Annexes Annex 1: Productivity of Non-farm Enterprises 1. Labor productivity is defined as the ratio between total enterprise sales and total number of workers (both employees and household members). Capital productivity is defined as the ratio between total sales and the reported values of current physical capital stock, including all tools, equipment, buildings, land, and vehicles for the business. Business costs are defined as the sum of: purchase of goods for sale; transport; insurance; rent; interest; raw materials; other. 2. Total factor productivity is derived from a simple Cobb-Douglas production function in logarithmic form. Value added (calculated as output minus the value of business costs) is the dependent variable and the independent variables are total capital (current physical capital stock) and total employment (both employees and paid household members). Business costs are the sum of the purchase of goods for sale (inventory), fuel for generator and raw materials. Total factor productivity is calculated based on the Levinsohn and Petrin (2003) model with the value of the transport costs used as the proxy for unobserved productivity shocks. The production function is estimated for each sector, namely manufacturing and services. Estimations using a stochastic frontier model are used for robustness checks. In this analysis, we use the production function estimates to construct measures of plant level productivity. This productivity for a firm 𝑖 in the sector 𝑗 is calculated as: (𝑦𝑖𝑡 − �𝑙𝑗 𝑙𝑖𝑡 − �𝑘𝑗 𝑘𝑖𝑡 ) �𝑖𝑡 = exp� Where the parameter estimates �𝑙𝑗 and �𝑘𝑗 are taken from the production function estimates. Table 1: Production Functions – Manufacturing and Services Levinsohn and Petrin (2003) Technical Efficiency Manufacturing Services Manufacturing Services Log Labor 0.453*** 0.272*** 0.604*** 0.619*** (0.154) (0.106) (0.157) (0.109) Log Capital 0.322*** 0.110*** 0.439*** 0.119*** (0.075) (0. 023) (0. 021) (0. 006) Observations 1150 3876 1168 3971 Chi2 1.72 32.16 486 412 Source: Authors’ calculations based on GHS 2010/11 and 2012/13 Table 2: Productivity Analysis – Gender Performance Gap Log Log Log Log Log Log Labor Emp. Sales Labor Emp. Sales Log TFP prod growth growth Log TFP prod growth growth W1 & 2 W1 & 2 W2 only W2 only W1 & 2 W1 & 2 W2 only W2 only * * *** * Log Age 0.0150 0.164 0.0419 -0.0966 0.0374 0.281 0.0855 -0.131 (0.84) (2.25) (0.76) (-1.27) (2.44) (4.48) (1.79) (-2.10) None 0 0 0 0 0 0 0 0 (.) (.) (.) (.) (.) (.) (.) (.) Primary 0.00535 0.0337 0.0154 0.129 0.00530 0.0452 0.0182 0.103 (0.39) (0.58) (0.43) (1.58) (0.39) (0.79) (0.51) (1.30) ** *** ** *** 0.0585 0.363 0.0173 0.165 0.0560 0.372 0.0187 0.121 * * Secondary (3.45) (5.09) (0.36) (1.79) (3.35) (5.40) (0.39) (1.39) *** *** ** *** Tertiary 0.104 0.543 0.0156 0.202 0.103 0.558 0.0131 0.163 (3.35) (5.14) (0.31) (1.74) (3.27) (5.29) (0.25) (1.41) * Rural (vs Urban) -0.0246 -0.161 -0.0596 -0.0202 -0.0209 -0.153 -0.0626 -0.0201 (-1.22) (-2.07) (-1.70) (-0.39) (-1.03) (-1.96) (-1.83) (-0.39) * * Log HH head years of 0.00941 0.0392 0.0179 -0.0721 0.0125 0.0484 0.0151 -0.0536 education (1.56) (1.48) (1.31) (-2.03) (2.15) (1.96) (1.08) (-1.57) ** ** Log Distance to the 0.0134 0.0291 0.0372 0.0275 0.0146 0.0301 0.0364 0.0304 nearest major road (1.64) (0.88) (2.72) (1.06) (1.81) (0.92) (2.65) (1.16) Log Distance to a city -0.0153 -0.0582 0.00567 -0.0488 -0.0151 -0.0604 0.00633 -0.0455 with more than (-1.79) (-1.72) (0.30) (-1.93) (-1.80) (-1.82) (0.34) (-1.79) 20,000 inhabitants *** *** *** *** Female owner -0.170 -0.625 0.0190 -0.120 -0.161 -0.647 0.0352 -0.0589 (-7.95) (-8.49) (0.57) (-1.81) (-10.74) (-12.77) (1.15) (-1.29) Male owner * 0.0106 0.0572 -0.00166 -0.0266 Polygamous (0.38) (0.66) (-0.04) (-0.38) ** Male owner * -0.109 -0.254 0.0754 0.113 Separated (-3.13) (-1.75) (1.32) (0.91) *** *** Male owner * Single -0.109 -0.396 0.0262 -0.120 (-4.31) (-3.93) (0.49) (-0.85) * Female owner * -0.0293 -0.233 0.0360 0.0813 Married (-1.04) (-2.34) (0.79) (0.91) (polygamous) ** Female owner * 0.0991 0.150 -0.0193 -0.168 Separated (2.64) (0.90) (-0.28) (-1.11) *** *** Female owner * 0.171 0.541 -0.134 0.0934 Single (4.33) (3.82) (-1.22) (0.60) Log (Children below 5 0.0116 0.00466 -0.0398 -0.0268 years) (0.62) (0.08) (-1.24) (-0.44) Female*Log Children -0.0194 -0.0959 -0.00370 0.108 (-0.95) (-1.23) (-0.10) (1.40) *** *** - -0.273 -0.0111 0.0218 - -0.285 -0.0126 0.0282 ** ** Operating in the HH 0.0476 0.0498 * * premise (-3.94) (-5.84) (-0.52) (0.49) (-4.12) (-6.03) (-0.58) (0.63) *** *** *** *** Being registered 0.142 0.527 -0.0657 0.139 0.147 0.541 -0.0675 0.146 (3.86) (5.52) (-1.43) (1.26) (4.01) (5.69) (-1.45) (1.34) *** *** *** *** Dummy for NFE 0.0289 -0.504 0.380 0.00745 0.0305 -0.499 0.382 0.00509 employs non-HH (1.42) (-7.12) (5.93) (0.08) (1.49) (-6.94) (5.74) (0.05) member ** *** ** *** 0.0531 0.360 0.0518 0.356 * * Wave 2 vs Wave 1 (3.95) (6.27) (3.86) (6.19) *** *** * *** *** ** Constant 0.887 9.341 -0.0846 0.891 0.685 8.848 -0.255 0.985 (9.73) (20.44) (-0.33) (2.20) (8.84) (20.74) (-1.09) (2.65) Industry dummy         r2 0.163 0.239 0.123 0.0594 0.155 0.231 0.116 0.0553 N 5117 5168 1858 1901 5117 5168 1858 1901 * ** *** t statistics in parentheses: p < 0.05, p < 0.01, p < 0.001 Table 3: Productivity Analysis – Gender Performance Gap by Zone National Zone NC Zone NE Zone NW Zone SE Zone SS Zone SW *** *** *** *** * *** Female-owned -0.161 -0.226 -0.221 -0.170 -0.107 -0.0566 -0.185 (-10.74) (-5.74) (-5.08) (-6.14) (-1.81) (-2.03) (-4.36) r2 0.155 0.209 0.266 0.166 0.227 0.257 0.261 N 5117 903 852 1091 604 662 1005 * ** *** t statistics in parentheses: p < 0.05, p < 0.01, p < 0.001 Table 4: Productivity Analysis – Geographical Variables Log Labor Log Labor Log Labor Log TFP prod Log TFP prod Log TFP prod Industry       Year       Zone × ×   × × State × × × ×   2 𝑅 0.163 0.239 0.175 0.257 0.209 0.312 N 5117 5168 5117 5168 5117 5168 * ** *** t statistics in parentheses: p < 0.05, p < 0.01, p < 0.001 Table 5: Productivity Analysis – IC Constraints Log TFP Log Labor Log Emp. Log Sales Log TFP Log Labor prod growth growth prod Owner characteristics * Log Age 0.0162 0.127 0.0166 -0.181 -0.0752 -0.0605 (0.91) (1.83) (0.42) (-2.16) (-0.99) (-0.21) None 0 0 0 0 0 0 (.) (.) (.) (.) (.) (.) Primary -0.000752 -0.00909 0.0000989 0.0165 -0.0436 -0.0722 (-0.06) (-0.16) (0.00) (0.19) (-1.24) (-0.55) ** Secondary 0.0343 0.205 -0.0106 0.0779 -0.0316 -0.0198 (1.95) (2.96) (-0.31) (0.81) (-0.67) (-0.11) ** * Tertiary 0.0565 0.279 0.00363 0.0823 0.140 0.473 (1.88) (2.90) (0.07) (0.69) (2.00) (1.79) Log HH head Years of 0.00638 0.0112 0.00678 -0.0524 education (1.08) (0.44) (0.62) (-1.48) *** *** *** * Female * Married -0.170 -0.692 -0.0130 -0.106 -0.233 -0.514 (monogamous) (-8.14) (-9.92) (-0.41) (-1.51) (-3.70) (-2.17) * Male * Married 0.0267 0.137 -0.0191 -0.0199 -0.122 -0.0872 (polygamous) (0.99) (1.77) (-0.54) (-0.28) (-2.09) (-0.40) *** * Male * Separated -0.132 -0.334 0.110 0.0926 -0.202 -0.438 (-4.11) (-2.50) (1.71) (0.78) (-1.79) (-1.03) *** *** Male * Single -0.103 -0.387 0.0253 -0.158 -0.0143 0.513 (-4.19) (-3.96) (0.56) (-1.17) (-0.13) (1.21) *** Female * Married -0.0472 -0.307 0.0254 0.0497 0.0876 -0.302 (polygamous) (-1.71) (-3.35) (0.66) (0.53) (1.18) (-1.08) ** Female * Separated 0.106 0.153 -0.0573 -0.142 0.228 0.432 (3.19) (1.03) (-0.78) (-0.98) (1.64) (0.83) *** *** Female * Single 0.149 0.478 -0.0851 0.0740 0.0918 -0.0438 (3.97) (3.37) (-1.09) (0.48) (0.66) (-0.08) Log Children below 5 0.0152 0.0510 -0.0325 -0.0488 0.0473 0.0855 yrs old (0.81) (0.87) (-1.13) (-0.83) (1.02) (0.49) Female * Log -0.0167 -0.0890 0.000910 0.115 -0.0187 -0.0302 Children below 5 (-0.82) (-1.17) (0.03) (1.54) (-0.32) (-0.14) Household characteristics Log Distance to a 0.0108 0.0344 0.00895 0.0268 main road (1.50) (1.30) (0.84) (1.17) * ** Log Distance to a -0.0157 -0.0796 0.000266 -0.0202 main city (-2.15) (-2.66) (0.02) (-0.88) * *** Does household own 0.0308 0.232 0.0480 0.0314 0.00578 0.0858 a mobile telephone? (2.00) (3.96) (1.50) (0.55) (0.16) (0.65) * *** Does household own 0.0794 0.446 -0.0151 0.132 0.0634 0.112 a computer? (2.44) (4.30) (-0.36) (1.03) (0.89) (0.42) Firm characteristics *** *** Operates in the HH -0.0548 -0.270 0.0122 0.0168 -0.0158 0.00729 premises (-4.26) (-6.14) (0.64) (0.36) (-0.51) (0.06) *** *** Being registered 0.120 0.427 -0.0663 0.182 0.0866 0.284 (3.38) (4.62) (-1.56) (1.66) (1.60) (1.40) *** * ** NFE employs non-HH -0.0568 -0.815 0.127 0.390 -0.319 -1.402 member (-0.76) (-3.32) (0.59) (1.03) (-2.22) (-2.60) IC variables Rural -0.0337 -0.0805 -0.0164 -0.0451 0.0623 -0.226 (-1.37) (-0.92) (-0.33) (-0.56) (0.42) (-0.41) Mean Transport -0.118 0.135 417.6 -326.8 0.426 0.00241 costs (-0.62) (0.13) (1.18) (-0.50) (0.72) (0.00) *** *** Rural* Mean -1.125 -5.617 -299.8 -588.2 -0.858 -2.706 transport costs (-4.22) (-4.24) (-0.82) (-0.79) (-0.63) (-0.52) *** ** ** Mean Market 0.406 1.242 -1.153 0.497 coverage (3.97) (2.76) (-3.32) (0.74) * *** * Mean Number of -0.0364 -0.167 0.151 -0.422 -0.00898 0.223 services (-2.37) (-1.78) (5.90) (-2.22) (-0.05) (0.35) * * HH Uses a Generator 0.0910 0.173 0.111 0.0665 0.229 0.846 (1.54) (1.09) (0.98) (0.36) (2.45) (2.40) * Mean Share of -0.140 -0.643 -0.222 2.515 0.0331 -0.728 Electricity access (-0.90) (-1.04) (-1.06) (2.01) (0.16) (-0.91) * * Generator * Mean -0.0655 -0.00237 -0.120 -0.0353 -0.320 -1.231 share of electricity (-0.75) (-0.01) (-0.75) (-0.13) (-2.15) (-2.20) * * Mean Bank coverage 0.734 4.042 -0.477 1.459 -0.0260 -1.780 (2.02) (2.25) (-1.62) (1.60) (-0.02) (-0.30) * Mean Share of -0.162 -0.361 0.738 -0.897 -0.155 -0.227 people with secondary education (-1.28) (-0.80) (2.45) (-1.63) (-0.45) (-0.17) Employs non-HH 0.311 0.960 -0.819 2.460 -0.390 -0.862 members * Mean (1.92) (1.51) (-1.37) (1.67) (-1.21) (-0.71) Share of people with secondary education * Employs non-HH 0.0577 0.378 -0.798 0.635 0.190 0.138 members * Mean (0.60) (1.04) (-2.56) (1.14) (0.90) (0.17) Share of literate people * Mean Share of -0.00499 0.0135 0.852 -1.766 0.678 1.525 literate people (-0.03) (0.02) (1.55) (-1.53) (2.08) (1.24) * *** * *** Wave 2 versus Wave 0.0277 0.286 0.0463 0.406 1 (1.99) (4.51) (2.14) (5.00) ** *** ** Constant 0.384 9.113 -0.300 2.880 0.632 8.086 (2.63) (11.90) (-1.46) (2.92) (0.52) (1.77) Industry Dummy     NA NA State Dummy     NA NA Firm fixed effect × × × ×   r2 0.216 0.322 0.304 0.130 0.0883 0.155 N 5111 5162 1858 1901 5112 5177 * ** *** t statistics in parentheses: p < 0.05, p < 0.01, p < 0.001 Annex 2: Enterprise Survey in Nigeria (2014/15) 1. This Annex provides additional information on the data collected in Nigeria between April 2014 and February 2015 under an initiative of the World Bank. As part of its strategic goal of building a climate for investment, job creation, and sustainable growth, the World Bank has promoted improving business environments as a key strategy for development, which has led to a systematic effort in collecting enterprise data across countries. The Enterprise Surveys (ES) are an ongoing World Bank project in collecting both objective data based on firms’ experiences and enterprises’ perception of the environment in which they operate. This project was expanded with increased geographic coverage during fieldwork. It began as a nine-state exercise and ten more states were added for a total of 19 states. For this reason, only the first nine states covered contain the Innovation Module. The General Enterprise Survey 2. The core questionnaire was implemented to 2,676 firms from April 2014 to February 2015. The following table summarizes the composition of the sample. 19 states have been sampled: Lagos, Kano, Abuja, Anambra, Kaduna, Cross-River, Kebbi, Nasarawa, Ogun, Gombe, Katsina, Abia, Niger, Kwara, Enugu, Jigawa, Zamfara, Oyo and Sokoto. 3. Three levels of stratification were used in this country: industry, region, and size. Industry stratification was designed in the way that follows: For panel firms, the universe was stratified into manufacturing industries and two service sectors (retail and other services). For fresh firms, the universe was stratified into seven manufacturing industries (food & beverage, garments, fabricated metal products, non-metallic mineral products, furniture, publishing, and other manufacturing) and six service sectors (retail, wholesale, transport, hotels & restaurants, repair of motor vehicles, and other services). All firms are formal. Sector firms Size firms Food 182 small(<20) 1753 Garment and Textile 125 medium(20-99) 728 Publishing, Printing 136 large(100 and over) 195 Non Metallic Mineral Products 139 Fabricated Metal Products 130 Furniture 151 Manufacturing Panel 319 Other Manufacturing 236 Services of Motor Vehicles 163 Wholesale 120 Retail Panel 237 Retail 227 Hotels and restaurants 121 Other Services Panel 227 Other Services 71 Transport 92 Total Manufacturing Firms Manufacturing firms with Productivity Data Firms Abia 125 65 3 Abuja 152 78 29 Anambra 149 65 11 Cross river 136 63 12 Enugu 125 64 16 Kaduna 139 64 14 Kano 195 87 24 Lagos 281 152 47 Oyo 119 59 1 Gombe 126 62 21 Jigawa 123 65 33 Katsina 125 69 21 Kebbi 131 71 7 Kwara 124 76 15 Nasarawa 130 81 12 Niger 125 79 27 Ogun 130 86 7 Sokoto 118 70 2 Zamfara 123 71 4 Note: Manufacturing firms are classified based on ISIC codes collected during the interview not information based on screener interview. Firms are considered to have productivity data if they provided enough information to calculate total factor productivity (sales value of capital, number of workers, and value-added). See Appendix 1 for descriptions of these variables. The Innovation Module 4. The data used is the 2014 enterprise survey and the matched innovation module implemented in 2015 to a large sample of the firms that were interviewed in the core questionnaire. Specifically, the core questionnaire was implemented to 1401 firms in 2014. Of these, 892 firms were re-interviewed for the innovation module in 2015. 5. The following Table shows the composition of the sample. The survey was implemented only to firms in nine Southern States. The main aggregate sectors were wholesale and retail, with 39 percent of firms and manufacturing with 36 percent. Also there is a 14 percent of firms in hotels and restaurants. Overall, most of the firms are in the services sectors. The subsectors with more representation are retail, food and hotels and restaurants; which represent almost half of the sample. Sector firms % Size firms % Food 104 12.08 small(<20) 539 60.43 Textiles 13 1.51 medium(20-99) 274 30.72 Garments 69 8.01 large(100 and over) 79 8.86 Wood, Paper 13 1.51 Region firms % Publishing, Printing 65 7.55 Abia 81 9.08 Chemicals 17 1.97 Abuja 97 10.87 Plastics 14 1.63 Anambra 124 13.9 nonmetallic mineral products 34 3.95 Cross River 80 8.97 Basic metals, products 36 4.18 Enugu 106 11.88 Machinery 4 0.46 Kaduna 80 8.97 Electronics 3 0.35 Kano 94 10.54 Furniture 60 6.97 Lagos 164 18.39 Construction 12 1.39 Oyo 66 7.4 Motor vehicle services 66 7.67 Wholesale 45 5.23 Retail 215 24.97 Hotels & Restaurants 91 10.57 6. By size, 60 percent of firms are small (less than 20 workers), 31 percent medium (between 20 and 99 workers) and only nine percent are large firms (more than 100 workers). All firms are in the formal sector. Annex 3: Measures of Firm Performance Based on the Enterprise Surveys 1. The study focuses on several measures of firm productivity. These are calculated in a uniform way in all countries with comparable Enterprise Survey data from between 2006 and 2011. Although some surveys were conducted before 2006, these surveys did not use the same consistent sampling methodologies and questionnaires as the later surveys. Moreover, weights were not collected for these firms. We therefore restrict the analysis to the later, consistent surveys. A list of countries and years of surveys included in the analysis are listed below. 2. The Enterprise Surveys collect financial data in the local currency in the country being surveyed. To make cross-country comparisons, we need to convert the financial data into a common currency in a single year (to control inflation and exchange rate differences). To do this, all values are converted into 2009 US dollars (US$). For firm surveys conducted between 2006 and 2009 (that is, surveys with accounting data from between 2005 and 2008), data are inflated to 2009 values in local currency using the GDP deflator. For surveys converted after 2010, the values are deflated in a similar way. The values in 2009 local currency are then converted into US$, using 2009 exchange rates. Since most firms in the sample sell their products primarily in local markets, exchange rates have to be close to their equilibrium values in 2009 for these comparisons to be very accurate. If the exchange rate in a given country is over- or under-valued, the comparisons will under- or overstate the actual value of the performance measure for that country. 3. The individual measures are constructed in the following way. a. Value-added. Value-added is value of the goods and services that the firm produces less the cost of the raw materials (such as iron or wood) and intermediate inputs (such as engine parts or textiles) used to produce the output. Output is measured in local currency not in physical units. We subtract the cost of raw materials, intermediate inputs, electricity and fuel in local currency from output to get value- added. Firms report electricity and fuel costs separately from raw materials and intermediate inputs. Firms that do not report sales or raw materials are dropped for this measure. Electricity and fuel costs are treated as if they are zero for firms that do not report electricity or fuel costs (firms are not dropped). This is done because dropping firms that do not report electricity or fuel costs would have a significant impact on sample size and because electricity and fuel costs are small relative to sales and raw materials. b. Number of workers. The number of workers is the number of permanent and temporary full-time workers. Temporary workers are weighted by the average length of employment for these workers. So, for example, if the average length of employment for a temporary worker was six months, the weight for temporary workers would be ½. Data on part-time workers is not collected in most countries and so part-time workers are omitted for reasons of comparability. In practice, for countries with data on part-time workers, including these workers does not have a large effect on relative rankings. Firms that do not report permanent or temporary workers are dropped for measures that use workers (for example, value-added per worker). c. Capital. The survey includes two measures of capital. The first measure is the book value of capital. For firms that keep detailed financial accounts, this measure should be the value of capital taken from those accounts. For other firms, it will either be omitted or estimated by the manager. As described in the ICA manual, this variable is: - The net book value represents the actual cost of assets at the time they were acquired, including all costs incurred in making the assets usable (such as transportation and installation) minus depreciation accumulated since the date of purchase. (World Bank, 2007)1 The second measure is the sales value of capital. The manager is asked to estimate the value of the capital if sold in its current condition. Although this is probably closer to the true value of the capital, it has some shortcomings. In particular, when markets for capital equipment are thin, it might be difficult for the manager to give an accurate estimate. The implementation manual notes: - Ask the manager to estimate the market value if all of the equipment, land and buildings were sold on the open market. If the respondent states that there is no market, ask how much the respondent would be willing to pay for the capital, knowing what it can produce in its current condition. Estimate how much it would cost to buy machinery in the current market which is similar in terms of age and characteristics. (World Bank, 2007) d. Capital productivity. This is the ratio of value-added to capital. It can be constructed either using the book value or sales value of capital. When a firm produces a lot of output with little capital, capital productivity will be high. Firms that do not provide information on capital or enough information to calculate value-added are dropped. e. Total Factor Productivity. This measure of productivity takes both labor and capital use into account. f. Unit labor costs. This measure is labor costs as a percent of value-added. Although it is an approximation to true unit labor costs (measures output in dollars rather than as physical measure of production), it can be calculated using information from the Enterprise Survey. Unit labor costs are higher when higher labor costs are not fully reflected in higher productivity. g. Sales and employment growth. These measures are the average annual sales and employment growth from the three years before the survey to the year before the survey. For Nigeria, this is 2011 and 2013. Following Davis and Haltiwanger (1999),2 we calculate this by dividing the change in sales (or employment) between 1 World Bank. 2007. World Bank's Enterprise Surveys: Understanding the questionnaire. Washington DC: World Bank. 2 Davis, S. J., and Haltiwanger, J. 1999. On the driving force between cyclical movements in employment and job reallocation. American Economic Review 89(2): 1234-1258. the two years by the average value of initial and final sales (or employment). We do this to reduce influence of outliers. Since there were two full years between the two points in time, we calculate the annual average sales and employment rates with the following formula: 1 (Lt−1 −Lt−3 ) 2 ×1 ,. (Lt−1 +Lt−3 ) 2 Because data on temporary workers is only available for the year of the survey, the employment growth rate is the growth rate of permanent full-time workers only.