is equation 1 that is estimated using the most recent UNHS data. 3.2 The Data As mentioned earlier, we use the most recent dataset collected by the Uganda Bureau of Statistics in 2005/2006. There are important reasons for focusing on this particular surveys and not earlier UNHS surveys. First, it is not possible to compare the labour module in the 2005/06 survey with earlier surveys due to the changes in the questionnaire. As highlighted by previous authors e.g. Sender and Uexkull (2009) and Fox (2009), the 2005/06 considered a broad range of employment categories not 7

9 considered in either the 2002/03 or 1999/2000 UNHS surveys. Second, the agricultural module of the 2005/06 surveys provides the most recent information on employment and earnings in the agricultural sector the sector that accounts fir the largest share of employment in Uganda. The 2005/06 survey, like previous national surveys, is modelled along the lines of the World Bank s Living Standards Measurement Surveys (LSMS), were meant to capture changes in the welfare status of Ugandans. The survey is nationally representative covering 7,427 households in 2005/2006. All the surveys are based on a two-stage simple random sampling design. In the first stage, the Enumeration Area (EA) is the principal sampling unit, and at the second stage, 10 households are randomly selected from each EA. The 2005/06 surveys have a socio-economic, community, and an agricultural module. In addition, the survey has a detailed labour force module which inquires about the activity status of all individual aged 5 years and above. The socio-economic modules capture information on household demographics (i.e. age, sex, marital status, and relation to the household), socio-economic characteristics (education attainment, household consumption, ownership of land, and ownership of other household assets). On the other hand, the labour force module capture information relating to the employment status (i.e. whether an individual is an employer, own account worker, employee, student, or unemployed), the main and secondary sectors of economic activities (e.g. agriculture, manufacturing, trade, public sector etc) as well as the occupations for persons employed (professional, clerical and semi-skilled or labourer unskilled employee) for all individuals aged 5 years and above. In addition, the survey inquires whether an individual receives a wage as well as the value of the 8

10 remuneration received. All the above information is captured for both the past 7 days as well as the previous 12 months preceding the survey. The community module captures availability and access to infrastructure in the locality. In the analysis, we match the individual worker characteristics to the household and community characteristics where the individual is resident. In the next sub-section, we describe the variables we use and the justification for their selection. 3.3 Variables included in the analysis Employment status: As earlier mentioned, for all usual and regular household members aged 5 years and above, the surveys inquire about the main employment status as well as the nature for those reporting employment. We define non-farm employment to include individuals who are: employers, own account workers, and employees not involved in agriculture. In addition, we consider other employment states such as working in the informal sector. We define informal sector employment to include persons who are self employed, domestic workers and wage earners who are not involved in professional or technical fields. 1 Furthermore, separate categories are used in the estimation of the multinomial logit regressions and the specific categories are defined in the results section. Demographics: In order to capture household demographic composition the following household characteristics were considered: sex, age, household size, and the number of other employed persons. The gender variable accounts for possibility of female discrimination in the labour market while the age variables accounts for the fact that middle aged persons are more likely to secure paid employment than for example 1 In accordance with international conventions the agricultural sector is excluded when defining the informal sector in developing countries. 9

11 recent school graduates. Finally, we include household size to account for the possibility of the presence of other household members who main also be active in the labour force. For example, as shown by the report by EPRC (2009), women in urban areas have lower labour force participation compared to rural women due to the presence of high earning males in the households. Socio-economic characteristics: In line with other studies analysing sociobehavioural outcomes in developing countries, consumption expenditure is used as the household welfare measure. Consumption expenditures adjusted for intrahousehold inequalities (household age and composition effects) using adult equivalence scales is the proxy measure of household socio-economic status. However, in the regression analysis, we include dummy variables for per capital expenditure quintiles as the explanatory variables. Furthermore, in the descriptive analysis, we also use indicators for per capita consumption expenditure quintiles in addition to the actual poverty status. Other socio-economic characteristics used relate to the education attainment of the individual. We consider both the highest number of years of formal education attained as well as four categories of education levels, namely: no education, some primary education, completed primary schooling, and secondary education and higher education. The education indicator represents accumulated human capital as well as skills acquired over time. In addition, we also include the value of household assets such as household land (agricultural and non-agricultural) owned measured in acres. In the literature, it is highlighted that the ownership of large tracts of land may act as hinder the search for non-farm employment (Diechmann et al., 2009). 10

12 Access to infrastructure: We account for community access to key infrastructure utilising the following variables: having a electricity within the community, distance to the nearest tarmac road, having a microfinance institution within 5 kilometres of the community centre, having a factory employing at least 10 people within 5kms of the community, and having either a general consumer goods, agricultural product or input market in the locality. Table 3 provides the means of the variables used in the estimation. [TABLE 3: HERE] 4.0 Results This section provides the econometric results of the paper. First, we profile the major sectors of employment for the households. This is followed by an analysis of extent of wage employment in Uganda by per capital quintiles in 2005/06. Next, we examine the extent of informal employment by various socio-economic characteristics. Finally, we examine the constraints to rural non-farm employment using the multinomial logit model. 4.1 Major sectors employment Table 4 presents the major sectors of employment for individuals in rural households in 2005/06. We categorise the households based on their status on the income distribution using per capita consumption quintiles. The table indicates that agricultural activities are the most important source of income. In particular, both crop and non-crop agriculture accounted for 73 percent of total employment in 2005/06. Other major sectors include trade and manufacturing sectors accounting for about 9% and 5 % of employment respectively. [TABLE 4: HERE] 11

13 The table also shows that individual from households in the poorest quintiles depend most on agricultural activities. In the bottom two quintiles, the proportion of individuals in the rural areas that report either crop or non crop agriculture as the main activity is over 85 percent. Agriculture is also the most important for even individual from the richest households; however, for these relatively well-to-do individuals, trade and public services are also important sectors of employment account for 19 and 13 percent respectively for individual employment in the top quintile. Overall, the above results are consistent with what was observed by previous authors on Uganda that households responded to declines in the agricultural sector by shifting to rural non-farm employment. However, as noted by Kigima et al. (2006) the shift of agricultural households to rural non-farm employment may not compensate fully for the loses in agricultural incomes due to the predominantly casual nature of non-farm employment. Also worth noting is the fact that the current sources of macro economic growth notably transport and communication is still dominated by households in the top quintile compared to the bottom quintile. 2 Overall, Table 4 shows that the sources of employment have remained agriculture despite implementation of economic reforms and diversification programs. 4.2 Extent of informal sector employment In SSA, informal sector employment has been a key driver of urban employment (Fox and Sekkel, 2008). In Table 6, we profile the distribution of female and male employment in the Ugandan informal sector. A number of issues emerge from the table. First, as expected, the poorest individuals are least likely to be employed in the informal sector as this particular sector exists predominantly in urban areas. Second, 2 According to the 2009/2010 Background to the Budget, growth in the transport and communication sector averaged 15.1 percent over the past 5 years ( ) compared to the overall GDP growth of 7.8 percent during the same period. 12

14 the proportion of men in the informal sector is about double that of women. Indeed, for the first four quintiles, the proportion of men in the informal sector is more than double that of men. Third, the gaps in informal sector employment by level of education are much narrowed compared to gaps per capita expenditures. For example, about 11 percent of women with no education worked in the formal sector compared to 13 percent for women with primary education and 17 percent for women who have completed primary education. On the other hand, the proportion of males working in the informal sector was about 30 percent regardless of the education level attained. Overall, the comparisons based on education attainment reveal that informal sector employment is only remotely related to education attainment. Finally, as earlier noted, informal sector employment is most prevalent in urban areas especially in Central Uganda. Related, women in Eastern and Western Uganda are least likely to work in the informal sector. This suggests that most of informal activities in these specific regions are highly manual and as such not suited to women. On the other hand, Northern Uganda exhibits higher than expected rates of female employment in the informal sector in 2005/06. This particular result may be partly explained by the conditions of civil war experienced in Northern Uganda since the late 1990s. The environment of insecurity and the associated large internal displacement and consequent confinement may have created opportunities for informal employment such as beer brewing activities that predominantly undertaken by women. 4.3 Constraints to employment in rural Uganda 13

15 The previous sections have shown that without increasing rural non-farm employment, poverty and inequality. To what extent is poverty a constraint for rural individuals to get into non-farm employment? In order to investigate this particular issue, we estimate multinomial logit model for employment choice. We contrast agricultural and off-farm employment choices in order to examine constraints to various spectrum of employment categories. The categories considered are: agricultural non wage employment (as the base category) 3 ; agricultural casual wage employment; regular non farm wage employment; causal non farm employment; non-farm self employment and those who are unemployed. As individuals in the rural and urban areas face a different labour environment (e.g. there is limited agricultural employment in urban areas due to shortage of land), we only focus on individuals in the rural areas in this particular analysis. The results for multinomial logit estimates are presented in Table 7. The columns show the estimated odd ratios and corresponding z-statistic in brackets. Odd ratios greater than one indicate that higher values (for continuous variables) or the explanatory variable taking on a value of one, increases the predicted probability of non-farm employment compared to agricultural non-wage employment while the opposite is true for coefficients less than one. For example, the coefficient of 1.3 for females for the choice of non-farm regular employment indicates that women have a 30 percent lower chance of working in this category compared to agricultural non wage employment. Similarly, the coefficient of for the third quintile indicates that the relative risk of an individual employed in non-farm regular employment is 3 Agricultural non wage employment includes individuals who are self employed in agriculture and those who are un-paid family workers. 14

16 11.3 percent higher for individuals in the third quintile compared to individuals from the bottom quintile. Overall, the table shows that gender matters for employment choice in rural Uganda. Specifically, women are 30 percent, 60.4 percent, and 4.6 percent less likely than men to be employed in non-farm regular employment, non-farm casual employment and non-farm self employment respectively. Other demographic characteristics show that increases in age significantly increases the likelihood of being unemployed. Education is also an important factor for employment choice in rural Uganda. For example, compared to individuals with no education, completing primary schooling increases the likelihood of non-farm regular wage employment by 25 percent. For the same employment category, attaining secondary education or higher increases the probability of non-farm regular employment by as much as 270 percent compared to individuals without any education. Nonetheless, the education coefficients for nonfarm casual wage employment are relatively much lower compared to non-farm regular wage employment. For example, attaining secondary or higher education increases the likelihood of non-farm casual employment by only 3.3 percent compared with individuals with no education. In order to determine the magnitudes of the impacts of the variables on the various employment choices, we estimate the marginal effects for the multinomial logit model. The results are reported in Table 8 and in the following discussion, we focus on the magnitudes of the poverty indicators as measured by the per capita expenditure quintiles. It is indicated in Table 8 that poverty matters for causal agricultural wage 15

17 employment, regular non-farm employment and self employment. In particular, an increase in an individual s welfare status (proxied by movement up the quintile ladder) is significantly associated with a reduced probability of working in the agricultural causal employment category. Compared to the poorest quintile, individuals from the richest quintile are 7.3 percent less likely to be involved in causal agricultural wage employment. On the other hand, increase in welfare status is associated with increased employment in non-farm regular employment. However, the magnitude of the changes are relatively small in comparison to the choice of agricultural casual wage employment. For non-farm causal wage employment, it is only individuals from the top quintile that are significantly more likely to be employed in this category. Similarly for nonfarm self employments, most of the coefficients are positive but insignificant. Indeed, it only individuals from the richest quintile who are 5 percent more likely to be employment in non-agricultural self employment compared to individuals from the bottom quintile. The above results suggest that poverty is major constraint to nonfarm employment in rural Uganda. Poor people can not get into regular or casual wage employment probably due to lower education attainment and also can not engage in non-farm self employment probably due to limited or no start up capital. 4. Conclusions and Implications This study examined the relationship between poverty, employment, and inequality in Uganda during the time the country registered declines in poverty albeit with declining agricultural growth. Using macro indicators together with the most recent national household survey, we found that women in rural areas are least likely to be employed in non-farm employment. Also, education and poverty status matter for 16

18 particular employment categories especially agricultural casual wage employment and non-farm regular employment. Also, ownership of large tracts of land reduces the likelihood of non-farm employment in rural Uganda although the magnitudes of the impacts are relatively small compared to either education or poverty status. Poverty remains a key constraint to rural non-farm employment. Specifically, we showed that individuals from poor households are most likely to acquire agricultural casual wage employment and are least represented among the ranks of non-farm regular employment or self employment. The above situation can linked to a low asset base for most rural households coupled with binding credit constraints despite the surge in microfinance institutions targeting rural individuals. Furthermore, most of the new opportunities for self employment (e.g. created by expansion in regional exports and informal cross border trade) have benefited urban individuals. With more than two thirds of the country dependent on agriculture, poverty is unlikely to reduce much further without increased agricultural incomes. Apart from weather shocks, increased agricultural incomes in Uganda will depend on diversification into new products as well as use of modern farming practices e.g. using hybrid seeds. Uganda s high population growth rate (over 3 percent per annum) and low primary school completion rates imply that the labour market is receiving an increasing number of unskilled graduates. Such a calibre of workers is unlikely to acquire employment with adequate incomes (NPA, 2010) and this in turn is likely to exacerbate the challenge of inequality. Indeed, poorly educated school leavers have limited opportunities mainly in agriculture or self employment. As such, this type of workforce is unlikely to grow into future employers of the expanding population. 17

19 Consequently, without attaining much higher primary school completion rates, Uganda will remain saddled with a large low quality labour force. Public policies to create new employment opportunities in Uganda have focussed more on interventions targeting the urban employed and less on rural individuals employed in agriculture. Activities such as vocational training, promotion of industrial parks, and entrepreneurship training are all geared towards the urban unemployed. If successful, such public programs run the risk of widening income inequality especially as agricultural incomes decline. On the other hand, recently proposed initiatives such as cluster development which aims to promote localised production of high value agricultural products, may have the ability to create new opportunities for rural non-farm employment. Although other public programs such as labour based road construction have the potential to generate non-farm employment, such programs are nonetheless heavily dependant on donor financing and as such are un-sustainable in the long term. 18

20 6. References Bridges, S and D. Lawson (2009) A Gender-based Investigation into the Determinants of Labour Market Outcomes: Evidence from Uganda Journal of African Economies Vol 18. No. 3: Canagarajah, S., C. Newman, and R. Bhattamishra (2001) Non-farm income, gender, and inequality: evidence from rural Ghana and Uganda Food Policy Vol 26. No. 4: Cook, S (1998) Who gets what jobs in China s countryside? A multinomial logit analysis Oxford Development Studies Vol. 26. No Deichmann, U., F. Shilpi, and R. Vakis (2009) Urban Proximity, Agricultural Potential and Rural Non-farm Employment: Evidence from Bangladesh World Development Vol 37. No.3: EPRC (2009) Gender and Productivity Survey: Analytical Report mimeo EPRC. Fox, L and M. Sekkel (2008) Working out of poverty: job creation and the quality of growth in Africa. Washington DC: The World Bank. Government of Uganda (2009), Background to the budget 2009/2010 Fiscal Year: Enhancing Strategic Interventions to Improving business, and revitalise production to achieve prosperity for all. Kampala, Ministry of Finance Planning and Economic Development., (2006) Uganda HIV/AIDS Sero-Behavioral Survey Ministry of Health and ORC Macro. Kampala. Haan, H. C Training for Work in the Informal Sector: New Evidence from Kenya, Tanzania and Uganda. In Focus Programme on Skills, Knowledge and Employability, Working Paper, International Labour Office, Geneva. Kappel, R. T., S. Steiner and J. Lay. (2005) Uganda: No more Pro-poor Growth? Development Policy Review Vol. 23, No. 1: Khan, A. R (2007) Asian Experience on Growth, Employment and Poverty. An overview with special reference to the findings of some recent case studies UNDP and ILO. Kijima, Y., T. Matsumoto and T. Yamano (2006) Non-farm employment, agricultural shocks and poverty dynamics: evidence from rural Uganda Agricultural Economics Vol. 35: supplement pp Ministry of Finance, Planning and Economic Development (2009) Budget Speech Delivered to the parliament of Uganda in June Okurut, N. F and S. Ssewanyana (2007) Determinants of Wage Earnings in Uganda The IUP Journal of Agricultural Economics. Vol 4. No.2: Ravallion, M (2009) Are there lessons for Africa from China s success against Poverty World Development Vol 37. No. 2: Ravallion, M., and S. Chen. (2007). China s (uneven) progress against poverty. Journal of Development Economics, 82(1), Sebbagala, R (2007) Wage Determination and Gender Discrimination in Uganda EPRC Working Paper No

21 Sender, J and E. Uexkull (2009) A rapid impact assessment of the global economic crisis on Uganda International Labour Organisation. UBoS (2007) Uganda National Household Survey : Social Economic Report. Kampala., (2003) Uganda National Household Survey 2002/2003. Uganda Bureau of Statistics, Entebbe, (2000) Uganda National Household Survey 1999/2000. Uganda Bureau of Statistics, Entebbe. World Bank (2009) Appraisal document for the Second Northern Uganda Social Action Fund (NUSAF-2), (2006) World Development Report 2006: Equity and Development. Washington, DC. World Bank. 20

22 Tables Table 1: Trends in Head count Poverty by sector of employment of household head, Survey Round 1999/ /3 2005/6 Uganda Sector of employment Crop Agriculture Non-Crop agriculture Mining Manufacturing Public utilities Construction Trade Hotels Transport/Communications Other services Government services Not employed Source: UBOS UNHS Reports 2002, and Table 2: Contribution to GDP by industry, 2001/ /09 (%) Industry 2001/ / / / / / / /09 Agriculture Cash crops Food crops Livestock Forestry Fishing Industry Manufacturing Formal Informal Construction Services Wholesale and Retail Trade Hotels and Resturants ransport and Communication Financial services Education Other services Other sectors Total Source: Bank of Uganda Annual Report 2009 and Background to the Budget 2009/

23 Table 4: Sector of employment of individuals in rural Uganda per capita welfare quintiles 2005/06 Quintiles Row Average Crop agriculture Non-crop agriculture Manufacturing/ Mining Construction Trade Hotels Transport/communication Government services Other sectors Sub Total Source: Author's calculations based on the 2005/06 UNHS survey Table 3: Statistics of the variables used (individuals resident in rural areas) Variable Mean Std. Dev. Min Max Gender (female=1) Age in years Education attainment No education Some Primary Education Completed Primary Secondary and higher Type of employment Agriculture unpaid worker Agriculture casual wage employment Non-farm regular employment Non-farm casual employment Non-farm own account worker Unemployed Household size Per capital expenditure quintiles quint_ quint_ quint_ quint_ quint_ Log of acres of land owned Community characteristics Availability of consumer market within 1km of the village Availability of product market within 1km of the village Availability of electricity within the community Availability of a factory that employs 10 people with 5kms of the community Availability of microfinance institution with 1km of the community Log of distance to trunk road Central Eastern Northern Western

24 Table 7: Multinomial logit model for employment choice in rural Uganda, 2005/06 Employment choice (agricultural non-wage employment is the base outcome) Agricultural Non farm regular Non farm Non farm Unemployed casual wage wage employment casual self employment employment employment Gender (female=1) [8.98]*** [7.84]*** [10.83]*** [13.1]*** [1.52] Log of age in years [0.82] [5.06]*** [5.86]*** [5.11]*** [12.72]*** Education attainment Some Primary [0.94] [0.81] [0.59] [31.9]** [6.02]*** Completed primary [0.02] [2.56]** [2.12]* [4.86]*** [4.31]*** Secondary education or higher [2.62]** [8.52]*** [4.07]*** [5.21]*** [2.22]* Household size [4.51]*** [0.37] [0.49] [0.48] [3.30]*** Per capita expenditure quintile Quintile [2.05]* [1.90] [0.92] [0.87] [0.13] Quintile [3.72]** [3.15]*** [0.38] [1.02] [0.41] Quintile [4.13]*** [4.03]*** [0.63] [1.36] [1.41] Quintile [5.16]*** [6.28]*** [2.03]* [4.81]*** [2.66]** Log of household land owned (acres) [7.05]*** [2.95]** [2.66]** [3.71]** [0.81] Agricultural consumer market within 5km [0.62] [2.30]* [1.28] [0.28] [0.59] Agricultural produce market within 5km [1.72] [1.20] [0.58] [2.67]* [0.41] Availability of electricity within LC [1.36] [3.30]*** [5.88]*** [6.69]*** [2.84]** Existence of major factory within 10km [1.48] [0.42] [0.47] [0.57] [0.39] Credit no collateral within 10km [3.12]** [2.19]* [2.74]** [0.93] [0.01] Log of distance to trunk road (kms) [0.97] [0.75] [0.72] [0.08] [1.81] Regions Central [7.2]*** [4.31]*** [0.67] [2.57]** [0.86] Eastern [7.23]** [0.64] [1.81] [3.86]** [1.24] Western [3.73]*** [1.03] [1.01] [3.69]*** [2.33]* Constant [1.15] [10.07]*** [0.21] [9.27]*** [15.08]*** Source: Author's calculations from 2005/06 UNHS 23

25 Table 8: Marginal effects for Mlogit estimates for employment choice in rural Uganda, 2005/06 Employment choice (agricultural non-wage employment is the base outcome) Agricultural Non farm regular Non farm Non farm Unemployed casual wage wage employment casual self employment employment employment Gender (female=1) [-7.49]*** [-5.27]*** [-10.21]*** [-12.64]*** [-1.16] Log of age in years [0.95] [5.04]*** [-6.03]*** [6.46]*** [12.75]*** Education attainment Some Primary [2.15]* [1.57] [1.74] [5.74]*** [-5.78]*** Completed primary [0.94] [3.65]*** [3.42]*** [7.75]*** [-4.43]*** Secondary education or higher [-2.02]* [7.48]*** [5.61]*** [8.37]*** [-2.32]* Household size [-4.69]*** [-0.11] [0.22] [-0.28] [3.11]*** Per capita expenditure quintile Quintile [-2.06]* [1.98]* [1.01] [-0.81] [0.16] Quintile [-3.93]*** [3.19]*** [0.44] [1.21] [0.41] Quintile [-4.36]*** [3.95]*** [-0.58] [1.60] [1.37] Quintile [-5.81]*** [5.27]*** [2.04]* [5.20]*** [2.49]** Log of household land owned (acres) [-6.42]*** [-2.13]* [-1.47] [-2.39]* [1.42] Agricultural consumer market within 5km [0.31] [2.27]* [1.34] 0 [0.41] Agricultural produce market within 5km [-1.97]* [-1.18] [0.36] [2.95]** [-0.34] Availability of electricity within LC [0.41] [2.11] [4.94]*** [5.81]*** [2.24]** Existence of major factory within 10km [-1.32] [0.51] [0.51] [-0.46] [-0.23] Credit no collateral within 10km [-3.55]*** [1.92]* [2.65]** [0.77] [-0.1] Log of distance to trunk road (kms) [-0.96] [1.02] [-0.58] [0.38] [1.89]* Regions Central [-6.62]*** [-3.65]*** [1.19] [-1.6]* [-0.03] Eastern [-6.54]*** [-0.39] [-1.19] [-3.02]* [-0.63] Western [-3.11]** [-0.84] [1.51] [-3.07]* [-1.93]* Constant [1.13] [-7.38]*** [0.41] [-9.66]*** [-13.38]*** Source: Author's calculations from 2005/06 UNHS 24