71810 Ghana: An Analysis of Firm Productivity Regional Program for Enterprise Development (RPED) Africa Private Sector Group (AFTPS) REVISED June 2006 Francis Teal (Oxford University) James Habyarimana (Georgetown University) Papa Thiam (AFTPS, World Bank) Ginger Turner (DECVP, World Bank) 1 Table of Contents Chapter 1 Introduction and Motivation .......................................................................... 3 1.1 Introduction ...................................................................................................................................... 3 1.2 Growth and investment .................................................................................................................... 3 1.3 The rise of the informal sector ......................................................................................................... 6 1.4 The implications of the rise of the informal sector for poverty and incomes ................................... 7 Chapter 2 The Performance of Ghana’s Firms in International Perspective ..... 11 2.1 Dimensions of comparative firm performance ................................................................................11 2.2 Comparative Firm Productivity .......................................................................................................11 2.3 Exporting .........................................................................................................................................18 2.4 Investment .......................................................................................................................................25 Chapter 3 Labor Markets in Ghana ............................................................................... 31 3.1 Introduction .....................................................................................................................................31 3.2 The macro context for wages ..........................................................................................................31 3.3 Wages in Ghanaian manufacturing in constant domestic prices and in US$ ..................................33 3.4 Wages for the Skilled and Unskilled ...............................................................................................34 3.5 Wages, firm size, labor market flexibility and the institutional environment .................................37 3.5.1 Firms size and wages .............................................................................................................37 3.5.2 Trade unions ..........................................................................................................................39 3.5.3 Labor market flexibility .........................................................................................................40 Chapter 4 Firm Growth and the Investment Climate in Ghana: Key Objectives for an Investment Climate Assessment.............................................................................. 43 4.1 Dimensions of firm performance ....................................................................................................43 4.2 Firm Growth in the Panel ................................................................................................................43 4.3 Firm Growth in the Population ........................................................................................................45 4.4 Firm Performance and the Investment Climate: key issues............................................................46 4.4.1 Access to Credit .....................................................................................................................48 4.4.2 Access to and Cost of Domestic Raw Materials ....................................................................57 2 Chapter 1 Introduction and Motivation 1.1 Introduction The objective set by the Government of Ghana (GoG) is for Ghana to be a middle-income country by 2020. In broad terms this requires a per capita income of US$6000 (in current purchasing power parity (PPP) terms). In 2000 Ghana’s income was just under US$ (PPP) 2,000. Reaching the goal of becoming a middle-income country by 2020 implies therefore an average per capita growth rate between 2000 and 2020 of more than 5 per cent per annum. This is over twice the growth rate achieved during the 1990s. While the period since the 1980s has seen a sustained growth of the Ghanaian economy, the rate is still too low to meet the stated objectives of the GoG. The focus of this study is an analysis of firm productivity in Ghana, based on panel of firms surveyed between 1996 and 2002, as well as other information. The analysis focuses on identifying the drivers of productivity and the factors behind the increasing informalization and the lack of expansion of firms in the Ghanaian private sector. Based on this analysis, the objective is to identify key hypotheses about the investment climate, to be tested in the forthcoming Investment Climate Assessment for Ghana. This hypothesis testing will lead to the identification of priority areas of the investment climate that need to be reformed in order to achieve a higher rate of growth in the private sector and the economy as a whole. 1.2 Growth and investment It will be argued in this study that more rapid increases in income require an increase both in the rate of investment and, of equal importance, in the returns on that investment. A key element in meeting both those objectives is a shift to private sector investment in export-oriented activities. As recent data for Ghanaian growth shows the growth rate of aggregate investment has been far higher than that for consumption or exports. Much of this growth has been predominantly in the private sector (see Figure 1 below). 3 Figure 1.1 Composition of Aggregate Investment Investment Rates: Ghana 30 25 20 % GDP 15 10 5 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Year Ghana Gross priv. invest. as share of GDP (%) Ghana Gross public invest. as share of GDP (%) Ghana Gross dom. invest. as share of GDP (%) Table 1.1 provides a summary of the information available from macro and micro data covering the period from 1987/88 to 1998/99 - this is the maximum period for which we can match macro data with information from household surveys. Over this period GDP per capita grew by 1.8 per cent per annum and investment by 10.6 per cent per annum. On the basis of macro data, private consumption grew by 1.5 per cent per annum and on the basis of household survey data by 1.3 per cent per annum.1 Growth rates in investment have massively exceeded those in consumption. While a breakdown of private investment is not available it is likely that the growth in investment was dominated by the non-tradeable sector. As will be shown below private 1 In GSO (1995, Table 2.1 p.6) the figure for consumption per capita is given as Cedis 215,000, as reported in Table1.1. At the time of the analysis of the fourth round this figures was revised down to Cedis 183,000, a reduction of 15 per cent. This figure can only be obtained from the data, not from the report, which gives figures in terms of adult equivalents rather than per capita and uses 1998 prices. In GSO (2000, p.3) there is a warning that “the results reported here are not strictly comparable with the previous report�. In this study the original data is used. A more detailed use of the data can be found in Teal (2001). 4 Table 1.1 Ghana at a Glance: Growth 1987/88-1998/99: Household Survey and Macro Data GLSS1 GLSS2 GLSS3 GLSS4 1987/88 1988/89 1991/92 1998/99 1 HHEXP/Capita (‘000 cedis 1991/92 prices) (a) 198.3 187.5 215.0 2 HHEXP/AE (‘000 cedis 1998/99 prices) (b) 1130.8 1412.1 3 Index 1998/99=100 73.9 69.9 80.1 100 HHEXP/Capita Weights used for GLSS 4 4 Nominal (‘000 Cedis) 87.0 107.9 208.9 1,336.3 5 CPI 1998/99 prices 6.8 8.6 15.8 88.7 6 CPI 1991/92 prices 43.3 54.6 100 561.2 7 Real (‘000 Cedis 1998/99 prices) 1,283.2 1,249.1 1,326.8 1,336.3 8 Real (‘000 Cedis) 1991/92 prices) 202.7 197.4 209.6 235.0 9 HHEXP/Capita 86.5 85.7 90.5 100 Index 1998/99=100 10 GDP per Capita (‘000 Cedis 1998/99) 822 842 890 992 11 Investment per Capita (‘000 Cedis 1998/99) 69 81 94 222 12 Consumption per Capita (‘000 Cedis 1998/99) 694 700 723 820 13 Consumption per Capita Macro Index 84.6 85.3 88.2 100 1998/99=100 Sources: GLSS Surveys and World Development Indicators (2004). As noted in the text the aggregate expenditure data for the third round of the survey were revised at the time the fourth round was analysed. We use throughout this study the original data so that we can compare out results with those published in GSO (1995). (a) Household Expenditure per Capita (HHEXP/Capita) is taken from GSO (1995, Table 2.1 p.6). (b) Household Expenditure per Adult Equivalent (HHEXP/AE) is taken from GSO (2000, Appendix 1, p.35). (c) Income in the Principal Job is obtained from the employment part of the GLSS surveys. Note on consumption estimates: Deaton (2003) compares survey estimates of consumption per capita with those from the national accounts for some 127 countries. He finds that “consumption estimated from the surveys is typically lower than consumption from the national accounts; the average ratio is 0.860 with a standard error of 0.029, or 0.779 (0.072) when weighted by population. (India has particularly low ratios.) The exception is sub-Saharan Africa, where the average ratio of survey to national accounts consumption is unity in the unweighted and greater than unity in the weighted calculations.�(p. 7) It will be noted from Table 1.1 that the estimates of consumption per capita from the survey data are much higher than the macro estimates. It appears that the Ghana data is an outlier in how high are the survey estimates relative to those in the national accounts. 5 investment, where it can be measured accurately as in the manufacturing sector, has been very low. The key problem facing Ghana is not the rate of growth of investment but its composition. 1.3 The rise of the informal sector In parallel with the limited rise in consumption has been the rapid expansion of self-employment activities in Ghana relative to wage employment. Table 1.2 uses the GSO surveys over the period from 1987/88 to 1998/99 to show how the composition of the labor force has changed over that period. Over this decade there was virtually no change in wage employment. The total number of Table 1.2 Employment in Ghana: 1987/88-1998/99 1987/88 1988/89 1991/92 1998/99 National % 000s % 000s % 000s % 000s Wage Employees 17.3 1,121 18.1 1,215 15.4 1,143 13.2 1,166 Government 8 518 7.9 530 7.8 579 5.9 521 State Enterprise 1.9 123 2.3 154 1.2 89 0.6 53 Private 7.4 480 7.9 530 6.4 475 6.7 592 Self-employment 19.5 1,264 24.2 1,624 23.5 1,744 27.3 2,411 Unpaid Family 2.2 143 1.1 74 1.3 96 0.3 26 Agriculture 58.7 3,804 54.6 3,664 56.7 4,207 55.7 4,918 Unemployed 2.2 143 1.9 127 3.2 237 3.5 309 Total Labor Force 100 6,480 100 6,710 100 7,420 100 8,830 1987/88 1988/89 1991/92 1998/99 Urban % 000s % 000s % 000s % 000s Wage Employees 33.8 727 34.0 739 30.6 725 23.6 681 Government 13.9 299 14.3 311 15.2 360 10.0 289 State Enterprise 3.7 80 4.3 93 2.0 47 1.1 32 Private 16.2 348 15.4 335 13.4 318 12.5 360 Self-employment 36.3 781 42.1 915 42.6 1,010 48.1 1,389 Unpaid Family 4.1 88 1.8 39 2.0 47 1.7 49 Agriculture 21.0 452 17.7 384 16.7 395 18.7 540 Unemployed 4.8 103 4.5 97 8.2 193 7.9 228 Urban Labor Force 100 2,151 100 2,174 100 2,370 100 2,887 Sources: Author calculations from GSO Surveys. wage jobs in the economy has remained constant at about 1.2 million, while the total labor force has expanded from 6.5 to 8.8 million. As the top part of Table 1.2 shows while this expansion did lead to some rise in measured unemployment (from 143,000 to 309,000) this was very modest compared with the massive rise in non-agricultural self-employment, where employment came close to doubling from 1.3 million to 2.4 jobs. The bottom part of Table 1.2 shows the breakdown for areas classified as urban in the survey 6 data. As can be seen by 1998/99 nearly 60 per cent of these non-agricultural self-employment jobs were in urban areas. In 1987/88 in urban areas there were as many wage jobs as self- employed, by 1998/99 there were twice as many in self-employment as in wage employment. This dramatic rise in the relative importance of non-agricultural self-employment is only one aspect of the rise of the informal sector. A second is the decline in jobs in state enterprises. As the top part of Table 1.2 shows the number of jobs in state enterprises fell from 123,000 to 53,000 over the period. In contrast the number of jobs in the government part of the public sector was virtually unchanged. While there has been no contraction of the government part of the public sector; there has been a very substantial contraction of state firms. Such firms will have been much larger than the typical Ghanaian private enterprise. The third aspect of the rise of the informal sector in Ghana has been the very large shift within the manufacturing sector towards smaller firms (see Chapter 4 below). As will be shown in Chapter 3, small firms pay substantially less than larger ones, even after we control for both observable and unobservable skills of the workforce. The collapse in employment in publicly owned enterprises combined with the rise in relative importance of small firms in the total number of enterprises implies a fall in average wages across enterprises relative to what they would have been in the size composition had not shifted towards the small scale.2 In summary the rise in the informal sector has three elements.  the rise in non-agricultural self-employment jobs,  the decline in state run enterprise employment,  the rise of the small firm. The three elements have in common that they can be explained by the failure of private sector investment to expand fast enough to absorb labor into new relatively large enterprises. 1.4 The implications of the rise of the informal sector for poverty and incomes The failure of investment to create jobs has implications for how growth is impacting both on consumption and poverty. It is very hard to measure income for the self-employed (either rural or urban) so comparisons based on household per capita expenditures give a more accurate account 2 It is difficult to determine the impact of this size shift by examining data at a point in time. It is possible that the firm size shift will allow the entry of more efficient and dynamic firms. In this sense, it is possible to think of the lowering of wages as a transitory phenomenon that would be reversed in the future. The kind of data required to answer this question is firm census data at two different points in time.. 7 of how the different activities compare. Table 1.3 shows household expenditures per capita taken from the GLSS surveys where households are classified by the occupation of the household head. In making comparisons across these groups it is most useful to focus on the means of the logarithms of incomes as the mean level will be influenced by the small number of large incomes in the survey data. The Table also gives figures in US$ (using actual exchange rates not PPP). Table 1.3 Household Expenditure per Capita (Annual Measures) 1987/88 1988/89 1991/92 1998/99 Wage Employees (N) 797 896 991 1046 1998 Cedis 1,739,173 1,671,170 1,814,395 2,041,369 (1,511,205) (1,526,970) (1,627,723) (1,784,568 Logs (1998 cedis) 14.11 14.06 14.12 14.28 (0.70) (0.72) (0.77) (0.73) US $ 659 613 707 812 (561) (559) (649) (708) Farmers (N) 1,649 1,655 2,299 2,940 1998 Cedis 1,001,534 960,789 969,044 1,007,263 (814,842) (846,274) (798,623) (831,395 Logs (1998 cedis) 13.58 13.52 13.54 13.55 (0.68) (0.69) (0.70) (0.70) US $ 384 350 384 382 (310) (304) (315) (318) Self employed (N) 517 720 985 797 1998 Cedis 1,487,194 1,426,714 1,592,886 1,802,173 (1,275,218) (1,328,677) (1,409,819) (1,527,338) Logs (1998 cedis) 13.94 13.88 14.02 14.13 (0.72) (0.74) (0.72) (0.77) US $ 657 522 617 718 (484) (484) (556) (619) All (a) (N) 2,963 3,271 4,275 5,465 1998 Cedis 1,284,687 1,257,936 1,308,746 1,454,805 (1,172,096) (1,219,156 (1,246,675) (1,348,863) Logs (1998 cedis) 13.78 13.75 13.79 13.87 (0.73) (0.75) (0.77) (0.79) US $ 489 460 512 570 (440) (444) (492) (538) N is the number of households, the figures in ( ) parentheses are standard deviations. (a) These figures are the totals for the three categories identified, not for all households in the survey. Sources: GSO Surveys. Several important findings emerge from the data. Households headed by a farmer have levels of per capita consumption approximately half of those headed by a wage employee. In contrast the gap (if we focus on the means of the logs) between households headed by a wage employee and 8 ones in which non-agricultural self-employment is small - of the order of 15 per cent in 1998/99. We know from the breakdown of the labor force given in Table 1.2 that the major change in the composition of the labor force was the rise in non-agricultural self-employment. How could expenditures have risen given that wage employment has fallen relative to self-employment? The answer from Table 1.3, which is shown graphically in Figure 1.1, is that the growth rate of self- employment expenditures was actually higher than that for expenditures in households headed by a wage employee. In contrast, on average per capita expenditures fell for households headed by a farmer. Figure 2.2 Consumption Growth in Ghana Rate of Growth of Expenditure per Capita by Occupation of Household Head: 1987/88-1998/99 Figure 1.1 is based 20 on the data from Table 1.3. 15 The growth rates are derived from 10 the means of the Percent per logs of expenditure Decade where households 5 are classified by the occupation of 0 the household head. -5 Farmer Self-Employed Wage Employee Household Expenditure per Capita from the GLSS surveys. The key to understanding the growth pattern in Ghana has been the successful growth of the non-agricultural self-employed. We have argued that the decline in the relative importance of wage employment (from 17 to 13 per cent of the labor force) can be explained by the insufficient levels and composition of private investment. In contrast the growth of self-employment is consistent with a pattern of investment leading to increased demand in the non-traded sector, construction and services which have created many more opportunities for the small scale self-employed person to gain an income. It is also possible that this increased range of self-employed opportunities has led to an increase in the labor supply in the household. Table 1.3 and Figure 1.2 present the means of survey data. It is important to note that the variance of expenditures for all types of household is very large. As will be shown in Chapter 3 there is a very substantial overlap between incomes from self-employment and incomes from wage jobs. 9 What can account for this overlap? We argue in Chapter 3 it is accounted for by the number of low wage jobs on offer and reflects the increasing informalization of the economy. Is this is a sustainable pattern for future growth? Or does it represent a short-term transition to a higher growth path? The limitations in the growth process are inherent in its pattern. Over the decade as a whole the total increase in per capita consumption was on the order of 15.6 per cent. This modest rise was insufficient to prevent falls in farmers per capita consumption who are the poorest class of households. In contrast, figures from China show rises of 5 per cent per annum for per capita consumption. While a pattern of investment geared to non-traded and primarily urban based employment opportunities can sustain some rise in income it is wholly inadequate to meet the objectives set by the GoG for growth rates in excess of 5 per cent per annum. The key to an acceleration in growth rates must be an expansion of private sector investment in the tradeable sector. The rest of this report is organized as follows--chapter 2 describes the performance of Ghana’s manufacturing firms in international perspective. Chapter 3 examines the features and performance of the labor and credit markets. In chapter 4, we present evidence on the evolution of productivity, export status and employment growth. Chapter 5 looks the impact of infrastructure and other investment climate variables on firm performance. We conclude with a chapter of policy recommendations. We speculate on the ability of Ghana to compete in export sectors associated with information technology in this chapter. 10 Chapter 2 The Performance of Ghana’s Firms in International Perspective 2.1 Dimensions of comparative firm performance In this chapter the main findings from the firm-level analysis of Ghana’s manufacturing sector are set out and placed in a comparative context. Firm level performance in terms of productivity, exporting and investment are compared between Ghana and four other African countries - Kenya, Tanzania, Nigeria and South Africa. The data used for this analysis is drawn from panel studies so it is possible to show how growth rates of firms differ both across countries and by the size and sector of the firm. Before focusing on these dynamic aspects of firm performance this chapter begins by setting out in section 2.2 how the firms compare in terms of underlying productivity and size. As has been extensively documented in similar data sets of Africa’s manufacturing sector labor productivity and capital intensity differ very substantially across firms by size. Large firms are oversampled in Ghana and in the four comparator countries. As will be shown in the next chapter the expansion in the number of firms in Ghana has been almost entirely confined to small firms. So in understanding how firm performance will have changed over time it is necessary to focus on why firms differ so substantially in size and the relationship between size and productivity.3 In later sections, size will be shown to be strongly associated with exporting (section 2.3) and with investment (section 2.4). The implications of these findings for the factors that limit the growth of manufacturing firms in Africa will be considered. A detailed discussion of the factors determining the performance of firms in Ghana is presented in Chapter 4. 2.2 Comparative Firm Productivity Tables 2.1 provides the summary statistics for the differences across manufacturing firms taken from surveys in Ghana (1991-2002), Kenya (1992-1999), Nigeria (1998-2003), Tanzania (1992- 3 It is important to point out that the results presented here rely on a panel of firms starting in 1991 and following them until 2002. Given that the shift in the size distribution of firms has been on-going throughout this period, it is possible that the sample of firms for which we have data is not representative. In particular, it is possible that younger and more dynamic firms have emerged in the interim. 11 2000) and in South Africa (1997-1998).4 For this report, both the Ghana and the Nigerian data have been extended to incorporate more recent data. In the case of Ghana, comparative data is now available over a twelve-year period. As at the end of the period for which there is panel data a census was carried out of the industrial sector it will be possible in the next chapter to use preliminary information from the census to assess how the industrial structure in Ghana has Table 2.1 Comparative Firm characteristics by Country Mean Mean Mean Raw Mean Mean Mean Mean Mean Firm N Output per Capital per Materials Indirect Employme Percent Percent Age Employee, Employee, per Cost per nt Exporters Foreign $ $ Employee, Employee, Ownership $ $ Ghana 3294.5 1152.9 1436.6 237.5 23.8 17 19 18.49 1563 Kenya 7332 5218.7 3641 632.7 28.8 33 18 20.47 900 Nigeria 9604.6 3751.8 4491.8 888.9 33.1 8 20 22.17 631 Tanzania 3041.2 1571.8 1436.6 330.3 18.2 14 15 15.42 967 South 40945.6 15214.4 18769.7 2186.4 135.6 69 24 20.51 313 Africa Notes: The data for output, capital, raw material and indirect costs have been converted to 1991/92 US$ which is the initial period for the data from Ghana, Kenya and Tanzania. The initial period for Nigeria is 1998 and for South Africa it is 1997. For Ghana, Kenya and Tanzania the data was first converted in constant domestic prices and then converted to US$ using for Ghana the 1991 exchange rate of the cedi for the US$ and for Tanzania and Kenya the 1992 exchange rate was used. For Nigeria and South Africa only consumer prices were available to deflate the firm output figures. These were then converted to US$ for 1998 and 1997 respectively. An adjustment was then made to these US$ figures to convert them to 1992 US$ by using the US GDP deflator. The capital stock for all countries except South Africa is accumulated from the investment data. Sources: Ghana, Kenya and Tanzania - RPED World Bank surveys, extended by CSAE with funding from DFID and UNIDO. The Nigeria data comes from two surveys funded by UNIDO in 2001 and 2004. The South African survey took place in 1999 and covered large firms (50+ employees) in the Greater Johannesburg Metropolitan Area (GJMA). This survey was undertaken by the World Bank in conjunction with the Greater Johannesburg Metropolitan Council (GJMC). One year of recall data was asked for and thus the South African data covers the period 1997- 1998. Information about the South African survey, as well as the data and a report on the survey can be found on the Trade and Industrial Policy Strategies (TIPS) website (www.tips.org.za). changed since the early 1990s. In the case of Nigeria a follow-up to a survey carried out in 2001 was undertaken in 2004/05. Nigeria is a useful comparator to Ghana as a result of a shared colonial and post-colonial experience that has shaped the investment climate in both countries. In 4 These are the countries used by Rankin (2004) in his analysis of the determinants of exports from Africa. 12 addition, the 2004/5 survey in Nigeria collects a similar set of infrastructure variables that allows for an informative comparison. In Table 2.1 it is clear that there are very large differences across the country samples. As is explained in the notes to the Table the figures presented are in 1991/92 US$ and all US$ references in this study refer to 1991/92 prices unless otherwise specified. Output per employee at US$ 40,135 (exp(10.62) in South Africa is more than ten times higher than in Tanzania at US$ 3,041 (exp(8.02). Ghana’s labor productivity is only marginally higher than Tanzania at US$ 3,294. The data is summarized in Figure 2.1 and in looking across the countries it is clear that the two countries with particularly low labor productivity are Ghana and Tanzania. There is a discrete rise in the relative labor productivity in South Africa. One of the key differences across the country data presented in Table 2.1 is differences in firm size distribution across countries. For South Africa the average firm size is 136 employees (exp(4.91), while for both Ghana and Tanzania it is less than 25. How much of the very large differences in labor productivity across the countries shown in Figure 2.1 is a country effect and how much the result of size? Figure 2.1 Comparative data on labor productivity and capital intensity Labour Productivity and Capital Intensity in US$ (1991/92 prices) 40,000 30,000 20,000 10,000 0 GHANA KENYA NIGERIA SOUTH AFRICATANZANIA Median Output per Employee Median Capital per Employee Table 2.2 gives a breakdown of the firm characteristics by size. As is shown in the Table there is a monotonic rise in output per employee, capital per employee and raw materials per employee across the size range. Clearly in terms of factor intensity the choice of technology differs across 13 firms by their size. As firms are smallest in the Ghana and Tanzania samples, some of the differences observed in labor productivity may reflect this difference in the size composition of the samples. As has already been noted, these samples are not representative of the population of firms in the countries. As small firms are greatly under-represented in the sample the sample average will overstate the average levels of productivity in the population. Further if small firms are growing in importance as a share of the population of firms average productivity will be falling even if the productivity within size classes of firms is not. Table 2.2 Comparative Firm Characteristics by Firm Size Firm Mean Mean Mean Mean Mean Mean Mean Mean N Size Output Capital Raw Indirect Employ Percent Percent Firm Age Catego per per Materials Cost per ment Exporters Foreign ry Employe Employe per Employe Ownershi e, $ e, $ Employe e, $ p e, $ Large 14328.4 10614.8 6502.9 1525.4 206.4 51 44 22.9 1,181 Medium 6247.9 4230.2 2807.4 544.6 36.6 21 16 20.1 1,264 Small 2617.6 665.1 1248.9 181.3 6.75 5 4 15.6 1,929 Why does factor choice differ by the size of firms and what are the implications of the very large differences observed in the data - labor productivity in large firms is more than five times that in small ones? One possibility is that firms differ in size as technology is related to the sector in which they operate. The technology of textile and wood firms is such that they require a larger size than garment firms. A second possibility is that the technology used changes as firms grow larger so that the optimal factor choice differs by firm size. Another possibility is that firms differ in the factor prices they face. If large firms have access to cheaper capital and unions, or other factors that push up wages for larger firms, then the large firm will use a more capital-intensive technology due to these differences in factor prices. These possible reasons why factor choice differs over the size range are not mutually exclusive and all may be present to some degree. The implications of these differences observed across firms by size for policy differ depending on their source. At one limit - if the differences reflect different technologies across sectors - then 14 only changes in sectoral composition will influence differences in labor productivity and capital intensity. At the other limit if the differences reflect different labor and capital costs across the size spectrum then there are large gains to be made by reforms to the labor and capital markets. To assess the issue of the relative importance of the factors influencing the changes over size and to answer how much of the labor productivity difference reflect differences in capital intensity and how much differences in underlying total factor productivity (TFP) it is necessary to estimate a production function. In Table 2.3 four equations are presented to show from a descriptive viewpoint how important are differences in TFP, export orientation, foreign ownership, firm size and age for explaining differences in labor productivity. In Column [1] a gross output production function is presented and in Column [2] a value-added production function. As firm characteristics such as ownership, size, age and export orientation may affect both factor choice as well as TFP in Column [3] an equation explaining labor productivity but with no controls for inputs is reported and finally in Column [4] the factors correlated with firm size are shown by means of an equation which regresses the log of employment on all the factors that are identified as possible determinants of TFP in Column [1]. There has been extensive discussion as to whether the production process should be modeled by means of a gross output or value-added production function (Bernard and Jones, 1996). In Table 2.3 both are presented so that a comparison can be made. In both specifications the inclusions of the log of employment allows for the possibility of non-constant returns to scale. The omitted country is Ghana so the country dummies can be interpreted as the differences in total factor productivity (TFP) across countries that cannot be explained by the observable differences across the countries which in the 15 Table 2.3 The Production Function (1) (2) (3) (4) Ln (Value- Ln (Output/ Ln Output/ added/ Ln (Employment) Employee) Employee) Employee) Ln (Capital/ 0.045 0.257 Employee) (6.72)** (15.31)** Ln (Raw Materials/ 0.645 Employee (36.09)** Ln (Indirect Costs/ 0.171 Employee (14.33)** Ln (Employment) 0.012 0.066 (1.51) (2.63)** Kenya -0.060 0.212 0.670 -0.058 (2.02)* (2.55)* (7.58)** (0.52) Tanzania -0.129 -0.244 -0.068 -0.270 (5.28)** (3.05)** (0.79) (2.49)* Nigeria 0.048 0.399 1.090 0.479 (1.17) (3.57)** (9.13)** (3.57)** South Africa 0.356 1.732 2.074 0.961 (8.09)** (15.68)** (19.59)** (7.53)** Dummy =1 if Firm Exports 0.071 0.229 0.560 1.174 (2.70)** (3.31)** (7.43)** (12.11)** Dummy=1 if Any Foreign 0.058 0.328 0.655 0.988 Ownership (1.99)* (3.94)** (7.66)** (8.72)** Firm Age 0.001 0.001 0.004 0.019 (1.23) (0.25) (1.78) (6.23)** Foods 0.017 0.239 0.652 0.241 (0.49) (2.67)** (6.42)** (1.90) Textile -0.099 -0.472 -0.250 0.942 (2.75)** (3.93)** (2.18)* (4.81)** Garment 0.051 -0.244 -0.796 -0.754 (1.71) (2.71)** (7.67)** (6.69)** Wood -0.039 -0.573 -0.645 0.258 (0.96) (5.02)** (5.11)** (1.64) Furniture 0.025 -0.195 -0.537 0.048 (0.84) (2.24)* (6.22)** (0.45) Textile and Garment (South -0.088 -0.367 -0.800 0.050 Africa) (0.97) (2.42)* (6.60)** (0.16) Constant 2.169 5.172 8.178 2.420 (18.26)** (28.76)** (64.48)** (17.28)** Observations 4374 4211 4374 4374 R-squared 0.92 0.48 0.46 0.43 Robust t statistics in parentheses * significant at 5%; ** significant at 1% Time dummies are included in the regression but not reported equation include sectoral dummies, whether or not the firm exports, foreign ownership and age. 16 In Figure 2.2 the differences in TFP across countries implied by the gross output production function in Column [1] are presented. Figure 2.2 Comparative data on total factor productivity Total Factor Productivity Differences .4 .3 .2 .1 0 -.1 KENYA NIGERIA SOUTH AFRICA TANZANIA Note: Total Factor Productivity is Measured Relative to Ghana The equation implies that while Ghana’s labor productivity is low relative to Kenya, its TFP is higher. Indeed the only country that has a substantially higher TFP than Ghana is South Africa. The country dummies shown in Figure 2.2 are essentially those aspects of TFP that the observables cannot explain. How important are the observables? Both exports and the existence of foreign ownership raise TFP by about 6-7 per cent in the gross output production function and between 23-33 per cent in the value-added production function. The effects associated with exporting and foreign ownership are much more important determinants of labor productivity and firm size than they are of TFP. Table 2.3 Columns [3] and [4] imply that exporting and foreign ownership are related to a two-fold increase in labor productivity and a three fold rise in size. There is no evidence from the production function estimation of substantial increasing returns to scale. Indeed the coefficient estimates for the gross output production in Table 2.3 Column [1] imply constant returns to scale. In summary there is a common finding across the countries that exporting and ownership are far 17 more strongly related to the size of the enterprise than to its efficiency measured either by labor productivity or TFP and size is not related to efficiency via increasing returns to scale. In the next sub-section the links between exporting and size are investigated further. 2.3 Exporting The results above suggest that export status is an important correlate of productivity and firm size. It is clear from the last sub-section that exporting firms are larger and have higher total factor productivity than non-exporting firms. In Table 2.4 data for exporting is presented both by country and by firm size. The Table provides a breakdown of exporting both by export propensity, by the percentage exported if the firm enters the export market and by destination. The dimension of destination that the data can address is whether or not the firm exports within or outside of Africa. On average across all the country samples 22 per cent of firms export and, conditional on exporting, they export 37 per cent of their output. The net effect of these two aspects of exporting - entering the export market and the degree of specialization in exporting, is that on average 8 per cent of output is exported. More firms export within Africa than outside although a small fraction do both. Focusing on the breakdown by exports-destination, 16 per cent of the sample export within Africa and 12 per cent export outside of Africa. Firms which export outside of Africa are more specialized than those which export to Africa - 44 per cent of output exported compared with 17 per cent within Africa. Even with this higher degree of specialization for firms which export outside of Africa, it is the case that less than half of their output is exported. This result is a puzzle although one common to other countries as well. If fixed costs are an important reason why small firms do not enter the export market then once firms cross a size threshold which makes exporting profitable why do they not specialise more? One possible reason is that they face a downward sloping demand curve in the export market. Another and, given the very small scale of exporting, a more likely explanation in the African context is that the same goods are exported as are sold domestically and the factors that limit exporting are the same factors that limit the scale of the firm. Before investigating this issue further we examine the differences across countries. 18 Table 2.4 Exporting to Africa and Outside Africa: By Country Mean % Mean % Mean % Mean % Mean % Mean % Mean % Mean % Mean % firms output output firms output output firms output output exporting exported exported exporting exported exported exporting exported exported condition to Africa to Africa to Africa out of out of out of al on any condition Africa Africa Africa exports al on any condition exports al on any exports Ghana 19 55 10 9 21 2 13 62 8 Kenya 39 28 11 34 18 6 12 38 5 Nigeria 8 33 3 6 29 2 4 21 1 Tanzania 17 24 4 13 13 2 7 30 2 South 70 18 13 62 10 6 44 15 6 Africa Large 49 37 18 35 15 5 28 45 13 Medium 21 36 8 15 19 3 11 44 5 Small 5 39 2 4 27 1 3 34 1 All firms 22 37 8 16 17 3 12 44 5 Figures 2.3 provide a summary of the key findings as to which countries export and how their exports divide between exports to other African countries and to countries outside of Africa. Percentage of Firms Exporting 80 60 40 20 0 GHANA KENYA NIGERIA SOUTH AFRICATANZANIA To all Destinations To Outside of Africa Within Africa The averages hide very important differences across the countries. South African firms are far more likely to export than those from the other countries. Among those other countries Kenyan 19 firms appear the most export oriented and Nigerian firms the least. However if we focus on exporting outside of Africa, Ghanaian firms are as export oriented as those from Kenya and more so than either Tanzania or Nigeria. To investigate further which factors are important in determining exports it is necessary to formalize the possible determinants of these differences in exporting outcomes. In Table 2.5, two regressions estimating the likelihood to enter the export market are presented - the propensity to export and then, conditional on exporting, the percentage of output exported. In Table 2.6 a similar specification is used for the decision to enter the market for exports within and outside of Africa. The equation identifies five broad factors as possible determinants of exporting. The first is the efficiency with which the firm operates. This is done by allowing both labor productivity and its determinants to enter as explanatory variables for exporting. As a first step to allowing for the possible endogeneity of these factors, they are entered as one-period lags. Thus the equation provides information on whether productivity lagged one period is informative as a determinant of exporting. If productivity is the sole factor associated with determining exporting then the output measure should enter with a positive sign and the inputs negatively. The second factor determining exports that the equation allows for is firm size which is measured by the log of employment. The third factor is foreign ownership, the fourth firm age and the fifth are sector dummies which allow for the possibility that some factor associated with sector, but not captured by any of the other variables is an important determinant of exporting. Two equations are presented in Table 2.5, one for the propensity to export which is a probit in which the coefficients reported are the marginal effects, the second an OLS equation of the percentage of output exported in which the sample is restricted to those firms in the export market. While such a procedure sets up a selectivity issue the determinants of exporting and the percentage exported appear so different as to make it essential to analyze the two dimensions of exporting separately. In explaining whether or not a firm exports the characteristic of the firm that dominates is its size. Figure 2.4 shows this graphically. There are no small firms in the South African sample but in the other four countries for which small firms are sampled in none does the export rate among such firms exceed 10 per cent. The vast majority of small firms are not in the export market. Among 20 large firms in contrast the export rate ranges from 78 per cent in South Africa, to 71 per cent in Kenya, 52 per cent in Ghana, 39 per cent in Tanzania and to 10 per cent in Nigeria. Ghana is in the middle of a large range of export orientation even within Africa. Table 2.5 Exporting to All Destinations (1) (2) (3) (4) (5) (6) Propensity Percentage Propensity Percentage Propensity Percentage to Export Exporting to Export Exporting to Export Exporting (conditiona Outside of (conditiona to Africa (conditiona l on any Africa l on any l on any exports) exports exports to outside of Africa) Africa) Ln 0.058 3.017 0.027 4.968 0.025 -0.909 (Output/Employee)(t-1) (2.20)* (0.67) (1.75) (0.77) (1.57) (0.23) Ln 0.027 0.188 0.010 -0.735 0.019 -0.354 (Capital/Employee)(t-1) (2.74)** (0.13) (1.66) (0.37) (2.90)** (0.39) Ln (Raw -0.032 -3.968 -0.028 -6.814 0.005 1.958 Materials/Employee)(t- 1) (1.69) (1.41) (2.72)** (1.82) (0.41) (0.70) Ln (Indirect 0.004 -0.104 0.010 2.439 -0.005 -1.071 Costs/Employee)(t-1) (0.36) (0.05) (1.43) (0.89) (0.67) (0.78) Ln (Employment)(t-1) 0.069 -0.357 0.028 1.693 0.039 -1.572 (6.73)** (0.15) (4.09)** (0.53) (5.63)** (1.37) Dummy =1 if Any 0.020 0.791 0.011 4.477 -0.011 -0.608 Foreign Ownership (0.62) (0.16) (0.54) (0.66) (0.58) (0.24) Firm Age -0.001 -0.102 0.000 -0.383 -0.000 0.035 (0.78) (0.72) (0.33) (2.03)* (0.20) (0.44) Foods -0.044 19.209 0.085 12.505 -0.059 5.088 (1.28) (2.92)** (2.86)** (1.08) (2.85)** (1.20) Textile 0.022 11.828 0.087 -16.442 0.039 8.002 (0.42) (1.67) (2.23)* (1.07) (1.04) (1.52) Garment 0.095 22.326 0.215 5.403 0.023 10.481 (1.60) (3.60)** (4.35)** (0.55) (0.62) (1.86) Wood -0.101 16.376 0.061 -0.678 -0.060 3.399 (2.15)* (1.76) (1.31) (0.04) (1.99)* (0.56) Furniture -0.042 10.783 0.064 13.719 -0.067 -3.434 (1.23) (1.13) (2.11)* (1.15) (3.72)** (1.21) Textile and Garment 0.046 -5.192 -0.034 -6.475 0.099 -1.579 (South Africa) 21 (0.37) (0.90) (0.77) (0.94) (0.96) (0.43) Wood_Ghana 0.767 34.510 0.592 39.868 0.210 -3.324 (Interaction term) (5.44)** (2.81)** (4.62)** (2.16)* (2.41)* (0.42) Kenya 0.230 -3.847 0.013 7.698 0.204 -1.176 (4.17)** (0.50) (0.42) (0.50) (4.47)** (0.24) Tanzania 0.078 -14.706 0.017 -3.501 0.059 -9.091 (1.67) (1.90) (0.60) (0.26) (1.70) (1.83) Nigeria -0.119 -9.735 -0.032 -11.441 -0.083 -4.780 (2.85)** (1.13) (1.22) (0.87) (3.11)** (0.68) South Africa 0.199 -1.874 0.291 -10.330 0.101 -4.676 (2.95)** (0.23) (4.53)** (0.66) (2.18)* (0.96) Constant 42.260 29.387 38.236 (1.65) (0.75) (2.39)* Observations 2515 551 2515 312 2515 391 R-squared 0.40 0.47 0.18 Robust z statistics in parentheses * significant at 5%; ** significant at 1% Time dummies are included in the regression but not reported 22 Percentage of Firms Exporting to All Destinations Large GHANA Medium Small Large KENYA Medium Small Large NIGERIA Medium Small Large SOUTH AFRICA Medium Large TANZANIA Medium Small 0 20 40 60 80 23 Percentage of Firms Exporting to Africa and Outside of Africa Large GHANA Medium Small Large KENYA Medium Small Large NIGERIA Medium Small Large SOUTH AFRICA Medium Large TANZANIA Medium Small 0 20 40 60 80 To Africa To Outside of Africa 24 2.4 Investment A striking feature of the data is the importance of exporting for large firms although there are substantial differences across countries. In Ghana nearly forty per cent of large firms export outside of Africa. Such firms have, as has already been documented, much higher capital labor ratios than small firms. Major proponents of the export-led growth strategies posit larger markets and improved technologies that raise investment in exporting firms. In this sub-section the possible links from exporting, firm size, efficiency, firm ownership and age to investing are investigated. As with exporting a two stage approach is adopted. First the factors related to the decision to invest are examined then, conditional of any investment, the amount of investment is examined. As with exporting this sets up a potential selectivity problem with the amount of investment model. In Table 2.7, the data for investment by country is presented and in Table 2.8 a breakdown by firm size is then reported. At the bottom of Table 2.8 the overall average across the countries is given. Figure 2.3a Percentage of Firms Investing Large GHANA Medium Small Large KENYA Medium Small Large SOUTH AFRICA Medium Large TANZANIA Medium Small 0 20 40 60 80 On average less than half the firms in the sample carry out any investment in any year. As Figure 25 2.6 shows this average hides substantial differences across firm size. Among large firms, most carry out some investment each year. As Table 2.8 shows the investment to capital ratio varies relatively little by firm size. Figure 2.7 reports the data in graphical form and the pattern for Ghana is particularly clear in that conditional on investment there is a strong inverse relationship with size for the amount of investment undertaken relative to the size of the capital stock. Table 2.7 Firm Level Investment Mean % Mean Mean Mean Mean Growth Growth Rate N Mean Investm of firms Capital Investment Investment Profits Rate of of Investment, to Capit investing Output to Value- to Capital per Output Employment $ Ratio Ratio Added Stock Employee (% pa) Ratio Ratio (lagged one period) Ghana 43.25 1.15 0.06 0.04 1,353 -0.06 -0.04 1,082 3,866 0.10 Kenya 55.03 1.65 0.10 0.05 3,837 1.55 -2.80 318 13,767 0.09 Nigeria 47.22 1.26 0.09 0.10 3,492 -0.29 -1.83 396 16,647 0.20 Tanzania 33.33 1.33 0.05 0.03 2,075 1.37 -0.51 372 1,620 0.08 South 83.45 0.71 0.10 0.10 11,143 -12.6 -3.63 139 135,944 0.12 Africa Large 68.21 1.58 0.11 0.07 5,203 -0.87 -1.70 670 98,715 0.10 Medium 42.17 1.44 0.08 0.05 2,542 -2.36 -1.81 709 6,634 0.11 Small 33.84 0.84 0.04 0.05 1,184 1.43 0.07 928 265 0.14 All firms 46.38 1.24 0.07 0.05 2,769 -0.40 -1.02 2307 8185 0.12 Figure 2.3b Investment to Capital Ratio: Conditional on Investing Large GHANA Medium Small Large KENYA Medium Small Large SOUTH AFRICA Medium Large TANZANIA Medium Small 0 .05 .1 .15 In order to pursue further the factors that influence both the decision to carry out any investment, a probit on the investment decision is presented in Table 2.9. The factors identified as possible determinants of the decision to invest are efficiency, modeled in a similar manner to that already 26 used for exporting, firm size, ownership and age and whether or not the firm exports. As will be documented in the next chapter by far the most common problem identified by the firm’s managers is access to credit. To investigate the salience of this issue a measure of the profit available to the firm is also included. Table 2.9 presents the results for the probit estimation of the decision to invest and Table 2.10 the regression estimation for the size of investment conditional on any investment being undertaken. We follow the same procedure as with exports in lagging the terms capturing the possible effect of efficiency on exporting. From Table 2.9 Column [1], there appears strong evidence that the efficiency with which a firm operates does influence whether or not it invests next period. As with exporting, firm size continues to be a highly significant determinant of investing even when we control for efficiency. There is evidence that older firms are less likely to invest. However, neither lagged export status, nor foreign ownership nor a measure of financial constraints - real profits per employee - appear to affect the decision to invest. One possible reason that we do not find an effect of financial constraints on the propensity to invest is that it is not possible to identify both an efficiency and a real profit effect - they are highly collinear. Table 2.9 Column [2] shows that if the efficiency terms are dropped the real profit term becomes significant. As it stands this cannot be regarded as strongly indicating that the true effect is working through profits rather than Table 2.9 The Decision to Invest (Probit Dummy=1 if Firm Invests) (1) (2) (3) Ln (Output/Employee)(t-1) 0.091 (2.26)* Ln (Capital/Employee)(t-1) -0.035 (3.58)** Ln (Raw Materials/Employee)(t-1) -0.030 (1.11) Ln (Indirect Costs/Employee)(t-1) 0.029 (1.91) Ln (Employment)(t-1) 0.075 0.079 (6.00)** (6.70)** (Real Profits/Employee)(t-1) 0.059 0.360 0.483 (0.31) (2.12)* (2.60)** Dummy = 1 if Export(t-1) 0.056 0.059 0.144 (1.53) (1.61) (3.93)** 27 Dummy = 1 if Any Foreign 0.059 0.085 0.162 Ownership (1.42) (1.98)* (3.64)** Firm Age -0.012 -0.013 -0.010 (3.85)** (3.93)** (2.90)** (Firm Age)2 0.000 0.000 0.000 (2.83)** (3.07)** (2.42)* Foods -0.016 0.004 0.016 (0.36) (0.09) (0.34) Textile -0.057 -0.090 -0.017 (0.75) (1.19) (0.23) Garment -0.059 -0.094 -0.150 (1.29) (2.23)* (3.54)** Wood -0.129 -0.149 -0.127 (2.01)* (2.32)* (1.94) Furniture 0.059 0.024 0.037 (1.31) (0.53) (0.82) Textile and Garment (South Africa) -0.227 -0.241 -0.223 (1.16) (1.26) (1.13) Kenya 0.132 0.146 0.153 (2.70)** (3.12)** (3.21)** Tanzania -0.091 -0.099 -0.118 (2.14)* (2.39)* (2.88)** Nigeria -0.012 0.030 0.065 (0.23) (0.60) (1.27) South Africa 0.160 0.240 0.305 (2.07)* (3.30)** (4.28)** Observations 2153 2153 2153 Robust z statistics in parentheses * significant at 5%; ** significant at 1% Time dummies are included in the regression but not reported. 28 Table 2.10 The amount of Investment (Conditional on any investment): Ln(Investment) (1) (2) (3) (4) Ln (Output/Employee)(t-1) 0.398 0.490 (2.61)** (1.89) Ln (Capital/Employee)(t-1) 0.391 -0.259 (8.59)** (1.59) Ln (Raw Materials/Employee)(t-1) -0.136 -0.253 (1.20) (1.45) Ln (Indirect Costs/Employee)(t-1) 0.124 -0.047 (1.91) (0.46) Ln (Employment)(t-1) 1.094 1.039 1.371 1.123 (21.72)** (3.55)** (24.53)** (4.88)** (Real Profits/Employee)(t-1) 0.391 -1.097 2.237 0.208 (0.87) (0.73) (3.18)** (0.16) Dummy = 1 if Export(t-1) 0.356 0.082 0.632 0.117 (2.64)** (0.33) (4.29)** (0.48) Dummy = 1 if Any Foreign -0.097 0.182 Ownership (0.60) (1.01) Firm Age -0.033 -0.020 -0.036 -0.027 (2.94)** (0.52) (2.51)* (0.78) (Firm Age)2 0.000 0.001 (2.49)* (2.26)* Foods 0.127 0.473 (0.70) (2.25)* Textile 0.139 0.025 (0.70) (0.10) Garment -0.127 -0.680 (0.60) (2.84)** Wood 0.230 -0.128 (1.06) (0.55) Furniture -0.259 -0.722 (1.45) (3.24)** Textile and Garment (South 0.067 -0.382 Africa) (0.34) (1.25) Kenya 0.339 0.798 (2.19)* (4.35)** Tanzania -0.200 -0.152 (0.89) (0.55) Nigeria 0.631 1.171 (2.57)* (4.42)** South Africa 0.652 1.423 (2.48)* (4.63)** Constant -1.255 5.936 4.293 5.566 (1.35) (2.19)* (4.45)** (5.04)** Observations 981 981 981 981 R-squared 0.80 0.13 0.74 0.11 Number of firm 571 571 OLS FE OLS FE Robust t statistics in parentheses* significant at 5%; ** significant at 1% Time dummies are included in the regression but not reported. 29 efficiency. If it were we would expect that efficiency would not remove the effect of profits as much as it does. In Table 2.9 Column [3] we drop size and now both exporting and foreign ownership come in highly significantly. We interpret this as showing that if there is an effect of export status or foreign ownership, it operates by their being associated with larger firms. In Table 2.10 the conditional export equation is run using the same specification as the probit equation. In column [1] the most general specification is presented and in Column [3] one which excludes the efficiency terms. Columns [2] and [4] present the fixed effects estimates for the first and third specification respectively. It will be noted that the fixed effect result in Column [2] significantly changes the interpretation of the cross-section results reported in Column [1]. The results in Column [2] are consistent with efficiency having a significant effect on the amount of investment undertaken. When we control for fixed effects, there is no evidence that the financial term - real profits per employee - matters for investment. The close to unity term on the log of employment in columns [1] and [2] implies that investment per employee is not a function of firm size. In summary while we have evidence that the efficiency with which a firm operates affects both its decision on whether to invest and its decision on how much to invest the evidence for a finance effect is very limited. As will be shown in chapter 4 credit constraints feature very prominently in the problems identified by firms so we need to consider how these two aspects of the data are to be reconciled. 30 Chapter 3 Labor Markets in Ghana 3.1 Introduction In this chapter labor markets will be reviewed to assess their role in enabling both the rate of investment and the returns on investment to be increased. We begin by setting out the evidence from the sample for how real wages have been changing over the decade and ask how far these changes can be related to skills. We will turn to the costs of capital to the firm in the next chapter. We wish to link both labor and capital market issues to understanding the sources of international competitiveness of the economy. For labor markets we need to answer some basic questions as to how wages have been changing in the Ghana economy over the last decade. How high are wages in Ghana relative to other countries and relative to the productivity with which firms in Ghana operate? How important are skills in explaining wage differences and how have the returns to skills been changing over time? We need to do this in a context which allows for how the wages of young workers have been changing. Many of those finding employment in the manufacturing sector for the first time start as apprentices. In many small firms most of the employees are apprentices particularly in the garment, furniture and metal working sectors. How does the pay of these apprentices compare with those of similar education and age working as production workers? We set out to use the data from the sample to answer these questions. 3.2 The macro context for wages Over the period from 1992 to 2003 (the period for which we have wage data) there have been very substantial changes in both the price level and the exchange rate. We begin by showing how the data from the firms for the change in wages relates to the changes in the price level. In the top part of Figure 3.1 we plot the path of nominal wages and the price level in the context of the implied rate of inflation. The inflation rate has been both high and highly variable. It went from a low of 10 per cent per annum at the beginning of the 1990s to an annual rate of nearly 50 per cent in the mid 1990s, then falling back before rising again. In the Figure we see that the growth of nominal wages has closely followed the price level. There was a fall in real wages in the mid 1990s and over the period after 2000 nominal wages rose above the price level implying a modest real wage gain. In parallel with the rate of inflation have been equally large changes in the nominal exchange 31 Figure 3.1 Wages and Prices 8 .5 7 .4 Prices .3 Inflation 6 Nominal wages Rate .2 5 .1 4 1990 1995 2000 2005 Year Log Nominal wage Log price Level Inflation Rate Exchange Rates and Prices .24 8000 .22 6000 Prices .2 Real Nominal Exchange 4000 Exchange .18 Rate Rate 2000 .16 .14 0 1990 1995 2000 2005 Year Nominal Exchange Rate Price Level Real Exchange Rate rate. The bottom part of Figure 3.1 shows the path of the nominal exchange rate and the price level. The real exchange rate (RER) in the Figure is defined as UrbanCPI RER  Xrate*US Prices Where Xrate is the nominal exchange rate and US prices are the unit value of US exports. The index is defined so that a fall in the RER is a real devaluation. It is apparent from the Figure that over this period there was a particularly dramatic real and nominal devaluation when the nominal 32 exchange rate went from 2314 Cedis per US$ in 1998 to 5455 in 2000. This massive nominal devaluation effected a real devaluation of 60 per cent. Just as with the rate of inflation there have been massive changes in the RER over this period. These changes in the real exchange rate have implications for the competitiveness of labor in Ghana. Changes in real wages in domestic prices can differ substantially from real wages in US$. As we are focusing on the linkage from investment to exporting it is the changes in the US$ figures that matter for the exporter. However the worker cares about the real wage in domestic prices. In the next sub-section we will examine change in wages both in real terms (ie nominal wages deflated by the price level) and in US$ terms (ie deflated by the nominal exchange rate). 3.3 Wages in Ghanaian manufacturing in constant domestic prices and in US$ How have wages changed over this period? In Figure 3.2 we use the data from the sample of workers carried out at the same time as the firms were surveyed to show how real wages have changed over the period. Figure 3.2 Real Wages and Wages in US$ 120 70 In the regression from which the figures are drawn we 110 control for all time invariant 60 aspects of the firm. This is 100 done in order to control for Real Wages 50 in Wages US$ changes in the sample 90 composition of firms, which 40 has changed substantially 80 over the 1991-2002 period. This way of proceeding 70 30 allows us to focus on how 1990 1995 2000 2005 Year wages would have changed Real Wages (Index 1992=100) Wages in US$ for the “typical� worker in the firms in the survey. The striking feature of this relatively long run of data is that while real wages have shown no trend change, they were higher in real domestic price terms at the end of the period than at the beginning. The right hand side axis for Figure 3.2 shows the wage data for an unskilled 33 production worker in terms of US$. The effects of the devaluation which occurred in the period from the 1998 to 2000 when the nominal exchange rate fell from 2314 cedis per US$ to 5455 are apparent in the data. This very substantial real devaluation meant that while wages rose in real terms in domestic prices, they fell in terms of US$. In 1992 this wage was nearly US$60 per month, by 2003 it had fallen to US$ 45, a decline of 25 per cent in nominal dollar terms. In the next section we show how wages differ by skills and note the differences between unskilled production workers and apprentices. 3.4 Wages for the Skilled and Unskilled How are skills rewarded in the Ghanaian labor market? We approach the issue of skilled labor in two ways. We look first at the wage differences between the three major types of labor in Ghanaian manufacturing firms - apprentices, unskilled (production, technical and masters) and skilled in Figure 3.3. We focus on apprentices as they are an important part of the labor force for small firms and they are young workers so they represent very much the conditions facing new entrants to the labor force. Figure 3.3 Wages in US$ per Month by Skill 110 100 90 80 70 60 50 40 30 20 10 0 199220002003 19922000 199220002003 199220002003 All Apprentice Skilled Unskilled We then look at the returns to education measured by the increment in wage which accrues to each additional year of education. Figure 3.3 gives a breakdown by type of labor for three years - 1992, 2000 and 2003 (wage data for apprentices is not available for 2003). The very large difference between types of labor is apparent from the figure. Skilled workers earn approximately 34 twice as much as unskilled workers across all the years shown in Figure 3.3. While Figure 3.3 shows a breakdown by the skill category of the worker in Figure 3.4 we show the returns to education defined as the percentage increase in earnings which accrue to an additional year of education. The increase in earnings associated with an additional year of schooling for workers with less than primary schooling is between 1-2%. Workers that have Figure 3.4 Rates of Return to Education .2 .15 Rate of Return .1 .05 0 0 5 10 15 20 Years of Education The shape of the earnings function is convex. This implies that the returns to education rise with its level. The return shown in the Figure is the Mincerian return and does not reflect the private or social return. completed 10 years of schooling enjoy a 5% increase in earnings for every additional year of schooling. In Figure 3.5 we present the data in terms of the wage that workers of average age and tenure would receive. In the sample the average age of workers is 33 years and tenure length is 7 years. It needs to be noted that earnings rise rapidly with age so Figure 3.5 is an average over the life- time earnings of an individual worker. 35 Figure 3.5 Wages in US$ per Month by Years of Education 110 100 90 80 70 60 50 40 30 20 10 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Note: The wages are calculated for an average worker over the period from 1992 to 2003 with the average characteristics of the sample for gender,age and tenure. Both Figures 3.4 and 3.5 show that by far the largest increases in wages accrue to those with higher levels of education. In the sample, the average years of education are 9.5. We see from the figures that at this level of education the rate of return is just under 5 per cent and the wage for the worker with average characteristics is just under US$40 per month. This is less than US$10 more than those with less than primary education. In contrast, those with very high levels of education have wages more than twice as much. The average return on education in Ghana reported in Söderbom, Teal and Wambugu (2005, Table 2) is 8.3 per cent which is high but lower than the average in Kenya of 10.4 per cent. The extent of the convexity in the earnings function implies that this average is misleading for those with average, or below average, levels of education. If skills are measured by the occupational classification of the worker, skilled workers earn about twice those of unskilled. If we use the years of education as the measure of skill - as in Figure 3.5 - the gap is wider. The earnings of a worker with a degree are about three times those of workers with primary education. Linking the returns to skills directly to productivity is difficult. However we showed in Chapter 2 above (Figure 2.2) that, by one measure, Ghanaian firms had a higher level of total factor productivity than Kenyan ones. There is little if any evidence that the supply of skills is not more than adequate for the sectors in which labor intensive exports would be located (see Chapter 5). 36 3.5 Wages, firm size, labor market flexibility and the institutional environment While wages rise with observed skills they also rise with the size of the firm as we now show. There has been extensive discussion as to differences within African economies between the formal and informal sectors. In this section we show how important the distinction between formal and informal firms is for the wages of their workers. 3.5.1 Firms size and wages To show how size premiums translate into effects on wages across firms, Figure 3.6 shows how wages vary across firms of different size. (The Appendix gives the regressions underlying the Figure). In the top part of Figure 3.6 we show for those workers with the average amount of human capital - measured by their age, education and tenure - how wages in US$ change as the worker moves up the size distribution of firms. Figure 3.6a and 3.6b: Wages in the formal and informal sectors Wages in US$ per Month by Firm Size 60 55 50 45 40 35 30 25 20 15 10 5 0 5 10 20 50 100 150 200 Note: The wages are calculated for an average worker over the period from 1992 to 2003 with the average amount of human capital measured by gender,age, education and tenure. 37 Wages in US$ per Month for Production Workers by Firm Size 70 65 60 55 50 45 40 35 30 25 20 15 10 5 0 5 10 20 50 100 150 200 Note: The wages are calculated for an average worker over the period from 1992 to 2003 with the average amount of human capital measured by gender,age, education and tenure. There is no hard and fast definition of what is meant by an “informal� relative to a “formal� firm. What the top part of Figure 3.6 shows is that there is a continuous rise in wages as the size of the firm increases. It needs to be stressed that these results control for the observed skills of the workers in the firms, so it is not the case that the rise reflects the fact - which is a fact - that larger firms tend to employ more educated workers. In round numbers wages double from US$25 per month to US$50 per month as you move from firms with 5 employees to firms with 150. It is also apparent from the Figure that the rise is most rapid over the size range from 5 to 50. If the sample is not so restricted then any size effect may reflect the presence in larger firms of certain types of more highly paid workers whose skills are not adequately captured by the human capital controls. We get around this potential interpretation by restricting the analysis to production workers (shown in the bottom part of Figure 3.6). It is still the case that wages vary substantially by firm size. The implication is that this is not explained either by their measured human capital or by the fact that there are unobserved skills correlated with occupation. Further these effects are observed across a wide range of firm size so it is misleading to classify firms into the “informal� and the “formal�. There is a size spectrum across which wages rise continuously. The wages-firm size relationship suggests two plausible explanations.5 On the one hand, given 5 Other explanations such as compensating wage differentials (Lester 1967, Scherer 1969) do not appear plausible given anecdotal evidence suggests that jobs in large firms are highly desirable and these firms are generally located in easily accessible parts of Accra. We don’t believe that better matching possibilities in 38 the firm size-export status relationship in chapter 2, it is likely that larger firms command rents that can be shared between owners and workers (Albaek et. al. (1998)). The sharing can be driven by a number of factors: firstly due to sharing norms between workers and owners (Thaler, 1995). Secondly due to increased bargaining power of workers given that capital and training intensities are higher among large firms (Black et. al 1999). Finally sharing can be induced as a deterrent to unionization. On the other hand it is possible that the wage-firm size relationship is due to efficiency wage considerations. Monitoring within larger firms using higher levels of capital intensity is generally costly. In order to counteract incentives to shirk, firms pay higher than reservation wages to workers (Shapiro and Stiglitz, 1984). Finally it is possible to push the analysis further for a sub-set of the workers in the firm surveys who have been observed continuously for several years in Ghana to ask if there are time-invariant unobserved skills that can explain the size premium. Söderbom, Teal and Wambugu (2005) show that with fixed effect at the individual worker level the size effect on wages remains large and significant. Their paper also provides a comparison with Kenya where a similar effect is observed. 3.5.2 Trade Unions Another potentially important distinction between the formal and informal sector is whether or not the workforce is unionized. Figure 3.7 shows the wages, again in US$ terms, for workers in unionized and non-unionized firms; first with no controls for human capital and then with controls. large firms as suggested by Kremer and Katz (1996) explains the variation that we observe even though our data cannot test the plausibility of this channel. 39 Figure 3.7 Wages in US$ per Month by Union Status 80 70 60 50 40 30 20 10 0 Non-Unionised FirmUnion Firm Non-Unionised FirmUnion Firm No controls for Skills Controls for Skills Note: The wages are calculated for an average worker over the period from 1992 to 2003. Controls for skills include gender,age, education and tenure. If we do not control for differences in human capital then there is a very large difference in wages between unionized and non-unionized firms with the former paying wages three times higher, US$ 75 as against US$25. However much of this differential disappears once we control for human capital although a substantial differential remains of some 60 per cent. While controlling for skills dramatically reduces the union effect, the remaining union premia are still very large by international standards. Indeed, the premia for Ghana are as high, or higher, than the average union effects found for South Africa. 3.5.3 Labor market flexibility In this section, we discuss two broad measures of labor market flexibility: firstly, we investigate the extent to which the labor market for manufacturing workers responds to changes in wages or labor supply and secondly, we investigate the effect of employment laws on the costs of hiring and firing. Kingdon, Sandefur and Teal (2004) identify three possible dimensions to the notion that labor markets may be inflexible. The first is that real wages may not adjust over time to excess supply of labor or to macroeconomic shocks. As stressed by Horton, et al (1994), the need for downward flexibility of real wages to achieve full employment in response to budget cuts and other demand reductions was seen as a crucial feature of structural adjustment programs. The second sense in 40 which labor markets may be inflexible is that, even without macroeconomic shocks, wages are unresponsive to high levels of unemployment. The third sense of the term is that there is a substantial differential across sectors distinguished either by whether or not the sector is unionized, or whether it is subject to minimum wage laws or whether the firms are simply large. These various meanings of the term inflexibility all have in common the notion that wages do not, for some reason, adjust to a market-clearing rate. However they differ both in the possible mechanisms by which wages may be inflexible and in their policy implications. If wages do not respond to unemployment then labor markets will not clear but there is no reason to think large firms will be disadvantaged by being made to pay higher wages than smaller ones. Kingdon, Sandefur and Teal (2004) examine a large range of evidence that suggests that labor markets in sub-Saharan Africa are flexible in the first two senses of the term but inflexible in the last. The evidence for Ghana is wholly consistent with their general conclusion. To understand the potential impact of employment regulations on the cost of hiring and firing, we use data from the World Bank’s Doing Business Survey that assesses the extent to which the regulatory framework affects the behavior of employers.6 The table below shows the averages of labor market rigidity indicators for Ghana, the four comparator countries and regions. 6 The difficulty of hiring index measures (i) whether term contracts can be used only for temporary tasks; (ii) the maximum duration of term contracts; and (iii) the ratio of the mandated minimum wage (or apprentice wage, if available) to the average value added per worker. A country is assigned a score of 1 if term contracts can be used only for temporary tasks, and a score of 0 if they can be used for any task. A score of 1 is assigned if the maximum duration of term contracts is 3 years or less; 0.5 if it is between 3 and 5 years; and 0 if term contracts can last more than 5 years. Finally, a score of 1 is assigned if the ratio of the minimum wage to the average value added per worker is higher than 0.75; 0.67 for a ratio between 0.50 and 0.75; 0.33 for a ratio between 0.25 and 0.50; and 0 for a ratio less than 0.25. The rigidity of hours index has 5 components: (i) whether night work is unrestricted; (ii) whether weekend work is allowed; (iii) whether the workweek can consist of 5.5 days; (iv) whether the workday can extend to 12 hours or more (including overtime); and (v) whether the annual paid vacation days are 21 or fewer. For each of these questions, if the answer is no, the country is assigned a score of 1; otherwise a score of 0 is assigned. The difficulty of firing index has 8 components: (i) whether redundancy is not considered fair grounds for dismissal; (ii) whether the employer needs to notify the labor union or the labor ministry to fire 1 redundant worker; (iii) whether the employer needs to notify the labor union or the labor ministry for group dismissals; (iv) whether the employer needs approval from the labor union or the labor ministry for firing 1 redundant worker; (v) whether the employer needs approval from the labor union or the labor ministry for group dismissals; (vi) whether the law mandates training or replacement before dismissal; (vii) whether priority rules apply for dismissals; and (viii) whether priority rules apply for reemployment. For each question, if the answer is yes, a score of 1 is assigned (for questions i and iv, a score of 2); otherwise a score of 0 is given. Questions (i) and (iv), as the most restrictive regulations, have double weight in the construction of the index. 41 Region or Country Difficulty of Rigidity of Difficulty of Rigidity of Hiring cost Firing costs Hiring Index Hours Index Firing Index Employment (% of salary) (weeks of Index wages) Ghana 11 40 50 34 12.5 24.9 Kenya 33 20 30 28 5 47 Nigeria 33 60 20 38 7.5 4 South Africa 56 40 60 52 2.6 37.5 Tanzania 67 80 60 69 16 38.4 Sub-Saharan Africa 48.1 63.2 47.8 53.1 11.8 53.4 East Asia & Pacific 26 29.6 23 26.2 8.8 44.2 Europe & Central 34.5 56.9 41.5 44.3 29.6 32.8 Asia Latin America & 40.5 50.9 29.5 40.3 15.9 62.9 Caribbean Middle East & 30.8 55 35 40.2 15.9 62.4 North Africa OECD: High 30.1 49.6 27.4 35.8 20.7 35.1 income South Asia 41.9 35 42.5 39.9 5.1 75 Relative to the comparator countries and the Sub-Saharan region, Ghana performs very well on all labor market rigidity dimensions. Only with the difficulty of firing, does Ghana perform worse than the Sub-Saharan average. The central problem that confronts policy in the labor market is the linkage from wages to firm size. Larger firms pay substantially more than smaller ones and this differential does not reflect observable skill differences. This linkage from firm size to wages ensures that larger firms are much more capital intensive than smaller ones (Chapter 2 Table 2.2) and thus that firms that can enter the export market have factor proportions inconsistent with exports being a source of substantial profits. 42 Chapter 4 Firm Growth and the Investment Climate in Ghana: Key Objectives for an Investment Climate Assessment 4.1 Dimensions of firm performance In chapter 2 the performance of Ghana’s manufacturing sector was placed in a comparative context. In this chapter we show how a range of the characteristics of firms within Ghana are of importance for understanding the limited success shown for investment and exporting. We begin by exploiting the panel dimension of the data to show how firms have changed over the twelve year period for which there is panel data. This evidence from the panel will then be linked to the data from the industrial census in section 4.3. In sections 4.4 and 4.5 we consider how two dimensions of the investment climate - the major problems which firm owners and managers identify as their principal problems and infrastructure - link to firm performance. 4.2 Firm Growth in the Panel The data presented in this sub-section is based on a panel survey of manufacturing firms carried out in Ghana from 1992 to 2003. In interpreting the results it is important to understand that the sample is not a random sample so the sample characteristics cannot be taken as reporting what has been happening in the population of all firms. The sample is also not a balanced panel and firms have exited from the sample and new firms have been added. The procedure we adopt is to report the results where we allow for all the aspects of the firm which have not changed over the period. The results tell us about changes within the sample over time. In Figure 4.1 the pattern of firm growth for the sample is shown where the measure of firm size is the number of employees. Figure 4.1 Firm Growth as measured by Employment Changes in Firm Employment 1992-2002 10 5 Percentage 0 -5 -10 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Changes in firm employment are the changes inthe log of employment for the sample with firm fixed effects NB: All years are relative to 1991 43 The percentage changes shown are all relative to 1991. The pattern shown by the data is of rises in the period 1992 to 1995 relative to the initial period 1991and then declines particularly in the period 2001-2002 after the major real devaluation which occurred after 1998 (see chapter 1). In Figure 4.2 we show changes in labor productivity over the twelve year period again based on a regression holding all other aspects of the firms constant. Over the whole period there is no change in the labor productivity of firms in the sample. This finding implies that the real devaluation led, at least in the short run, to a fall in output. Figure 4.2 Changes in Labor Productivity 1992-2002 30 20 Percentage 10 0 -10 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Changes in labor productivity are the changes in the log of output per employee with firm fixed effects NB: All years are relative to 1991 In looking at Figures 4.1 and 4.2, three points stand out. The first is the lack of any long term growth in labor productivity, the second is that, within the sample, firms seem to be getting smaller when size is measured by the number of employees and thirdly the major devaluation after 1998 has not on average had a positive effect on firm performance. It is not the case that this sample can necessarily be taken as a good guide to what was happening to all firms in the economy. We cannot know from our sample how many firms may have entered or exited from the sector over the period. This pattern of entry and exit will have implications for how the size composition of the population of firms will have been changing. All these aspects of the industrial structure have important implications for what will have been happening to wages and firm size. While firms in the sample may have been getting smaller it is possible that firms outside the sample have been growing and the picture from the panel is misleading. It is now possible to use some of the preliminary results from the Industrial Census 44 which was conducted in 2003/04 to see how the population of firms in Ghana has been changing. 4.3 Firm Growth in the Population Why is the size distribution of firms important? As was shown in chapters 2 and 3, firms of different size have very different levels of labor productivity, capital intensity and wages. As firms expand in size they do not use the same technology ie they do not use factor inputs in the same proportion. Larger firms have far more capital per employee than small firms. For any given level of investment if expansion occurs with the “small� firm technology far more jobs will be created than if a “large� firm technology is used. Understanding the changes in firm size distribution allows us to predict the impact of investment on wages and poverty eradication in general.. Table 4.1: Census Data on Manufacturing Firms 1987 2003 Employ- Employ- Size Firms % % Firms % % ment. ment. 1-4 2,884 35 7,400 5 14,352 55 35,834 15 5-9 3,391 41 21,264 14 7,829 30 48,982 20 10-19 1,101 13 14,306 9 2,427 9 30,784 13 20-29 310 4 7,235 5 541 2 12,405 5 30-49 232 3 8,594 5 401 2 14,538 6 50-99 191 2 13,116 8 287 1 18,270 8 100-199 114 1 15,866 10 124 <1 16,819 7 200-499 74 1 22,596 14 87 <1 26,240 11 500+ 52 1 46,707 30 40 <1 39,644 16 Total 8,351 100 157,084 100 26,088 100 243,516 100 Ave. Size 19 9 Source: Ghana Statistical Service, National Industrial Census 1987, Phase I Report, and National Industrial Census Bulletin No. 1, 2005. Note: Size categories and average size refer to employees per establishment. In Table 4.1 we present a comparison of the results from the 1987 Industrial Census with the results reported by the GSO from the listing which preceded the full scale census. In 1987 the Census counted 8,351 enterprises which has risen more than three fold to 26,088 in 2003. This represents a growth rate in enterprise formation of 7 per cent per year. In contrast to this explosive growth in the number of establishments, overall employment grew by only 55 per cent, a growth rate of 2.7 per year, very close to the growth rate of the labor supply (see chapter 1). It is clear from a comparison of the two censuses that small firms have dramatically increased both in numbers and as a proportion of the stock of firms. In 1987, 35 per cent of firms employed between 1 and 4 employees (ie in very small scale enterprises), by 2003 this percentage had risen 45 to 55 per cent. In 1987 these micro enterprise employed just over 7000 people, by 2003 this had risen to nearly 36,000. If we look at firms employing more than 100 employees then the number of such enterprises was virtually unchanged between 1987 and 2003 (from 240 to 251), while employment had fallen from 85,169 to 82,703 employees. In summary a comparison of the two industrial censuses shows that average firm size in Ghana has collapsed from an average of 19 employees in 1987 to 9 in 2003. Firms in Ghana have always been small and they have been getting much smaller. As long as this trend in firm size distribution continues, potential workers will face depressed wage prospects. It is likely that this phenomenon is self-reinforcing with potential workers choosing self-employment over low wage jobs. 4.4 Firm Performance and the Investment Climate: Key Issues for Further Analysis In this and the remaining sub-sections we focus on key aspects of the investment climate, based on the results discussed thus far. The business climate for firms will be the focus of the upcoming Investment Climate Assessment for Ghana. This Assessment will be based on a new firm survey scheduled to be conducted in 2007, that will enable benchmarking with appropriate comparator countries from around the world. The survey will be enumerated using a core questionnaire that is used around the world, that will enable us to compile information on the costs of doing business as well as information on factor markets (labor and finance), the regulatory environment and the obstacles that firms face while exporting and importing goods. A related worker survey will gather wage data and other information on the characteristics of the labor market. The reforms in Ghana to its trading and exchange rate regimes have eliminated much of the protection, which it could be argued previously limited competition. Yet, as has been shown, firms remain conspicuously unsuccessful through the early part of this decade. Few firms export, investment rates are low and, for the sample of firms used in this study, output has fallen in the period 2000-2002. However, this situation may have changed since 2002 and available evidence indicates that things are in fact improving in terms of investment and private sector growth. We need to understand the key factors constraining firms, whether these have changed since the period of the last survey. We also need to assess those constraints that still need to be addressed. 46 What are the factors that constrain the ability of firms to grow? In each survey since the mid 1990s managers of the firms have been asked to identify their three most important problems facing the firm. In Table 4.2, their answers over the period 1996 - 2002 are summarized. The data shows the percentage of firms who reported one of the identified problems as one of their three most important problems over the period from 1996 to 2002. It will be noted that five issues - corruption, difficulty in obtaining licenses, gaining investment benefits, labor regulations and price controls - were mentioned by less than 1 per cent of firms on average. Table 4.2 Problems Firms Face: 1996-2002 1996 1998 2002 OWNERSHIP REGULATIONS 1.1 0 1.8 TAXES 4.9 3.6 11.7 GOVERNMENT RESTRICTIONS ON ACTIVITIES 3.8 4.2 2.7 GAINING INVESTMENT BENEFITS 0.5 0 2.7 LABOUR REGULATIONS 0 0.6 0 DIFFICULTY IN OBTAINING LICENSES 1.6 1.2 0 CORRUPTION 0.5 0.6 1.8 PRICE CONTROLS 0.5 0.6 0 LACK OF BUSINESS SUPPORT SERVICES 1.6 1.8 11.7 LACK OF INFRASTRUCTURE 18.7 3.6 5.4 ACCESS TO IMPORTED RAW MATERIALS 3.8 4.2 4.5 COST OF IMPORTED RAW MATERIALS 9.3 12.5 8.1 ACCESS TO DOMESTIC RAW MATERIALS 15.9 20.8 17.1 COST OF DOMESTIC RAW MATERIALS 14.3 17.3 29.7 UTILITY PRICES 3.8 0 2.7 ACCESS TO CREDIT 54.9 58.9 42.3 HIGH INTEREST RATES 11.5 10.7 10.8 INFLATION 6.6 16.7 6.3 INSUFFICIENT DEMAND 26.4 20.8 20.7 ACCESS TO FOREIGN EXCHANGE 1.6 7.7 1.8 HIGH EXCHANGE RATES 3.3 14.9 0.9 COMPETITION FROM IMPORTS 6.6 4.2 3.6 COMPETITION FROM LOCAL FIRMS 3.3 6 9.9 UNCERTAINTY ABOUT GOVERNMENT INDUSTRY POLICIES 2.2 4.8 0.9 LACK OF SKILLED LABOUR 2.7 3.6 1.8 Notes: Averages reported are the percentage of firms reporting the problem as one of the 3 most important constraints facing the firm. 47 Figure 4.3 Firm's Major Problems Access to Credit Access to Domestic Raw Materials Access to Foreign Exchange Access to Imported Raw Materials Competition from Imports Competition from Local Firms Cost of Domestic Raw Materials Cost of Imported Raw Materials Government Restrictions on Activities High Exchange Rates High Interest Rates Inflation Insufficient Demand Lack of Business Support Services Lack of Infrastructure Lack of Skilled Labour Ownership Regulations Taxes Uncertainty about Government Industrial Policies Utility Prices 0 10 20 30 40 50 Percentage of Firms Citing Problem The percentages refer to the firms which cited the problem as one of the three most important problems that they currently face. i It may also be that some of the problems, for example gaining investment benefits, are not thought of as problems as, for other reasons, the firms do not wish to undertake substantial investment. That this may be the case is suggested by two of the reasons that do get frequently cited - access to credit and lack of demand. 4.4.1 Access to Credit In Figure 4.3 the results are summarized for all those problems which were mentioned by more than 1 per cent of firms. It is clear from the figure that the factor most commonly identified as a problem is access to credit. This is given as a major problem by more than 50 per cent of firms in 1996 and 1998 and by 43% in 2002. Figure 4.4 shows the differences between large and small 48 firms for the firms citing lack of demand and access to credit as one of the major problems they face. 49 Figure 4.4 Credit and Demand as Firm's Major Problems Large Medium Small 0 20 40 60 80 Percentage of Firms Citing Problem Credit Demand The percentages refer to the firms which cited the problem as one of the three most important problems that they currently face. As is common across all surveys of firms in Africa, small firms are far more likely to cite access to credit and demand as problems than larger firms. As Figure 4.4 shows this difference across firms by size is much more pronounced for credit than for demand. For small firms, access to credit is cited as a problem by nearly 70 per cent of the firms while for large firms the problem is cited by only just over 20 per cent of firms. While small firms are more likely than large ones to cite demand as a problem the differences across firms by size are much less marked than is the case with access to credit. In order to assess whether the financial infrastructure is particularly adverse in Ghana requires comparative data. All surveys show that small firms find accessing credit more difficult than larger firms. Figure 4.5 shows comparative data drawn from the same countries as in chapter 2 to show how rates of return on capital differ across countries once we control for differences in firm size. 50 Figure 4.5 Rates of Return on Capital by Firm Size Across Countries GHANA KENYA Large NIGERIA TANZANIA SOUTH AFRICA GHANA KENYA Medium NIGERIA TANZANIA SOUTH AFRICA GHANA KENYA Small NIGERIA TANZANIA 0 .5 1 1.5 Median Rate of Return on Capital Note: a small firm is on employing less than 20 people, a medium firm has from 20 to 75 and a large firm employs more than 75. The rate of return on capital is measured as profits (value-added less wages) as a proportion of the value of the capital stock. It is a measure of how profitable investment should be for the firm - if the firm had access to the credit market. For all four countries for which we can compare rates of return across large and small firms, the rate of return is much higher in small than large firms. Further it is particularly striking that the rate of return on capital for small firms in Nigeria and Ghana is twice that for firms in Tanzania and Kenya. It would appear from this comparative data that the financial climate is far more adverse for small Ghanaian firms than in some other African countries. In order to establish the extent to which the financial infrastructure in Ghana is deficient, we move from the perceptions realm to the reality of firms sources of finance. In economies where financial infrastructure is very weak we expect to find a large proportion of financing dominated by retained earnings or informal sources. The table below shows the sources of financing for plant and building investment over the last 5 years. In particular, an examination of the share of investments financed by retained earnings and bank loans suggests that there has been little improvement in the financial infrastructure. On average, conditional on making any investments, Ghanaian firms finance between 60-70% of investments from personal and firm savings and receive only about 13% from the formal credit market. Informal lending provides an average of 1.5% while trade credit accounts for 2% of total financing needs. 51 Table 4.3: Sources of Financing Sources of Financing 1998 1999 2000 2001 2002 Company retained earnings 62.1 59.0 43.1 49.6 62.1 Personal savings 5.7 5.8 11.5 8.0 2.2 Borrowed from friends and relations 0.3 2.5 0.0 2.4 2.9 Bank loan or overdraft 12.2 14.0 16.5 18.0 10.6 Suppliers credit 2.7 1.6 0.0 2.4 3.2 Borrowed from money lender 0.0 0.0 0.0 0.0 1.6 Borrowed from parent or holding co. 3.0 0.6 0.0 0.0 0.0 Sale of equity 0.0 0.0 0.0 0.0 0.0 New partner 0.0 0.0 0.0 0.0 0.0 Other 1.7 2.2 1.9 0.0 1.3 A comparison with our set of comparator countries suggests that Ghana is not very different from a number of Sub Saharan economies at similar levels of development. In Uganda, Tanzania and Madagascar, firm and individual savings account for about 70% of investment needs, while banks finance between 11-17% of total financing needs. To the extent that firms in Kenya and Mauritius can finance nearly one third of their financing needs from the formal credit market, these countries have a financial infrastructure that is clearly superior to Ghana’s. Table 4.4 Sources of Financing for New Investments: Mean Percentages by Source Source: Kenya Tanzania Uganda Mauritius Madagascar South Ghana Africa Retained earnings 52.7% 68.0% 71.1% 56.0% 76% 58.4% 62.1% Banks 36.6% 17.4% 11.6% 28.0% 11.9% 16.5% 10.6% Trade credit 4.0% 2.0% 0.5% 0.0% 2% 0.6% 3.2% Equity 0.3% 4.8% 2.0% 0.0% 2.00% 0.13% 0.0% Informal sources 1.5% 3.7% 1.5% 0.0% 0.05% 0.19% 4.5% Other 8.5% 4.2% 4.5% 16% 7% 6.9% 19.6% Alternatively we can look at the fraction of firms that have access to various sources of external financing for financing new investment and operating costs. The graph below shows the pattern of access throughout the 1990s during the reform era. It is clear from this graph that the formal credit market has not seen any marked improvements as far as providing credit to the private sector is concerned. In the early 1990s access to formal loans appears to have increased from about 10% to 20% of firms by 1994. However, the trend flattens out from the mid 1990s to the 52 end of the period for which we have data. It is also noteworthy that the pattern of demand for informal loans tracks that of formal loans. This probably indicates a complementarity between formal and informal credit and further suggests that firms are rationed in the market for formal loans. The pattern of overdraft access has been much more stable with about 25% of all firms reporting having access to an overdraft. Access to trade credit has been stable throughout the entire period with about half of all firms reporting receiving credit from suppliers. The graph below highlights in a very powerful way the limits of liberalization reform of the early 1990s. The graph suggests that a number of constraints were not targeted by privatization and improvements in regulatory oversight of the financial market. Figure 4.6: Access to External Finance 70 60 50 40 % 30 20 10 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Overdrafts Formal loans Informal Loans Trade Credit In order to understand some of the constraints associated with formal credit market we use the survey data to present a summary of reasons why firms do not borrow and why other firms access the informal credit market as a response to these difficulties. Given that rates of return on capital are directly proportional to firm size, we expect that the underlying cost of capital in the economy and the firm size distribution will affect borrowing behavior. We present this data by year in the tables below. The table below shows the distribution of firms across different reasons for not borrowing. About 53 27% of firms did not need a loan in 1999 compared to 45% in 2002. 32% and 23% do not borrow because they think the interest rate is too high in 1999 and 2002 respectively. Collateral requirements are an impediment to only 6% of firms not applying for credit while 12% and 8% of firms find the process too cumbersome in 1999 and 2002 respectively. Table 4.5: Reasons that Firm Did Not Apply for a Loan Reason did not apply for a Loan 1999 2002 Percent Percent Inadequate Collateral 6.61 6.33 Do not want debt 6.61 20.25 Process too difficult 11.57 7.59 Didn't need one 20.66 25.32 Didn't think would get one 12.40 6.33 Interest rate too high 32.23 22.78 Already too much debt 3.31 1.27 Other 6.61 10.13 Total 100.00 100.00 A large percentage of firms are either discouraged by the procedural requirements for obtaining credit or the cost of credit. By this definition, about 64% and 43% of firms not applying are discouraged in 1999 and 2002 respectively. For firms that cannot use firm or personal savings to meet financing needs, the informal credit market is the alternative of choice. Firms were asked why they borrowed from the informal credit market. The table below provides the distribution of responses for 1998-2002. The ease of borrowing is the most frequently cited reason for firms accessing the informal market. More than half of firms in 1998 and 1999 report that borrowing formalities were the reason they chose to borrow from the informal credit market. While the numbers fall in 2000- 2002, an average of 24% of firms report that borrowing requirements were the primary reason for accessing informal loans. The second most important reason is the price of credit. Between 20- 30% of firms find the price of informal loans to be more competitive than formal loans. This is particularly surprising given that the formal credit market is likely to be much bigger (per lender) than the informal market. The flexibility of payback appears to have become an important reason for accessing informal credit. Less than 10% of firms report this before or during 1999, but an average of 21% of firms report this after 1999. 54 Table 4.6: Reasons for Informal Borrowing 1998 1999. 2000 2001. 2002 Most favorable interest rate 25.93 29.03 21.43 23.08 20.00 Easier formalities 55.56 54.84 21.43 15.38 35.00 No collateral required 7.41 9.68 21.43 15.38 10.00 Flexible payback 3.70 6.45 14.29 23.08 25.00 Other 3.70 - 14.29 15.38 10.00 Total 100.00 100.00 100.00 100.00 100.00 It is important to apply some caution in the interpretation of reported data. As it is, this data suggests that banks and other lending institutions in the formal credit market have very onerous requirements for firms to acquire credit and in addition charge a very high price for each cedi borrowed. These two implications are consistent with a residual inefficiency in formal lending institutions as well as an inefficient use of information about credit worthiness. In the upcoming Assessment, one of the main tasks will be to evaluate whether these constraints are still present in the financial sector and whether access to credit is in fact a central problem for small firms. We already have some recent data about the ease of borrowing for firms in Ghana. Data from the Doing Business Survey (World Bank, 2006) highlights two important pillars of the formal credit market--property rights enforcement and the availability of information about the credit worthiness of borrowers. In order for banks to lend at the optimal price of debt, a number of factors must be in place. The legal system must be able to assign and enforce clear property rights that make it possible for firms to present easily verifiable collateral and for banks to be able to re-possess any security in the event that borrowers default. It will be useful to understand how much of this has been put in place since 2002. However, in order for banks to determine both the level and cost of credit, they need to have information about the credit-worthiness of the borrower. This information can be collected by the bank through its dealings with a particular borrower (private information) or can be obtained from a public registry. The public registry serves as a repository for borrowers’ performance on a wide range of credit obligations over the borrowers lifetime and is a considerably cheaper and much more informative than private information. In the upcoming Assessment, we will focus on whether Ghana has been successful in increasing the supply of information about borrowers, via a viable public registry system or other means. 55 Table 4.7: Doing Business 2006--Determinants of Ease of Access to Formal Financing Legal Rights Index Credit Information Public registry Private bureau Index coverage (% adults) coverage (% adults) Region/Country Ghana 5 0 0 0 Kenya 8 5 0 0.1 Nigeria 7 3 0 0.3 South Africa 5 5 0 63.4 Tanzania 5 0 0 0 Sub-Saharan Africa 4.4 1.5 0.8 3.5 East Asia & Pacific 5.3 1.8 1.7 9.6 Europe & Central Asia 5.6 2.5 1.4 6.6 Latin America & Caribbean 3.8 4.5 11.5 31.2 Middle East & North Africa 4.1 2 1.9 1.7 OECD: High income 6.3 5 7.5 59 South Asia 3.8 1.8 0.1 0.6 The table above shows the averages for the performance of the legal system and the existence of information registers in Ghana and the comparator countries. With a legal rights index average of 5, Ghana is on par with South Africa and Tanzania, lower than Kenya and Nigeria but slightly better than the Sub-Saharan African average of 4.4. Ghana performs considerably worse than the comparator countries on the ease of obtaining information on the credit-worthiness of borrowers. Ghana and Tanzania both score 0 on the credit information index compared to 3 for Nigeria and 5 for Kenya and South Africa. No country has a public registry for individuals. The private sector does no better in 4 of the 5 countries--only South Africa has private credit bureaus that collect information on a substantial percentage of individuals. The presence of a well-functioning credit registry is strongly associated with the performance of the formal credit market as suggested by the numbers in Table 4.7 above (% investment financed by banks). While the data here cannot provide incontrovertible evidence as to whether Ghana’s formal credit market could be improved by the establishment of a credit bureau alone; it is strongly suggestive. In the upcoming Assessment, we will look at efforts made in this area and whether they are adequate in terms of solving the information problem associated with smaller firms and their abilty to access credit. 56 4.4.2 Access to and Cost of Domestic Raw Materials how the problems of access to and cost of domestic raw materials differ across firms by size. As is clear from the figure below, large firms focus on the problem of access while small firms focus on the problem of cost. Figure 4.7 Access and Cost of Domestic Raw Materials as Firm's Major Problem Large Medium Small 0 10 20 30 Percentage of Firms Citing Problem Access Cost The percentages refer to the firms which cited the problem as one of the three most important problems that they currently face. To understand what underlies the problems shown in Figure 4.7, we present in Figure 4.8, a breakdown of this problem by sector. 57 Figure 4.8 Access and Cost of Domestic Raw Materials Non-wood Wood 0 20 40 60 Percentage of Firms Citing Problem Access Cost The percentages refer to the firms which cited the problem as one of the three most important problems that they currently face. As is clear from Figure 4.8, the problem of access to raw materials is overwhelmingly a problem for the wood sector where it is cited as a problem by more than 60 per cent of firms while cost is less of a problem for wood than for other sectors. What accounts for this pattern? As was shown in Chapter 2, the Ghanaian wood sector is one of the most export-oriented sectors in the countries for which we have comparative data. It was also shown there that large firms are those most likely to enter the export market. So for large, export-oriented firms, cost is not the issue--costs do reflect the large changes in the nominal exchange rate but as they are in the export market the exchange rate also changes the output prices too. Access to raw materials is an issue and the larger the firm, the larger the problem. For large, export oriented firms, issues of property rights were also of major concern. Another concern of the larger firms was the activities of smaller firms who were logging if not illegally then certainly on the margins of the law. For these smaller firms access could be gained but they focused on the cost. The issue of cost for raw materials is of much wider application than simply for the wood sector as in most sectors raw materials inputs are a major part of the cost structure. The upcoming Assessment will enable the cost structure of firms to be carefully constructed, and compared with cost structures in comparator countries. Apart from the cost of raw materials, the investment climate firm survey will also enable computations on the “indirect costs� that firms face, including the cost of electricity, transport, and other inputs into the production process. 58 4.4.3 Taxes While taxes are not an important problem before 2000, 12% of firms report taxes as one of the leading investment climate constraints in 2002. Available data from the World Bank’s Doing Business surveys show the extent to which regulatory requirements as relates to taxation are cumbersome. While the average economy in Sub-Saharan Africa makes 41.4 different tax payments, Ghana has 35. This is comparable to Nigeria and South Africa but much greater than Kenya’s 17 and considerably less than Tanzania’s 48. However, despite having to make more payments, the time spent preparing and paying taxes in Ghana is considerably less than Kenya. In addition, a crude measure of the tax rate in Ghana is comparable to South Africa and lower than Tanzania and Kenya. The Investment Climate Assessment will enable us to benchmark Ghana’s tax system further, both in terms of tax rates and in terms of the quality of tax administration. Table 4.8: Taxes Region/Economy Payments (number) Time (hours) Total tax payable (% gross profit) Ghana 35 304 45.3 Kenya 17 372 68.2 Nigeria 36 1,120 27.1 South Africa 32 350 43.8 Tanzania 48 248 51.3 Sub-Saharan Africa 41.4 394 58.1 East Asia & Pacific 28.2 249.9 31.2 Europe & Central Asia 46.9 431.5 50.2 Latin America & Caribbean 48.2 529.3 52.8 Middle East & North 27.3 241.9 35.1 Africa OECD: High income 16.3 197.2 45.4 South Asia 25.8 331.7 35.3 Source: Doing Business Survey 2005 4.4.4 Corruption Ghana has made real progress in dealing with corruption, as per the data shown in Table 4.9 below. The table shows the average index of government’s ability to control corruption as well as the percentage of countries with worse performance (percentile (in bold)) below. The table shows a clear rise in Ghana’s ranking or relative improvement in its ability to control corruption between 1996 and 2004. The magnitude of the improvement in Ghana’s ranking is only second to Tanzania. 59 Control of Corruption 2004 2002 2000 1998 1996 Ghana -0.17 -0.39 -0.34 -0.44 -0.47 51.7 41.8 44.6 42.1 35.3 Nigeria -1.11 -1.32 -1.06 -1.01 -1.2 8.9 3.1 6.5 5.5 5.3 Kenya -0.89 -1.09 -1.04 -0.92 -1.05 18.7 8.7 9.1 9.8 8 South Africa 0.32 0.11 0.28 0.21 0.35 60.9 58.2 64.2 67 66.9 Tanzania -0.57 -0.97 -0.97 -0.95 -1.03 36 15.3 12.9 8.7 9.3 The Investment Climate Assessment will enable us to benchmark the results of these efforts, and gather information on the problem of corruption specific to the private sector. In particular, we will ask firms the percentage of sales spent in “informal payments to get things done� and the number of visits of government officials to the firm. These data will again be benchmarked with comparator countries, to highlight the cost of corruption in Ghana relative to other locations. 4.4.5 Infrastructure There have been significant improvements in the quality of Ghana’s infrastructure over the last decade. The data for 1996-2002 show that particularly for electricity, there has been a large improvement in the reliability of electricity services in the run up to 2002. In 1996, more than 40% of firms reported receiving electricity for an average of 3 days or less /week. In 1998, this number had fallen eight-fold to just over 5% and has remained at this level until 2002. However, there is recent, anecdotal evidence of a much higher frequency of disruption of electricity services resulting from sporadic rainfall patterns and operating deficiencies. The performance of water and telephone services has also been more mixed. Phone and water services improved between 1996 and 1998 but have become much worse since 1998. Evidence on these variables is not available in a consolidated form. However, existing research and various Investment Climate Assessments conducted in Africa and elsewhere show that these variables significantly affect firm performance. The upcoming Assessment will gather a large amount of information on the quality, quantity, and cost of electricity supplied to firms, on access to 60 transportation, and on infrastructure related to exporting including customs and ports. This information will be benchmarked with comparator countries as well as used to calculate the cost structure of firms. It will enable us to show the effect of the cost and quality of infrastructure on the competitiveness of firms. 61 4.4.6 Labor Market Flexibility The data from previous surveys show that larger firms pay substantially more than smaller ones and this differential does not necessarily reflect skill differences. This linkage from firm size to wages ensures that larger firms are much more capital intensive than smaller ones (Chapter 2 Table 2.2) and thus that firms that can enter the export market have factor proportions inconsistent with exports being a source of substantial profits. This lack of profitability in the export market ensures that growth is limited and in particular there is little job creation for the relatively unskilled. A key question in the upcoming Assessment is—how much of this has changed? Available evidence from the Doing Business database indicates that the flexibility of hiring and firing workers has increased—on a scale of 0-100, Ghana has a difficulty of hiring index of 11 and a difficulty of firing index of 50. The rigidity of employment index is 34, on a scale of 0-100. There is still some work to be done--Ghana ranks 48th in the world in terms of the ease of hiring and firing in 2006. But overall, the picture is that of a fairly flexible labor market, when measured on a global scale. The upcoming Assessment will enable this claim to be tested rigorously. Data will be gathered from workers employed at the factories that are surveyed in order to analyze the determinants of wages, the degree of labor mobility and other measures of labor market flexibility. Information will also be collected from firm managers regarding workers and the regulatory environment that surrounds the formal labor force. Conclusion Existing data for the period 1996-2002 shows that firms in the private sector have not grown very much in Ghana. Underneath this is a set of factors that is constraining the private sector. Some of this may have changed as a result of policy reforms, other obstacles may still be unaddressed. The key objective of the Investment Climate Assessment is to evaluate the changes in the private sector since the last round of data collection in 2002, and to analyze the business climate in order to update our understanding of the constraints that firms face. To this extent, the available evidence helps us direct our attention to areas such as access to credit and labor market flexibility. But there is also evidence that some things have changed for the better, while other aspects of the investment climate, such as the supply of infrastructure, have changed for the worse. A rigorous Assessment, based on a new round of data collection, will help us identify the key constraints facing firms and the policy reforms that must be undertaken to address these constraints. It will also enable us to think about strategic directions for the private sector, the role of relatively newer sectors such as ICT, and some of the choices that might facilitate Ghana’s goal of becoming a middle-income country by 2020. 62 References Albaek, Karsten, Mahmood Arai, Rita Asplund, Erling, Barth and Erik Madsen (1998). Measuring Wage Effects of Plant Size, Labour Economics. 5: 425-48. Black, Dan A., Brett J. Noel and Zheng Wang (1999). On-the-job Training, Establishment Size, and Firm Size: Evidence for Economies of Scale in the Production of Human Capital, Southern Economic Journal. 66: 82-100. Deaton, A. (2003) “Measuring poverty in a growing world (or measuring growth in a poor world)�, NBER Working Paper 9822. Demery, L. and Lyn Squire (1996) “Macroeconomic adjustment and poverty in Africa: an emerging picture�, The World Bank Research Observer, vol. 11, no. 1, February, pp.39- 59. Ghana Statistical Service (GSO) (1995) The Pattern of Poverty in Ghana: 1988-1992, Accra, November 1995. Ghana Statistical Service (GSO) (2000) Poverty Trends in Ghana in the 1990s, Accra, October 2000. Horton, Susan, Dipak Mazumdar and Ravi Kanbur. 1994. “Overview,� in Labor Markets in an Era of Adjustment, Vol. 1. Washington D.C.: World Bank. Kingdon, G. Sandefur, J. and F. Teal (2004) Patterns of Labor Demand (Africa Region Employment Issues – Regional Stocktaking Review) http://www.gprg.org/pubs/reports/pdfs/2005-11-kingdon-sandefur-teal.pdf Kremer, Michael and Eric Maskin (1996). Segregation by Skill and the Rise in Inequality, National Bureau of Economic Research, Working Paper 5718, Cambridge (MA). Lester, Richard A. (1967). Pay Differentials by Size of Establishment, Industrial Relations. 7:57- 67. Scherer, F.M. (1976). Industrial Structure, Scale Economies and Worker Alienation, in: R.T. Masson and P.D. Qualls (eds.), Essays on Industrial Organization in Honour of Joe S.Bain, Cambridge (Mass.): Ballinger. 63 Shapiro, Carl and Joseph E. Stiglitz (1984). Equilibrium Unemployment as a Worker Discipline Device, American Economic Review. 74: 433-44. Söderbom, M. and F. Teal (2004). “Size and Efficiency in African Manufacturing Firms: Evidence from Firm-Level Panel Data,� Journal of Development Economics 73, pp.36- 394. Söderbom, M. and F. Teal and A. Wambugu (2005) “Unobserved heterogeneity and the relation between earnings and firm size: evidence from two developing countries�, Economics Letters, 87, pp.153-159. Teal, F. (2001) Education, incomes, poverty and inequality in Ghana in the 1990s, CSAE Working paper No. 2001.21. Teal, F. (2002) “Export growth and trade policy in Ghana in the twentieth century� The World Economy, 25(9), pp. 1319-1337. World Bank (2004) World Development Indicators. World Bank (2006) Doing Business. 64 Appendix: Tables Table 3.1 Comparative Firm characteristics by Country Mean Std. Dev. Min Max Ghana Ln Output/Employee 8.10 1.25 2.67 11.85 Ln Capital/Employee 7.05 1.97 2.30 12.38 Ln Raw Materials/Employee 7.27 1.42 -0.56 11.51 Ln Indirect Cost/Employee 5.47 1.75 -0.47 10.06 Ln Employment 3.17 1.43 0 7.50 Dummy =1 if Firm Exports 0.17 0.37 0 1 Dummy = 1 if Any foreign Ownership 0.19 0.39 0 1 Firm's Age 18.49 12.22 1 73 Span of Data 1991 2002 N=1563 Kenya Ln Output/Employee 8.90 1.25 5.12 13.13 Ln Capital/Employee 8.56 1.67 3.41 13.13 Ln Raw Materials/Employee 8.20 1.44 1.76 12.64 Ln Indirect Cost/Employee 6.45 1.24 1.60 11.32 Ln Employment 3.36 1.69 0 7.85 Dummy =1 if Firm Exports 0.33 0.47 0 1 Dummy = 1 if Any foreign Ownership 0.18 0.38 0 1 Firm's Age 20.47 13.43 1 74 Span of Data 1992 1999 N=900 Nigeria Ln Output/Employee 9.17 1.40 5.02 13.71 Ln Capital/Employee 8.23 2.08 2.73 12.86 Ln Raw Materials/Employee 8.41 1.68 2.88 13.64 Ln Indirect Cost/Employee 6.79 1.76 0.33 11.07 Ln Employment 3.50 1.70 0 8.53 Dummy =1 if Firm Exports 0.08 0.27 0 1 Dummy = 1 if Any foreign Ownership 0.20 0.40 0 1 Firm's Age 22.17 10.95 1 59 Span of Data 1998 2003 N=631 Tanzania Ln Output/Employee 8.02 1.31 4.47 12.08 Ln Capital/Employee 7.36 1.98 0.52 12.94 Ln Raw Materials/Employee 7.27 1.52 3.41 11.81 Ln Indirect Cost/Employee 5.80 1.38 1.57 10.05 Ln Employment 2.90 1.49 0 7.86 65 Dummy =1 if Firm Exports 0.14 0.35 0 1 Dummy = 1 if Any foreign Ownership 0.15 0.36 0 1 Firm's Age 15.42 11.95 1 94 Span of Data 1992 2000 N=967 South Africa Ln Output/Employee 10.62 0.68 8.30 12.44 Ln Capital/Employee 9.63 1.23 6.19 12.75 Ln Raw Materials/Employee 9.84 0.89 7.20 12.20 Ln Indirect Cost/Employee 7.69 0.83 5.60 10.36 Ln Employment 4.91 0.89 3.40 7.75 Dummy =1 if Firm Exports 0.69 0.46 0 1 Dummy = 1 if Any foreign Ownership 0.24 0.43 0 1 Firm's Age 20.51 17.07 1 94 Span of Data 1997 1998 N=313 Table 2.2 Comparative Firm Characteristics by Firm Size Mean Std. Dev. Min Max Large Ln Output/Employee 9.57 1.27 5.41 13.71 Ln Capital/Employee 9.27 1.31 3.72 13.13 Ln Raw Materials/Employee 8.78 1.49 3.71 13.64 Ln Indirect Cost/Employee 7.33 1.28 -0.47 11.32 Ln Employment 5.33 0.90 3.14 8.53 Dummy =1 if Firm Exports 0.51 0.50 0 1 Dummy = 1 if Any foreign 0.44 0.50 0 1 Ownership Firm's Age 22.90 14.41 1 94 N=1,181 Medium Ln Output/Employee 8.74 1.32 4.45 13.13 Ln Capital/Employee 8.35 1.63 2.79 12.94 Ln Raw Materials/Employee 7.94 1.56 1.76 12.67 Ln Indirect Cost/Employee 6.30 1.46 0.23 10.29 Ln Employment 3.60 0.50 0.69 5.08 Dummy =1 if Firm Exports 0.21 0.41 0 1 Dummy = 1 if Any foreign 0.16 0.37 0 1 Ownership Firm's Age 20.14 12.25 1 74 N=1,264 Small Ln Output/Employee 7.87 1.22 2.67 12.96 66 Ln Capital/Employee 6.50 1.87 0.52 12.13 Ln Raw Materials/Employee 7.13 1.43 -0.56 12.85 Ln Indirect Cost/Employee 5.20 1.45 0.23 10.01 Ln Employment 1.91 0.74 0 3.69 Dummy =1 if Firm Exports 0.05 0.21 0 1 Dummy = 1 if Any foreign 0.04 0.21 0 1 Ownership Firm's Age 15.63 11.28 1 94 N=1,929 Table 2.4 Exporting to Africa and Outside Africa: By Country Std. Obs Mean Dev. Ghana Percentage of Firms Exporting 1205 19 39 Percentage of Output Exported Conditional on any Exports 226 55 37 Percentage of Output Exported 1205 10 27 Percentage of Firms Exporting to Africa 1205 9 29 Percentage of Output Exported to Africa Conditional on any Exports 113 21 24 Percentage of Output Exported to Africa 1205 2 9 Percentage of Firms Exporting Out of Africa 1205 13 34 Percentage of Output Exported Out of Africa Conditional on any Exports 161 62 35 Percentage of Output Exported Out of Africa 1205 8 25 Kenya Percentage of Firms Exporting 329 39 49 Percentage of Output Exported Conditional on any Exports 126 28 30 Percentage of Output Exported 329 11 23 Percentage of Firms Exporting to Africa 329 34 48 Percentage of Output Exported to Africa Conditional on any Exports 113 18 19 Percentage of Output Exported to Africa 329 6 14 Percentage of Firms Exporting Out of Africa 329 12 33 Percentage of Output Exported Out of Africa Conditional on any Exports 40 38 36 Percentage of Output Exported Out of Africa 329 5 18 Nigeria Percentage of Firms Exporting 465 8 27 Percentage of Output Exported Conditional on any Exports 37 33 32 Percentage of Output Exported 465 3 13 Percentage of Firms Exporting to Africa 465 6 24 Percentage of Output Exported to Africa Conditional on any Exports 28 29 23 Percentage of Output Exported to Africa 465 2 9 67 Percentage of Firms Exporting Out of Africa 465 4 20 Percentage of Output Exported Out of Africa Conditional on any Exports 20 21 29 Percentage of Output Exported Out of Africa 465 1 7 Tanzania Percentage of Firms Exporting 371 17 38 Percentage of Output Exported Conditional on any Exports 60 24 27 Percentage of Output Exported 371 4 14 Percentage of Firms Exporting to Africa 371 13 33 Percentage of Output Exported to Africa Conditional on any Exports 47 13 12 Percentage of Output Exported to Africa 371 2 6 Percentage of Firms Exporting Out of Africa 371 7 26 Percentage of Output Exported Out of Africa Conditional on any Exports 27 30 35 Percentage of Output Exported Out of Africa 371 2 12 Std. South Africa Obs Mean Dev. Percentage of Firms Exporting 145 70 46 Percentage of Output Exported Conditional on any Exports 102 18 20 Percentage of Output Exported 145 13 19 Percentage of Firms Exporting to Africa 145 62 49 Percentage of Output Exported to Africa Conditional on any Exports 90 10 14 Percentage of Output Exported to Africa 145 6 12 Percentage of Firms Exporting Out of Africa 145 44 50 Percentage of Output Exported Out of Africa Conditional on any Exports 64 15 20 Percentage of Output Exported Out of Africa 145 6 15 Exporting to Africa and Outside Africa: By Firm Size Large Percentage of Firms Exporting 705 49 50 Percentage of Output Exported Conditional on any Exports 342 37 36 Percentage of Output Exported 705 18 31 Percentage of Firms Exporting to Africa 705 35 48 Percentage of Output Exported to Africa Conditional on any Exports 245 15 17 Percentage of Output Exported to Africa 705 5 12 Percentage of Firms Exporting Out of Africa 705 28 45 Percentage of Output Exported Out of Africa Conditional on any Exports 199 45 40 Percentage of Output Exported Out of Africa 705 13 30 Medium Percentage of Firms Exporting 728 21 41 Percentage of Output Exported Conditional on any Exports 151 36 32 Percentage of Output Exported 728 8 21 Percentage of Firms Exporting to Africa 728 15 35 Percentage of Output Exported to Africa Conditional on any Exports 106 19 21 Percentage of Output Exported to Africa 728 3 11 68 Percentage of Firms Exporting Out of Africa 728 11 31 Percentage of Output Exported Out of Africa Conditional on any Exports 79 44 32 Percentage of Output Exported Out of Africa 728 5 17 Small l Percentage of Firms Exporting 1082 5 23 Percentage of Output Exported Conditional on any Exports 58 39 33 Percentage of Output Exported 1082 2 12 Percentage of Firms Exporting to Africa 1082 4 19 Percentage of Output Exported to Africa Conditional on any Exports 40 27 27 Percentage of Output Exported to Africa 1082 1 7 Percentage of Firms Exporting Out of Africa 1082 3 17 Percentage of Output Exported Out of Africa Conditional on any Exports 34 34 34 Percentage of Output Exported Out of Africa 1082 1 8 Table 2.7 Firm Level Investment Mean Std. Dev. Min Max Ghana Percentage of Firms Investing 43.25 49.57 0 100 Capital Output Ratio 1.15 2.17 0.01 18.06 Investment to Value-added Ratio 0.06 0.15 0 1.00 Investment to Capital Stock Ratio 0.04 0.10 0 0.90 Profits per Employee (lagged One period) 1,353 3,160 -5,675 48,924 Growth Rate of Output (%pa) -0.06 39.63 -98.66 99.99 Growth Rate of Employment (%pa) -0.04 25.95 -98.08 97.08 Span of Data 1992 2002 N=1,082 Ln (investment) 8.26 3.02 0.02 14.99 Investment to Capital Ratio 0.10 0.14 0 0.90 (conditional on any investment) N=468 Kenya Percentage of Firms Investing 55.03 49.82 0 100 Capital Output Ratio 1.65 2.52 0.02 18.60 Investment to Value-added Ratio 0.10 0.19 0 1.00 Investment to Capital Stock Ratio 0.05 0.11 0 1.19 Profits per Employee (lagged One period) 3,837 11,849 -67,472 118,250 Growth Rate of Output (%pa) 1.55 36.15 -98.05 97.20 Growth Rate of Employment (%pa) -2.80 27.16 -99.85 95.78 Span of Data 1993 1999 N=318 Ln (investment) 9.53 2.80 2.34 14.31 Investment to Capital Ratio 0.09 0.13 0.00 1.19 (conditional on any investment) (N=175) 69 Nigeria Percentage of Firms Investing 47.22 49.99 0 100 Capital Output Ratio 1.26 1.93 0.01 9.93 Investment to Value-added Ratio 0.09 0.18 0 0.98 Investment to Capital Stock Ratio 0.10 0.29 0 3.63 Profits per Employee (lagged One period) 3,492 5,133 -10,204 33,074 Growth Rate of Output (%pa) -0.29 34.46 -98.06 99.73 Growth Rate of Employment (%pa) -1.83 27.25 -98.08 91.63 Span of Data 1999 2003 N=396 Ln (investment) 9.72 3.04 2.38 17.49 Investment to Capital Ratio 0.20 0.39 0.00 3.63 (conditional on any investment) N=187 Tanzania Percentage of Firms Investing 33.33 47.20 0 100 Capital Output Ratio 1.33 2.21 0.01 19.78 Investment to Value-added Ratio 0.05 0.15 0 0.97 Investment to Capital Stock Ratio 0.03 0.07 0 0.44 Profits per Employee (lagged One period) 2,075 5,470 -872 40,075 Growth Rate of Output (%pa) 1.37 37.44 -99.86 97.55 Growth Rate of Employment (%pa) -0.51 30.48 -95.55 91.63 Span of Data 1993 2000 N=372 Ln (investment) 7.39 3.81 -2.29 17.32 Investment to Capital Ratio 0.08 0.09 0.00 0.44 (conditional on any investment) N=124 South Africa Percentage of Firms Investing 83.45 37.29 0 100 Capital Output Ratio 0.71 0.83 0.03 4.35 Investment to Value-added Ratio 0.10 0.15 0 0.79 Investment to Capital Stock Ratio 0.10 0.16 0 1.08 Profits per Employee (lagged One period) 11,143 19,035 -132,521 86,735 Growth Rate of Output (%pa) -12.60 24.46 -78.20 98.78 Growth Rate of Employment (%pa) -3.63 21.71 -69.31 92.80 Span of Data 1998 1998 N=139 Ln (investment) 11.82 1.69 8.05 16.14 Investment to Capital Ratio 0.12 0.17 0.00 1.08 (conditional on any investment) N=116 70 Large Mean Std. Dev. Min Max Percentage of Firms Investing 68.21 46.60 0 100 Capital Output Ratio 1.58 2.31 0.01 18.60 Investment to Value-added Ratio 0.11 0.17 0 1.00 Investment to Capital Stock Ratio 0.07 0.19 0 3.63 Profits per Employee (lagged One period) 5,203 12,506 -132,521 118,250 Growth Rate of Output (%pa) -0.87 32.20 -95.44 92.77 Growth Rate of Employment (%pa) -1.70 21.27 -78.17 95.78 Span of Data 1992 2003 N=670 Ln (investment) 11.50 2.09 2.95 17.49 Investment to Capital Ratio 0.10 0.23 0.00 3.63 (conditional on any investment) N=457 Medium Percentage of Firms Investing 42.17 49.42 0 100 Capital Output Ratio 1.44 2.23 0.01 19.78 Investment to Value-added Ratio 0.08 0.18 0 0.98 Investment to Capital Stock Ratio 0.05 0.12 0 1.19 Profits per Employee (lagged One period) 2,542 5,518 -7,430 47,645 Growth Rate of Output (%pa) -2.36 37.22 -98.51 98.78 Growth Rate of Employment (%pa) -1.81 24.45 -95.55 97.08 N=709 Span of Data 1992 2003 Ln (investment) 8.80 2.05 1.93 13.54 Investment to Capital Ratio 0.11 0.16 0.00 1.19 (conditional on any investment) N=299 Small Percentage of Firms Investing 33.84 47.34 0 100 Capital Output Ratio 0.84 1.88 0.01 16.84 Investment to Value-added Ratio 0.04 0.13 0 1.00 Investment to Capital Stock Ratio 0.05 0.14 0 1.88 Profits per Employee (lagged One period) 1,184 2,736 -10,106 33,018 Growth Rate of Output (%pa) 1.43 40.58 -99.86 99.99 Growth Rate of Employment (%pa) 0.07 31.85 -99.85 91.63 N=928 Span of Data 1992 2003 Ln (investment) 5.58 2.06 -2.29 10.80 Investment to Capital Ratio 0.14 0.22 0.00 1.88 (conditional on any investment) N=314 All Firms 71 Percentage of Firms Investing 46.38 49.88 0 100 Capital Output Ratio 1.24 2.14 0.01 19.78 Investment to Value-added Ratio 0.07 0.16 0 1.00 Investment to Capital Stock Ratio 0.05 0.15 0 3.63 Profits per Employee (lagged One period) 2,769 7,777 -132,521 118,250 Growth Rate of Output (%pa) -0.40 37.29 -99.86 99.99 Growth Rate of Employment (%pa) -1.02 26.90 -99.85 97.08 Span of Data 1992 2003 N=2307 Ln (investment) 9.01 3.22 -2.29 17.49 Investment to Capital Ratio 0.12 0.21 0.00 3.63 (conditional on any investment) N=1070 72