Bank of Uganda. Working Paper Series. Working Paper No. 03/2016. Structural Change and Economic Growth in Uganda. July 2016

Transcription

1 Bank of Uganda Working Paper Series Working Paper No. 03/2016 Structural Change and Economic Growth in Uganda Martin Brownbridge and Thomas Bwire Bank of Uganda July 2016 Working papers describe on-going research by the author(s) and are published to elicit comments and to further debate. The views expressed in the working paper series are those of the author(s) and do not in any way represent the official position of the Bank of Uganda. This paper should not therefore be reported as representing the views of the Bank of Uganda or its management

2 Bank of Uganda WP No. 03/2016 \ Structural Change and Economic Growth in Uganda Prepared by Martin Brownbridge and Thomas Bwire Bank of Uganda July 2016 Abstract This paper provides an empirical assessment of structural change in Uganda since the late 1990s. It utilizes a range of indicators, including the sectoral composition of output and exports, the composition of the labour force, labour productivity and data on registered businesses and large firms. Although the Ugandan economy has realized sustained real growth since the 1990s, structural transformation has been limited. In particular there has not been a large scale shift of labour from low productivity traditional activities into wage employment in high productivity formal sector firms. Consequently labour productivity growth has been weak, at around 2 percent per annum in real terms. The paper concludes by discussing some of the constraints to structural transformation in Uganda. JEL Classification: O10, O14, O49, O55. KEY WORDS: Economic growth, Structural Transformation, Uganda. To cite this article, Martin, B., and Bwire, T. (2016), Structural Change and Economic Growth in Uganda, Bank of Uganda Working Paper No. 03/2016. Correspondence Address: Research Department, Bank of Uganda, P.O. Box 7120, Kampala, Tel , Fax Authors addresses: MBrownbridge@bou.or.ug and TBwire@bou.or.ug. The views as expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the Bank of Uganda. 1 P a g e

3 Introduction The key characteristic of structural change, which is integral to socio-economic development, is growth in labour productivity which enables living standards to be raised in a continuous, dynamic manner. Although it is possible for living standards to be raised in the absence of structural transformation, for example through the exploitation of natural resources, such increases are usually not sustainable. The normal historical pattern of structural transformation has involved large scale shifts in the nature of employment, with labour shifting from traditional informal sectors, such as peasant agriculture or petty commodity production, to modern industries, as in the stylised two sectors Lewis model of economic development (Lewis, 1954). The traditional sectors are characterised by rudimentary levels of technology and little use of capital, often involving production for subsistence rather than the market, a very small scale of production and, consequently, low and stagnant labour productivity. In contrast, the modern sectors are characterised by capitalist production, in which wage labour is combined with capital in profit oriented firms. The much more intensive use of capital together with the larger scale of production means that labour productivity is much higher in the modern sector than in the traditional sector. Even more importantly, the competitive nature of a capitalist market economy dictates that firms must re-invest the bulk of their profits to expand production, improve their efficiency and develop new products, thereby imparting a dynamic which raises labour productivity through time. Large scale shifts of labour from the traditional to the modern sectors of the economy have usually been accompanied with, or even preceded by, the modernisation of agriculture which raises productivity in that sector and by a demographic transition which entails lower fertility rates and lower age dependency ratios. Such patterns of structural transformation characterise the type of rapid growth and development which has taken place in many East Asian countries over the last few decades (Rodrik, 2013; Zhu, 2012). The stylised facts of structural change thus include: i) a shift in the composition of GDP, with the share of agriculture falling and that of industry and services increasing; ii) a shift in the composition of the labour force from self-employment, or employment in household enterprises, to wage employment in the formal sector; iii) rising labour productivity; iv) growth in the number of formal sector firms and in their contribution to GDP; and v) a 2 P a g e

4 diversification of export earnings away from primary commodities and towards processed and manufactured products. Several authors have identified a lack of structural change in sub-saharan Africa (SSA) as a major challenge for development (e.g. Ajakaiye and Page, 2012; Fox et al, 2013; Jedwab, 2013; Vries, Timmer and Vries, 2014). Manufacturing output as a share of GDP has stagnated in SSA, the vast majority of the workforce still works in the informal sector, and the shifts in the structure of output and (to a lesser extent) employment that have occurred have mainly entailed a shift from agriculture to informal sector services. The shift of labour into modern, tradeable goods industries of the type which has characterised East Asian development has not occurred in SSA. Most of the labour force in SSA still works in self-employment and household enterprises rather than as wage workers in formal sector firms. The objective of this paper is to examine and analyse the empirical evidence on the nature and extent of structural change in Uganda. In section 2, we explore several different indicators of structural change over the last years. We look at sectoral shares in GDP and the composition of employment, in terms of sector of activity and whether workers are selfemployed or wage workers. We also look at trends in employment in registered businesses and at more recent data on the value added, investment and employment by the large firms covered by the Private Sector Investment Survey. We also look at trends in the composition of merchandise exports. Section 3 concludes by drawing together the evidence from section 2 and then discussing why there has been only very limited progress towards structural change in Uganda. 2. Structural Change in Uganda: What does the data tell us? 2.1 Sectoral shares in GDP As shown in figure 1, Uganda s economic growth since the late 1990s has mainly been driven by services. The share of services in GDP increased from 42 percent in 1998/99 to 50.2 percent in 2014/15. The downside is that some of these activities and, in particular, telecommunications, finance and real estate are not employment intensive and instead rely on a relatively small number of skilled workers. The industrial sector s share of GDP has also grown, but only modestly so, increasing from 17 percent in 1998/99 to 18.7 percent in 2014/15. The growth in the industrial sector s share of GDP is concentrated in the 3 P a g e

5 construction sub-sector. In contrast, the share of manufacturing in GDP remained stagnant at between 8 and 9 percent. The share of the agricultural sector in GDP has fallen from 41 percent in 1998/99 to 23.5 percent 2014/15. Figure 1: Percentage contribution to GDP by economic activity (constant 2009/10 prices) Agriculture, forestry and fishing Industry Services Source: UBOS and Author s calculations 2.2 Changes in the labour market Changes in the way in which people earn their living are at the heart of structural change. In this sub-section, we ask three questions. First, have there been changes in the way in which people earn their living, involving a shift from the self-employment, which is typical of informal, traditional activities, to paid employment which characterizes most modern sector activities? Second, have there been changes in the sectoral composition of employment? Third, how fast has labour productivity increased? We utilize data, presented in Table 1, from four Uganda National Household Surveys (UNHS) to analyze changes in the labour market. Two caveats, however, must be borne in mind. The first is that the Uganda Bureau of Statistics (UBOS) has not presented labour market data from all the household surveys conducted in Uganda in a wholly consistent manner; in particular there are major methodological differences between the 2012/13 presentation of the data and the three earlier surveys. As a result the authors of this paper have constructed some of the variables shown in Table 1 from the data published by UBOS but there may be inconsistencies in the coverage of some of the variables across the four household surveys. Secondly, as discussed by Fox and Pimhidzai (2013), the primary data itself may not be consistent across household surveys because of differences in the framing of questionnaires, 4 P a g e

6 recall periods and computation of data. At the aggregate level, these problems may affect, in particular, estimates of the Labour Force Participation Rate (LFPR) 1 and the sector of employment. In addition, the data on the labour force and working population include working students. It is reasonable to assume that working students are mostly part time workers. UBOS has published data on the working population excluding working students for 2005/06 and 2009/10, 2 but not for 2002/03 and 2012/13. In 2005/06 working students accounted for 12.3 percent of the working population, a share which had risen to 14.5 percent in 2009/10; an increase which is consistent with the expansion in the share of the population in fulltime education. A growing share of working students in the working population implies that a larger share of the working population works part time, and hence the total amount of hours worked will not have increased as fast as the size of the working population. The working age population is classified as all people in the age group of 14 to 64 years. The labour force is a subset of the working age population, and comprises persons engaged in any economic activity or who are unemployed but looking for work. As shown in Table 1, Uganda s working age population rose from 12.2 million in 2002/3 to 16.5 million in 2012/13. However, as a share of the total population, the working age population was the same, 48 percent, in 2012/13 as it had been in 2002/03. The age dependency ratio was therefore 107 dependents (people of non-working age) per 100 persons of working age; one of the highest in the world. This is because the Total Fertility Rate (TFR) in Uganda is still very high at 5.8 births per woman (UBOS, 2014B). As such, Uganda has barely begun its demographic transition. It will not derive a demographic dividend until the TFR has fallen substantially, thereby reducing the age dependency ratio. The size of the labour force expanded from 9.8 million people in 2002/03 to 14.7 million in 2012/13, at an average annual rate of 4.1 percent. This was faster than the growth of the working age population because of a marked increase in the LFPR from 80 percent in 2002/03 1 LFPR measures the proportion of the country s population that engages actively in economic activities, i.e. those either employed (including subsistence production) or looking for work (unemployed), as a percentage share of the working age population. It excludes those who were not working and not actively looking for work, such as retired people. 2 See UBOS (2012), Table 2/.3.1 p18. 5 P a g e

7 to 89 percent in 2012/13, with most of the rise occurring between 2005/06 and 2009/10 (see table 1). It is possible that the increase in the LFPR in the second half of the 2000s is related to the return to the labour force of people who had been internally displaced and were living in camps, because of the Lords Resistance Army (LRA) insurgency in the north of the country. The number of internally displaced people (IDPs), living in official IDP camps in June 2004 was 1.6 million, although the actual number of IDPs may have been around 2 million (Global IDP database, 2005). If the share of the working age population in the total population of IDPs mirrored that of the country as a whole, the IDPs of working age would have been about 780,00 in the mid-2000s, based on the figures of IDPs in official camps. That was about 6 percent of the working age population of the country. Hence it is probably the case that the LFPR was about six percentage points lower; at least, in the first half of the 2000s, than it would have been if the LRA insurgency had not occurred. In the second half of the 2000s, when peace was restored to the North, most of the IDPs left the camps which would have enabled the LFPR to rise by approximately 6 percentage points, which is more than 60 percent of the recorded rise in LFPR between 2005/06 and 2009/10. If IDPs who were not living in official camps are also taken into account, the contribution of IDPs returning to the labour force to the rise in the LFPR may have been higher. However, the estimated rise in the LFPR may reflect changes in the methodologies or questionnaires used in the household surveys, rather than an actual increase in the LFPR. The working population, which consists of the labour force minus the unemployed, is classified as being in either self-employment or in paid employment. Self-employment accounts for the majority of the working population, although its share has fallen. Conversely, the share of the working population which is in paid employment has risen from 15 percent in 2002/3 to 27 percent in 2012/13. A shift from self-employment, which includes subsistence agriculture, to paid employment is an indicator of positive structural change in the labour market. 6 P a g e

8 Table 1: Labour Market Indicators 2002/ /13 Indicator category 2002/ / / /13 Population (000's) 24, , , ,759 3 Size of working age population (000's) 12,235 13,272 14,598 16,502 Labour force participation rate (LFPR, %) Size of labour force (000's) 9,800 10,883 13,357 14,713 o/w Size of unemployed working age population (000's) o/w Persons in employment (000's) 9,457 10,676 12,890 13,896 a) Persons in paid employment (000's) 1,428 1,740 2,668 3,738 b) Persons in self-employments (000's) 8,029 8,936 10,235 10,158 Unemployment rate Activity Status of the Work Force a) Self-employed (including those in subsistence production) 1, % b) Paid employee 2, % Sector of employment a) Agriculture, forestry and fishing b) Service c) Manufacturing Total Age Dependency Ratio Source: UBOS (UNHS, various) and Author s calculations. However, not all paid jobs are in the formal sector or are of high productivity; many are in micro and small scale informal enterprises or are in domestic employment. The formal sector workforce in Uganda comprises government workers and workers in formal sector business enterprises, plus those working for other public entities, NGOs, international organizations, foreign missions etc. In 2011 there were 275,000 civil servants and 430,000 workers in formal sector business with fixed premises (see section 2.4 below). Together these amounted to just over 700,000 workers. The numbers of other formal sector workers employed by NGOs, etc are unlikely to be more than a few tens of thousands at most, hence the number of formal sector employees in 2011 was probably between 710,000 and 750,000, which was approximately 5-6 percent of the total working population. 3 Therefore, of the 27 percent of the working population characterized as being in paid employment in the latest UNHS, only about one fifth had a formal sector job, with the remaining 80 percent, which was about 21 percent of the working population, working in informal sector jobs. 3 Data on contributions to the National Social Security Fund (NSSF) suggest an even lower number of formal sector workers. In 2011, about 320,000 employees paid contributions to the NSSF. Contributions are mandatory for workers in establishments of five or more workers, except for Government employees. If we add the civil servants to the workers paying contributions to the NSSF, formal employment would have been only about 600,000 in P a g e

9 Table 1 also provides data on the composition of the workforce by sector. These data pertain to the sector of the primary activity of the working population. It does not capture secondary activities, which have become increasingly important, as discussed further below. Between 2002/03 and 2012/13 there were small changes in the shares of the working population by sectors of employment. The share working in agriculture fell by 3 percentage points to 62 percent while there was an increase of similar magnitude in the share employed in manufacturing, which rose to 9 percent. The sharp increase in the share of the employed population in manufacturing between 2009/10 and 2012/13 is not consistent with the sector s stagnant share in GDO over this period and it may simply reflect a reclassification of workers from the service to the manufacturing sector. The share employed in services was 28 percent in 2012/13, the same as in 2002/03. It is difficult to explain why there was an increase in the share of the working population in agriculture between 2002/03 and 2005/06 which was subsequently reversed. These data only capture the primary income generating activity of the working population and, for the reasons discussed in section 2.3 below, they may underestimate the inter-sectoral shifts in labour inputs which took place since the turn of the century. 2.3 Labour Productivity The trends in labour force productivity must be treated with caution. They are derived by dividing real output by the number of workers, either for the economy as a whole or for major sectors. However, if changes in survey methodologies have led to more accurate estimates of the LFPR in later surveys, as suggested by Fox and Pimhadzia (2013), the growth in the labour force between 2002/03 and 2012/13 may be overestimated and hence labour productivity growth will be underestimated. Furthermore, if the share of working students in the working population has risen, it is likely that the share of part time workers in the working population also increased through time, which implies that the hours worked per worker have not increased as fast as the size of the working population. Hence productivity growth in terms of output per hour worked will be faster than that estimated by dividing output by the number of workers. On the other hand, the data in the 2009/10 and 2005/06 UNHS indicate an increasing share of the working population undertaking a secondary economic activity, in addition to their 8 P a g e

10 primary activity. In 2005/06, 22 percent of the working population undertook a secondary activity, compared to 28 percent in 2009/10. In the latter year, rural inhabitants were much more likely (31 percent) to engage in secondary activities than urban inhabitants. As a large majority of rural inhabitants cite agriculture as their primary activity, it follows that over time, an increasing number are undertaking a secondary, non-agricultural activity. This suggests that the hours devoted to non-agricultural activities (e.g. manufacturing and services) are actually greater than the data on primary activities indicate, and that this trend has risen through time as the share of the working population engaged in secondary activity has increased. It could also mean that the hours devoted to agriculture are lower than indicated by the data on primary activity, if workers have substituted hours spent in non-agricultural activity for hours spent in agricultural activity. But it could also mean that total hours per worker have increased, if workers whose primary activity is agriculture do not reduce the hours devoted to this activity but work more hours to accommodate non-agricultural secondary activities. Table 2: Labour productivity (constant 2009/10 prices, billions of shillings) 2002/ / / /13 GDP at market prices (Billions of Shillings) 23,143 28,994 38,017 44,620 o/w Agriculture, forestry and fishing 9,568 9,964 10,732 11,364 o/w Manufacturing 2,140 2,672 3,481 3,759 o/w Services 11,435 16,358 23,804 29,497 Working Population (000's) 9,457 10,676 12,890 13,896 o/w in Agriculture, forestry and fishing 6,194 7,644 8,456 8,674 o/w in Services 2,648 2,562 3,609 3,920 o/w in Manufacturing ,301 Value of Labour Productivity (Millions of Shillings) Overall output per worker Agriculture output /worker Manufacturing output/worker Services output/worker Annual labour productivity growth 2002/ /13 (%) Output 2.0 Agriculture, forestry and fishing -1.6 Manufacturing -1.8 Services 4.1 Source: UBOS and authors calculations Table 2 depicts trends in labour productivity from 2002/03 to 2012/13. At the level of the whole economy, labour productivity increased from Shs 2.6 million per worker per year, in constant 2009/10 prices, to Shs 3.2 million over this 10 year period. This is an average annual 9 P a g e

11 increase of only 2 percent, which is mediocre for a developing economy. By contrast labour productivity has increased at an average of nearly 5 percent and nearly 8 percent in south Asia and East Asia respectively. However, as discussed above, it may be an under estimate of the true increase in labour productivity in Uganda. Labour productivity in agriculture actually fell between 2002/03 and 2012/13, from Shs 1.5 million to Shs 1.3 million per worker per year in constant 2009/10 prices. However, in addition to the caveats already mentioned, it is possible that agricultural workers are also devoting less time to agriculture and more time to secondary, non-agricultural, activities as discussed above, including such activities as motor cycle riding. The 2012/13 UNHS breaks down the labour employed in agriculture into those working in subsistence agriculture and those employed for pay of profit. 4 Two thirds of the agricultural labour force works in the subsistence sector. In the National Income Accounts, there is a breakdown of agricultural value added into marketed and non-marketed value added. If we make the assumption that subsistence workers produce non-marketed output whereas workers employed for pay of profit produce marketed output, it is possible to derive some tentative estimates of labour productivity in the marketed and non-marketed agricultural sectors, as shown in Table 3. The estimated labour productivity in marketed agriculture was nearly five times higher than in non-marketed or subsistence agriculture. Table 3: Structure of Output and the Workforce in Agriculture; 2012/13 Agricultural value added Shs bns 2009/10 prices 11,364 o/w Market 7,817 o/w Non market 3547 Agricultural labour force (000s) 8674 o/w For pay or profit 2665 o/w Subsistence workers 6009 Labour productivity (shs Mns/worker p.a. 2009/10 prices 1.3 o/w For pay or profit 2.9 o/w Subsistence workers 0.6 Source: authors estimates derived from UNHS and National Income Accounts In contrast to agriculture, there were quite strong increases in labour productivity in services, from Shs 5 million per worker per annum in 2002/03 to Shs 7.5 million in 2012/13, an average increase of 4.1 percent per annum. Part of this growth has been driven by the 4 The 2012/13 UNHS was the first household survey from which a breakdown of the agricultural labour force into subsistence workers and workers for pay and profit was derived. 10 P a g e

12 expansion of high productivity services such as telecommunications and finance. It is also possible that services output has benefitted the most from secondary activities of workers, so that productivity growth per hour worked might not be as fast as the data derived from primary activities indicate. Labour productivity in manufacturing actually fell over the period which is difficult to explain other than by a possible reclassification of workers from services to manufacturing referred to above. 2.4 The Business Sector UBOS has conducted two comprehensive surveys of registered businesses since the turn of the millennium, which cover all businesses with fixed premises (but not businesses operated from residential properties).these surveys, called the Census of Business Establishments (COBE) were conducted in 2001/02 and 2010/11. However, the COBE does not collect data on the value added of registered businesses, so it is not possible to compute their contribution to GDP. Table 4: Characteristics of Registered Businesses 2001/ /11 No of registered businesses 160, ,106 o/w No of manufacturing businesses 11,968 31,757 No of employees in registered businesses 444,118 1,074,000 o/w No of employees in manufacturing businesses 87, ,097 Average size of workforce per business Average size of workforce per manufacturing business No of employees in informal businesses (<5 workers) 233, ,400 No of employees in formal sector businesses (5 or more workers) 210, ,600 No of workers in formal manufacturing (five or more workers) 68,358 89,326* Share of formal business employees in working population 2.3% 3.3% Share of manufacturing employment in business employment 20% 13% *estimated Source: UBOS Data from the COBEs are shown in Table 4. The number of registered businesses tripled between 2001/02 and 2010/11 and the number of workers employed more than doubled. The average size of the workforce of these businesses fell from 2.8 to 2.3 workers between 2001/02 and 2010/11. UBOS classifies businesses as either informal or formal according to the size of their workforce, with the former having a workforce of 1-4 workers and the latter 5 or more workers. As a share of all workers in registered businesses, workers in formal sector businesses fell from 47 percent to 40 percent between 2001/02 and 2010/11. There were only 429,000 workers in formal sector businesses in 2010/11 which is only 3.3 percent of the total 11 P a g e

13 employed population. There was strong growth in registered businesses between 2001/02 and , but that growth was fastest among very small, informal sector businesses. Table 4 also provides data on registered manufacturing businesses. Although the number of registered manufacturing businesses, expanded by 165 percent between 2001/02 and 2010/11, employment in manufacturing business as a whole increased by about 60 percent, rising in absolute numbers from 87,131 employees in 2001/02 (20 percent of employees in registered businesses) to 139,097 employees in 2010/11 (13 percent of employees in registered businesses). Consequently the average number of workers per manufacturing business fell from 7.3 to 4.4. Employment in formal sector manufacturing expanded at a much slower rate of 31 percent relative to informal manufacturing employment which increased by 165 percent over the 10 years between the two censuses. Formal sector employment in manufacturing in 2010/11 comprised only about 12 percent of the total employed population in manufacturing implied by the 2009/10 UNHS data; the rest are working in either informal registered businesses or in businesses without fixed business addresses such as artisanal workers working from their homes or other unregistered addresses. The data from the COBEs indicate that the formal private sector firms, which in effect comprise the modern commercial sector of the economy, make a small contribution to the economy, especially in terms of employment. Additional data on the modern sector of the economy can be obtained from the Bank of Uganda s annual survey of large firms, which is called the Private Sector Investment Survey (PSIS). This covers around 650 of the largest firms in the economy. These firms are a subset of the formal sector firms covered by the COBE. Since 2011, the PSIS has provided data on the value added, gross capital investment and the workers employed in the firms covered in the survey. As shown in Table 5, these firms contributed about 5 percent to GDP on average over the four calendar years from 2011 to They contributed about 12 percent of the economy s gross fixed capital formation and a negligible 0.8 percent of total employment. Gross investment by the firms in the PSIS was equivalent to only 2.5 percent of GDP, although this estimate omits investment by the oil companies which have made substantial investments in oil exploration in recent years. Labour productivity in the firms covered by the PSIS was, on average, more than 6 times higher than that of the economy as a whole during P a g e

14 Table 5: Value Added, Investment and Employment by firms covered by the PSIS PSIS firms Value added Shs bns 1,753 2,816 3,900 3,212 Share of GDP (%) 3.4% 4.9% 6.4% 4.9% Fixed investment Shs bns 2,193 2,180 2,078 1,897 Share of GFCF (%) 14.7% 12.7% 11.6% 10.3% Workers 146, , , ,459 Share of Working Population (%) 1.1% 0.8% 0.7% 0.7% Labour productivity PSIS firms Shs mns/worker Labour productivity total economy Shs mns/worker Source: Bank of Uganda The data from the COBE and PSIS do not suggest that the modern business sectors of the economy are very dynamic. Their share of output and employment is small and growing only slowly. 2.5 Exports Uganda s export performance since the turn of the century offers some evidence of positive structural change. Table 6 provides data on exports by SITC category. SITC categories 0-3 are essentially primary commodities such as coffee, categories 4, 5 and 6 are processed primary products such as vegetable oils, categories 7 and 8 are manufactured products and category 9 is gold. Table 6: Exports by SITX category; SITC 0,1,2,3.9 SITC 4,5,6 SITC 7,8 SITC 9 $ mns % share $ mns % share $ mns % share $ mns % share Source Bank of Uganda 13 P a g e

15 In 2002, processed and manufactured products together amounted to $35 million which was only 8 percent of merchandise exports. By 2013 they had increased to $812 million which was more than a third of merchandise exports. 5 The processed and manufactured exports include sugar, vegetable oil, beer and soft drinks and building materials. Most of these products involve goods which were mainly produced for the domestic market, where they enjoy a degree of natural protection, but have increasingly been sold on regional markets, especially that of South Sudan, in recent years. The fall in processed and manufactured exports in 2014 and 2015 reflects the impact of the civil war in South Sudan on demand in that market. 3. What has held back structural change? The evidence reviewed in section 2 suggests that structural change in the Ugandan economy over the last years has been very modest. In particular, the modern, private sector of the economy, comprising formal sector firms, still commands a very small share of output and especially of the labour force. The majority of the labour force still works in agriculture and two thirds of the agricultural workforce comprises subsistence framers. Structural change in Uganda has been impeded by weaknesses in two important areas. First, there has been no green revolution in agriculture. The growth in agricultural output that has occurred has been the result mainly of increases in the area of cultivated land and the labour force employed in agriculture, hence yields per acre and labour productivity have been stagnant. The vast majority of agricultural output in Uganda is produced by smallholder farmers, who use rudimentary farm technologies and produce mainly for subsistence rather than the market. The failure to modernize smallholder agriculture reflects a raft of self-reenforcing constraints. The low rates of adoption of productivity enhancing technologies are caused by smallholders lack of resources and knowledge and inability to bear risks. In turn this means yields are low, soils are often depleted and marketable surpluses are small. Small marketable surpluses raise transactions costs and depress farm gate prices. Consequently returns to farming are low, which leaves smallholders in poverty, with little or no savings and no capacity to bear the risks of trying to modernize their farming (Drew, 2010). 5 The exports of processed and manufactured products comprised just over 10 percent of the gross output of the manufacturing sector in P a g e

16 The second major impediment to structural transformation is the lack of private investment in labour intensive industries. There have been quite substantial levels of private gross fixed investment in recent years, probably in the region of 19 percent of GDP. 6 However most of this investment has been in residential and commercial buildings and in oil exploration, which provides very little employment. Investment in residential buildings alone accounted for almost 40 percent of private sector investment during 2008/ /15. The large firms covered by the PSIS, which have probably accounted for most of the investment undertaken by formal sector firms outside of the oil sector, undertook investment equivalent to just over 3 percent of GDP during It is these firms which provide the bulk of high productivity jobs in the economy. They are not investing enough and, therefore, not expanding quickly enough, to absorb more than a miniscule share of the labour force. The problem of the dearth of private investment in formal sector, labour intensive industries such as manufacturing, is not unique to Uganda; rather it affects most economies on the African continent. There are three theories of why there is so little private investment in manufacturing industry in SSA. One theory emphasises deficiencies in physical infrastructure which raise the cost of doing business; e.g. by increasing the cost of power. The second emphasises the costs of labour in SSA relative to those of potential competitors in manufacturing exports elsewhere in the world, arguing that labour costs are higher in the former and that this differential is not matched by a differential in relative labour productivity (Gelb, Meyer and Ramachandran, 2013). This might reflect overvalued real exchange rates in Africa, because of natural resource flows or large inflows of aid and/or remittances. The third explanation emphasises weaknesses in the institutional environment for business which increase the risks for private investors (Lim, 2013). It must be acknowledged that empirical research has not yet been able to conclusively identify the binding constraint to private sector investment in manufacturing and other, labour intensive, modern sectors of the economy, in Africa. The fact that Uganda has barely begun its demographic transition is also associated with the lack of structural transformation. Uganda s very high age dependency ratio, which results from its high total fertility rate, unavoidably depresses savings rates which, in turn, constrains 6 Since it rebased the National Income Accounts to the 2009/10 base year in 2014, UBOS has not disaggregated gross fixed capital formation (GFCF) into private and public sectors. Total GFCF averaged about 26 percent of GDP at market prices from 2008/09 to 2014/15. Data from the fiscal accounts indicate that public investment averaged about 6 percent of GDP, implying that private sector fixed investment averaged about 19 percent of GDP in this period. 15 P a g e

17 the resources available f or investment in human and physical capital. Countries which have experienced rapid economic growth and structural transformation, such as those in East Asia, have far lower TFRs and dependency ratios. Although the relationship between demographic and structural change is not uni-directional, it is difficult to believe that Uganda s retarded demographic transition has not contributed to its economy s failure to achieve more rapid structural transformation. 4. Conclusion Since the late 1990s, there have been some positive structural shifts in the economy. The share of services and construction in the economy has increased at the expense of agriculture, with the latter s share of GDP falling by 14 percentage points. There has also been a shift from self-employment to wage employment, although the former still accounted for nearly three quarters of the working population in 2012/13. Moreover, the shift from selfemployment to paid employment mainly comprises workers moving into informal paid employment. Formal sector employment, in both public and private sectors, only comprises about 20 percent of wage employment or about 5-6 percent of the total working population. More than 40 percent of the labour force, and two thirds of the agricultural labour force, are subsistence farmers. There was rapid growth of firms in the first decade of the century, but this growth was concentrated among informal and micro-enterprises. The number of employees of firms with a fixed business address rose from 440,000 in 2000/01 to nearly 1.1 million in 2010/11, but employment growth was fastest in firms with less than five employees, which are classified by UBOS as informal. Employment in formal sector firms, with five or more employees, increased by 7 percent per annum on average during the first decade of the century, but by 2010/11 it still only accounted for 3.3 percent of the working population. The larger formal sector firms, which are covered by the BOU s Private Sector Investment Survey, contributed an average 5 percent of GDP and 0.8 percent of total employment during The performance of the manufacturing sector has been mixed. The sector s share of GDP has stagnated, at between 8 and 9 percent, since the late 1990s. Manufacturing employment in firms with fixed businesses premises grew by less than 5 percent per annum between 2000/01 and 2010/11 and by the latter year, employment in formal sector manufacturing accounted for only 0.7 percent of the working population, slightly lower than in 2001/02. On the other hand, 16 P a g e

18 exports of processed primary commodities and manufactured products grew rapidly between 2002 and 2015, and increased their combined share of merchandise export earnings from 8 percent in 2002 to 31 percent in This growth was boosted by access to regional markets, especially that of South Sudan. The economy has not achieved strong growth in labour productivity. At the level of the whole economy, labour productivity increased from Shs 2.6 million per worker in 2002/03 to Shs 3.2 million in 2012/13, in constant 2009/10 prices; an average annual real increase of only 2 percent, which is mediocre for a developing economy. This reflects the fact that most of the shifts which have occurred in the composition of the workforce have involved workers moving from one low productivity sector to another, such as from self-employment in agriculture to self-employment in services or artisanal manufacturing or paid employment in informal microenterprises in the services or artisanal manufacturing sectors. Rapid labour productivity growth would require labour being absorbed in large numbers into high productivity formal sector firms, but that has not taken place. 17 P a g e

19 References Ajakaiye, Olu and John Page (2012), Industrialization and Economic Transformation in Africa: Introduction and Overview, Journal of African Economies 21, AERC Supplement 2, ppii3-ii18. Drew, Clive (2010), Past Successes and Failures in Supporting Agricultural Commercialization in Uganda, mimeo prepared for the World Bank. Fox, Louise and Obert Pimhidzai (2013), Different Dreams Same Bed: Collecting, Using and Interpreting Employment Statistics in sub-saharan Africa - the case of Uganda, Policy Research Working Paper 6436, World Bank, Washington DC. Fox, Louise, Cleary Haines, Jorge Huerta Munoz and Alun Thomas (2013), Africa s Got work to Do: Employment Prospects in the New Century Working Paper WP/13/201, International Monetary Fund. Gelb, Alan, Meyer Christian and Ramachandran Vijaya (2013), Does Poor Mean Cheap? A Comparative Look at Africa s Industrial Labour Cists, Working Paper 325, Centre for Global Development. Global IDP Database (2005), Profile of Internal Displacement: Uganda, Norwegian Refugee Council. Jedwab, Remi (2013), Urbanization without Structural Transformation: Evidence from consumption Cities in Africa mimeo, Department of Economics, George Washington University. Lewis, Arthur (1954). Economic Development with Unlimited Supplies of Labour, The Manchester School, pp Lim, Jerome (2013), Institutional and Structural Determinants of Investment Worldwide Policy Research Working Paper WP 6591, World Bank. Rodrik, Dani (2013), The Past, Present and Future of Economic Growth, Working Paper 1, Global Citizen Foundation Uganda Bureau of Statistics (2012), Statistical Abstract, Kampala. Uganda Bureau of Statistics (2014A), Uganda National Household Survey 2012/13, Kampala Uganda Bureau of Statistics (2014B), National Housing and Population Census, Provisional Results, Kampala Vries, Gaiten de, Marcel Timmer and Klaas de Vries (2013), Structural transformation in Africa: Static gains, dynamic losses, Groningen Growth and Development Centre Research Memorandum 136, University of Groningen Zhu, Xiaodong (2012), Understanding China s Growth: Past, Present and Future, The Journal of Economic Perspectives 26 (4), P a g e