Lake Chad Regional Economic Memorandum  |  Development for Peace Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region Mathilde Lebrand (World Bank) 192 Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region 5.1 Introduction Infrastructure investments can support economic national and regional perspective, and with a particular development through both capital accumulation and focus on the Lake Chad area. structural transformation.340 Structural change, the movement of workers from lower to higher productivity All four countries, while being at different stages of employment, is seen as essential to growth in low-income development, face a similar challenge to generate countries. This paper investigates how investments in more adequate-quality jobs through economic electricity, internet, transport infrastructure, and their transformation. Employment in nonagricultural sectors interactions, affect economic development through is currently around 55 percent in Cameroon, 65 percent productivity gains and structural change in countries in Nigeria, and less than 30 percent in Niger and Chad around the Lake Chad. Roads provide access to markets, (World Bank data, see Figure A5.1 in the appendix). increasing both economic opportunities for local firms The share of employment in agriculture in Cameroon and competition from other locations. Electricity and and Nigeria has been declining since the 1990s, while internet allow for modern production technologies it has stagnated at high levels for Niger and Chad. and complement roads by boosting firm productivity. Overall, the pace of structural change has remained While the literature has studied specific infrastructure slower than expected in the region, hence the need to expansions as potential drivers of development, little has better understand the role of infrastructure as a driver of been done on the associated structural change and how structural transformation. the combinations of such investments will matter. Because the Lake Chad region is characterized by This paper focuses on the impact of infrastructure strong historical trade, ethnic, cultural, and political on economic development for the countries ties, which makes the areas within countries that around the Lake Chad area, an economically- and comprise the region economically interdependent, socially-integrated area in north-west Africa that we finally consider regional integration and cross- has development potential, but which has been border linkages when assessing the impact of further undermined by multiple and interrelated drivers of infrastructure investments. The region and its vicinities fragility, conflict, and violence. The Lake Chad region host some key cities—such as Maiduguri in Nigeria, comprises a set of administrative areas across Cameroon, Maroua and Kousseri in Cameroon, N’Djamena in Chad, Niger, and Nigeria that surround Lake Chad, with Chad, and Diffa in Niger—that could serve as a trade an estimated 17 million to 19 million people, who are hub that could potentially drive the regional economy. primarily involved in agriculture and fishing activities. However, connectivity across the borders (or between The region has one of the largest concentrations of these cities within the national boundaries) is poor due extreme poverty in Sub-Saharan Africa and the world to a lack of road infrastructure and volatile security and lags in human development outcomes and access situations that make trade or transportation of goods too to key public services. The paper analyzes the impact of costly. The paper complements the economic analysis of infrastructure in Cameroon, Chad and Nigeria, from a key infrastructure investments with additional border investments. 340 There are two approaches for explaining economic growth (McMillan et al., 2017). The first one assumes that the accumulation of skills, capital, and broad institutional capabilities are needed to generate sustained productivity growth. The second approach assumes a dual economy where long-run growth is driven by the flow of resources to the modern economic activities that operate at higher levels of productivity. 5.1 Introduction 193 Lake Chad Regional Economic Memorandum  |  Development for Peace The paper is divided in two parts. The first part uses work for countries in the Horn of Africa (Herrera Dappe reduced-form analysis to quantify the impacts of past and Lebrand, 2021). investments in electricity, internet, and road infrastructure on the sectoral structure of employment in Cameroon, Our paper is related to a number of different strands Nigeria, and Chad. The second part uses a spatial general of research. First, our work contributes to research on equilibrium model, based on Moneke (2020), to assess the the different impacts of infrastructure. Several papers aggregate and spatial impacts of planned infrastructure have examined the impact of infrastructure investment investments in the region. Reduced-form results capture on sectoral employment at the micro-level (Adukia et al., the localized effects in the areas that have been affected, 2020; Asher and Novosad, 2018; Gertler et al., 2016). In but do not capture the general-equilibrium effects and the case of roads, lower transport costs empower women spillovers due to the network nature of infrastructure by opening labor market opportunities and increase their such as roads. The general-equilibrium model captures employment in the non-agricultural sector (Gertler et the spillover effects that a localized investment has on al., 2016). However most papers have considered the the rest of the country and all the countries in the Lake gains from energy, transport and digital investments Chad region and generates welfare estimates for the entire in isolation or bundled in a unique infrastructure region and all its subregions. index. The aggregate impact of infrastructure has been measured through the elasticity of output with respect to We first provide evidence on how past investments a synthetic infrastructure index, which includes transport and their combinations mattered for structural along with electricity and telecommunications (Calderon transformation in countries around the Lake Chad, et al., 2015). More recently, Moneke (2020) shows that which includes Cameroon, Nigeria, Niger, and interactions of transport and electricity investments are Chad. Facing endogenous infrastructure investments complementary and gave rise to large effects on economic with respect to sectoral employment outcomes across development in the context of Ethiopia. He finds starkly time and space, we use several instrumental variables to different patterns of big push infrastructure on sectoral overcome these endogeneity concerns. We then study the employment compared to only road investments: roads average and heterogenous effects to understand whether alone cause services employment to increase at the expense leading or lagging regions benefit differently from such of agriculture and, especially, manufacturing employment. investments. As the pace of structural change remains In contrast, the interaction of roads and electrification slow in the region, we provide counterfactual evidence to causes a strong reversal in manufacturing employment. which extent a push for more regional integration through Our paper is similar but includes investments in digital the expansion of transport and trade infrastructure infrastructure, and covers more countries. would support economic development and structural transformation. Finally, our paper contributes to the long literature using quantitative spatial general equilibrium The objective of this work is to extend our models to provide counterfactuals for infrastructure understanding on the impact of infrastructure investments. We look at the general equilibrium effects investments across several neighboring countries of road and border investments via changes in trade costs in Africa. The novelties of this work are to assess the that lead labor to reallocate across locations such as Allen interactions between different infrastructures and how and Arkolakis (2014) and Redding (2016). Bustos et it affects the sectoral structure of employment at the al. (2016) and Fried and Lagakos (2020) study general district level, and to assess the impacts of several planned equilibrium implications of electrification via its effect on transport investments and trade facilitation measures productivity. Michaels et al. (2011) look at changes in in neighboring countries that are at different stages of sectoral employment as outcome of interest that captures development. A companion paper undertakes similar the underlying infrastructure-induced effects. Several 194 5.1 Introduction Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region papers provide policy counterfactuals for future road and border infrastructure improvements for the Belt and Road Initiative (Lall and Lebrand, 2020; Bird et al., 2020), in Bangladesh (Herrera Dappe and Lebrand, 2019), and between Bangladesh and India (Herrera Dappe et al., 2021). The paper is structured as follows. Section 2 presents the data. Section 3 presents the empirical strategy and results. Section 4 develops a spatial general-equilibrium model to produce counterfactuals for more regional integration. Section 5 concludes. 5.1 Introduction 195 Lake Chad Regional Economic Memorandum  |  Development for Peace 5.2 Data In this paper, we complement household survey networks around the year 2007 based on surveys and data that have been geolocalized with new spatial governments sources. We complement each network with  n order to study links between access infrastructure data i data from recent government surveys for Cameroon and to infrastructure and structure of the local economies, Chad, and from the work of Ali et al. (2015) for Nigeria. and the complementarities between the different types of In the latest, geographic information system road network infrastructure. data are combined with road survey data from the Nigeria Federal Roads Maintenance Agency (FERMA) and the World Bank’s Fadama project.341 Panels of roads from 5.2.1 Sources of data the same source are rare. Related works include a similar paper applied to the Horn of Africa (Herrera Dappe and Infrastructure. We start by collecting new information Lebrand, 2021) and Moneke (2020) whose focus on ‘all- on road network expansions, access to the electricity weather’ (i.e. gravel, asphalt or bitumen surface) roads is network, and access to the internet fiber backbone. Table closer to ours. 5.1 summarizes the data sources and years of coverage. Second, we use two methods to map access to the First, we collect geospatial maps of road expansion electricity network, nighttime lights as a proxy for using governmental sources but also previously- access to electricity and maps of the power transmission harmonized collections of road networks  (Foster and grids. Nighttime lights are available for most years Briceno-Garmendia, 2010; Jedwab and Storeygard, and locations but convey imperfect information on 2020). The quality of the network and associated features, electrification. Historical maps of electricity grids are and the frequency of updates vary across countries. For more difficult to find and use in a consistent way. First, all three countries, we first rely on data from Foster and we use satellite images of annualized nighttime lights Briceno-Garmendia (2010) that cover all African roads and population rasters to calculate the percentage of Table 5.1: Summary of Infrastructure data Infrastructure Country Year Source Nigeria 1991 Jedwab and Storeygard (2020) Nigeria 2009 Jedwab and Storeygard (2020) around 2013 Ali et al. (2015) Roads Cameroon 2009 Foster and Briceno-Garmendia (2010) 2018 Road authorities Chad 2009 Foster and Briceno-Garmendia (2010) 2018 Road authorities Electricity All vary across countries Nighttime lights/Population raster Electricity grid All around 2006 Foster and Briceno-Garmendia (2010) All most recent gridfinder.org and Arderne et al. (2020) Internet All 2009–2019 Africa Bandwidth Maps 2009–19 341 To "ground truth" and take advantage of first hand local knowledge, government offices across Nigeria were surveyed about the conditions of specific road segments near them. 196 5.2 Data Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region electrified population, meaning population in settlements from the Africa Infrastructure Country Diagnostic that produce some lights at night. We compare the results (AICD) which collected primary data covering network for two metrics: a dummy that is equal to one if at least service infrastructures from 2001 to 2006 in 24 selected 50 percent of the population has access to electricity African countries. To complement these data, we rely from as defined previously, and the percentage of electrified on a recent effort by the World Bank, Facebook and other population per district. Sensitivity checks show that the institutions (KTH, ESMAP, WRI and the University of choice of the threshold does not significantly change the Massachusetts Amherst) to use remote sensing, machine results.342 Such methods have been used before to estimate learning and big data to map connected populations and electricity access in remote areas and guide grid extension the systems that support them. They create an algorithm programs.343 They assume that locations that emit lights at for estimating the location of existing medium-voltage night are settlements that have electricity access, and that infrastructure based on nighttime lights and the location their electricity is most likely supplied from an electrical of roads assuming that medium voltage lines are more grid. It also assumes that small off-grid systems do not likely to follow (or be followed by) main roads.345 Figure emit enough light to be captured by satellites, but larger A5.21 in the appendix shows the grid for the Lake Chad isolated power networks certainly do. We cross-check the using the 2009 grid from Foster and Briceno-Garmendia numbers that we get with country estimates of electrified (2010) and using the most recent grid. population over years from the World Bank.344 Figure A5.2 in the appendix shows the results for the percentage Internet infrastructure is proxied by access to the fiber of electrified population for all countries. broadband backbone network. We obtained data for all Africa for the period 2009–2020 with the exact location We also collect information on transmission of fiber nodes along the backbone network from Africa grids based on past efforts to harmonize data for Bandwidth Maps. We construct a proxy for access to the infrastructure from primary sources and recent fiber backbone which is equal to one if there is a node mapping strategies to infer the electricity grids based of the backbone in the location of interest. Each node on satellite data. For past data, we use electricity grids has a year attribute which allows us to build a panel for Figure 5.1: Route-Kilometers of Terrestrial Transmission Network, Africa 2009–2019 1.4M– 1.2M– 1.0M– 800K– 600K– 400K– 200K– 0– 9 09 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -0 -1 -1 -1 -1 -1 -1 -1 -1 -1 - Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 J Fiber operational J Fiber under construction J Fiber planned J Fiber proposed Source: http://www.africabandwidthmaps.com/ 342 There is a large concentration of data points around 0 and 100 percent, which explains that the choice of the threshold, being the mid-point of the average, does not have a strong impact on the analysis. 343 An example of mapping rural electrification based on nighttime lights can be found at http://india.nightlights.io/ 344 The World Bank reports access to electricity (percent of population) for most countries for a long time period at https://data.worldbank.org/indicator/ EG.ELC.ACCS.ZS. 345 More details can be found on the blog https://blogs.worldbank.org/energy/using-night-lights-map-electricalgrid-infrastructure and in the paper Arderne et al. (2020). 5.2 Data 197 Lake Chad Regional Economic Memorandum  |  Development for Peace access to the backbone. We assume that access before round. The DHS-provided GPS coordinates for EAs 2008 is null everywhere, which is supported by World locations are not perfectly reliable due to the common Bank data on access to internet which reports that less random displacement applied to GPS coordinates for than 4 percent of individuals in Sub-Saharan countries anonymity.347 We aggregate EAs per geographic location. (including high-income countries) have access to internet in 2008. Figure 5.1 shows the growth of the network for Table 5.2: Survey data the whole of Africa, and Figures A5.8, A5.12, A5.18 show Data Country Year Source the geographic evolution per country. However, internet 1990, 2003, access remains unequal across countries as some countries Nigeria 2008, 2013, have a broad internet coverage while others are still 2018 Household DHS lagging. We cross-check our numbers using the World surveys 1991, 2004, Cameroon 2011, 2018 Bank indicators reporting the percentage of individuals using the internet.346 Chad 2014 Table 5.3: Additional district-level data Employment. We are interested in structural transformation, which we interpret in line with the Population GHSL Herrendorf et al., 2014) as changes in sectoral literature ( Land ESA Land Cover employment. Following Moneke (2020), we require Distance to the coast GSHHG information on sectoral employment shares, which we Distance to the border Aiddata database derive from the Demographic & Health Survey (DHS). Access to a city The DHS produces harmonized survey data with GPS >50,000 inhabitants The Malaria Atlas Project coordinates available for most surveys and is available for Land Processes several rounds per country. The DHS is a repeated cross- Temperature Distributed Active Archive Center section of enumeration areas (EA), with approximately Elevation CGIAR-CSI 20 to 30 households enumerated per EA. In total, five rounds in Nigeria [1990, 2003, 2008, 2013, 2018], four rounds in Cameroon [1991, 2004, 2011, 2018], and one Usages of infrastructure. In addition of access to round in Chad, summarized in Table 5.2. Unfortunately infrastructure, we consider how usages linked to roads, we are not able to cover Niger. electricity and ICT infrastructure impact the structure of employment. We include variables from the DHS We use DHS data for which we have access to the surveys that cover access to electricity as reported by the occupation of the individuals as well as a proxy for their households, ownership of cars and motorcycles, ownership location. In order to construct the shares of employment of land phone and mobile phone, and use of internet. We per sector, we use respondents’ answer to questions about aggregate the answers at the subnational level of interest their current occupation. We first compute the shares of as percentage of individuals that have access to electricity, non-working individuals, and then we group the working own a car, a motorcycle, a land phone, a mobile phone, individuals into three sectors, agriculture, manufacturing or use internet. The analysis of usage complements the and services. We aggregate individual responses to the analysis of infrastructure investments. Not all variables are enumeration area and then generate an unbalanced panel available for the whole period so we restrict our analysis of districts that have at least one EA during a survey to the period 2008–2018. 346 The World Bank reports access to internet (percent of population) for most countries for a long time period at https://data.worldbank.org/indicator/WeT. NET.USER.ZS. based on International Telecommunication Union, World Telecommunication/ICT Development Report and database. 347 DHS coordinates of rural (urban) EAs are randomly displaced within a 0–10km (0–5km) radius. 198 5.2 Data Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region District characteristics. We use additional data to control In Cameroon, 60 percent of the communes and for district heterogeneity as described in Table 5.3. First, 80 percent of the population are connected to a paved we use population from satellite data from GHSL348, road. The coverage falls when restricting access to fair or land categories from ESA land cover349, distance from good paved roads. Access to electricity and internet covers the coast from GSHHG350, distance to the border351, a small number of communes but the most populated access to a city larger than 50,000 inhabitants from the ones as the percentage of population is almost twice the Malaria Atlas Project,352 temperature from Land Processes percentage of communes. Between 2003 and 2018, the Distributed Active Archive Center353, and elevation from number of communes covered has more than double CGIAR-CSI354. but the percentage of population covered has increased by only 7 percent percentage points from 41 percent. The additional communes that have received electricity 5.2.2 Access to infrastructure coverage over the last two decades are much less populated. We compare access to infrastructure for paved roads, Access to infrastructure in Chad is very limited electricity, and internet broadband for Nigeria, compared to other countries. In 2014, only 2.6 percent Cameroon and Chad after 2000. Figures 5.2, 5.3, of communes had access to a paved road, 3 percent to and 5.4 report summary statistics of the infrastructure the internet broadband, and 6 percent to the electricity variables used in the next sections at the country level, network. The covered communes are the most populated and Figure 5.5 focuses on the smaller area around the ones, as 20 percent of the population has access to lake Chad. electricity and 15 percent of the population has access to the internet broadband. Recent improvements of the Nigeria has the highest level of access to paved roads paved road network since 2014 as shown in Figure A5.17 and electricity with more than 90 percent of the show that the percentage of communes and population districts and population having access to a paved road having access to a paved road has largely increased. in 2018. While access to paved roads has barely changed since 2000, access to electricity has increased significantly The Lake Chad area, which includes The Extreme from 35 percent to 56 percent of districts having access North region in Cameroon, the regions of Kanem, to electricity between 2003 and 2018.355 23 percent of Lac, Hadjer-Lamis, and Chari-Baguirmi in Chad, districts are connected to the fiber network as defined by and the regions of Borno, Yobe and Adamawa in the presence of a node from the fiber backbone. Nigeria as depicted on Figure A5.20, is characterized by a limited access to infrastructure. Only 30 percent 348 GHSL: Population count from the Global Human Settlement Layer. Based on population data from Gridded Population of the World v4.10 polygons, distributed across cells using the Global Human Settlement Layer global layer. Source data provided in 9 arc-second (250m) grid cells. 349 ESA land cover: Defourny, P. (2017): ESA Land Cover Climate Change Initiative (Land Cover cci): Land Cover Maps, v2.0.7. Centre for Environmental Data Analysis, 7/2017. 350 Distance to coast (on land only), measured in meters. Derived using World Vector Shorelines. Wessel, P., and W. H. F. Smith, A Global Self-consistent, Hierarchical, High-resolution Shoreline Database, J. Geophys. Res., 101, B4, pp. 8741-8743, 1996. 351 Distance to country borders, measured in meters. Derived using GADM 2.8 ADM0 (Country) boundaries. 352 Incorporates data from Open Street Map (OSM) data and the Google roads database. D.J. Weiss, A. Nelson, H.S. Gibson, W. Temperley, S. Peedell, A. Lieber, M. Hancher, E. Poyart, S. Belchior, N. Fullman, B. Mappin, U. Dalrymple, J. Rozier, T.C.D. Lucas, R.E. Howes, L.S. Tusting, S.Y. Kang, E. Cameron, D. Bisanzio, K.E. Battle, S. Bhatt, and P.W. Gething. A global map of travel time to cities to assess inequalities in accessibility in 2015. (2018). Nature. doi:10.1038/nature25181. 353 Yearly daytime land surface temperature. Wan, Z., Hook, S., Hulley, G. (2015). MOD11C3 MODWeS/Terra Land Surface Temperature/Emissivity Monthly L3 Global 0.05Deg CMG V006 [Data set]. NASA EOSDWeS LP DAAC. doi: 10.5067/MODWeS/MOD11C3.006. 354 Global elevation (in meters) from Shuttle Radar Topography Mission (SRTM) dataset (v4.1) at 500m resolution. Jarvis A., H. We. Reuter, A. Nelson, E. Guevara, 2008, Hole-filled seamless SRTM data V4, International Centre for Tropical Agriculture (CIAT), available from http://srtm.csi.cgiar.org. 355 A district is classified as electrified when at least 50 percent of its population has access to electricity as monitored with nighttime lights. 5.2 Data 199 Lake Chad Regional Economic Memorandum  |  Development for Peace Figure 5.2: Access to infrastructure in Nigeria a. At the district level b. At the population level Percentage of districts with access to infrastructure Percentage of population with access to infrastructure 100– 100– 94.8 95.3 86.6 86.5 88.9 88.6 80– 80– 79.9 76.3 60– 60– 61.5 56.3 40– 40– 42.9 35.5 28.4 20– 23.2 20– 0.0 0.0 0– 0– Paved road Good paved road Electricity Internet Paved road Good paved road Electricity Internet J 2003 J 2018 J 2003 J 2018 Note: Authors’ calculations using data sources listed in previous section. Good paved roads include roads of fair or good condition. The left graph shows the percentage of districts (admin 2 for Nigeria), the right graph the percentage of population using the 2015 district population. Figure 5.3: Access to infrastructure in Cameroon a. At the district level b. At the population level Percentage of districts with access to infrastructure Percentage of population with access to infrastructure 100– 100– 80– 80– 82.0 75.8 60– 61.9 60– 57.5 51.7 53.1 47.7 48.9 40– 40– 41.3 36.5 32.5 29.8 20– 22.8 20– 9.3 0.0 0.0 0– 0– Paved road Good paved road Electricity Internet Paved road Good paved road Electricity Internet J 2004 J 2018 J 2004 J 2018 Note: Authors’ calculations using data sources listed in previous section. The left graph shows the percentage of districts (admin 3 for Cameroon), the right graph the percentage of population using the 2015 district population. Figure 5.4: Access to infrastructure in Chad (2014) of the 80 locations—districts in Nigeria, communes in Percentage Cameroon and Chad—have a paved road, 16 percent 20– 20.0 access to electricity, and 10 percent access to internet. Only half of the population has access to a paved road, 15– 14.9 20 percent to electricity, and 30 percent to the internet broadband. 10– 6.3 5– 2.9 2.6 4.3 0– Electricity Internet Paved road J Districts J Population Note: Authors’ calculations using data sources listed in previous section. 200 5.2 Data Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region Figure 5.5: Access to infrastructure in the Lake Chad region a. At the district level b. At the population level Percentage of districts with access to infrastructure Percentage of population with access to infrastructure 100– 100– 80– 80– 60– 60– 60.3 40– 40– 32.5 29.8 20– 20– 21.0 16.3 10.0 0– 0– Paved road Electricity Internet Paved road Electricity Internet Note: Authors’ calculations using data sources listed in previous section. The Lake Chad area is defined as the area depicted in Figure A5.20 around the lake. The left graph shows the percentage of districts (admin 2 for Nigeria, admin 3 for Chad and Cameroon), the right graph the percentage of population using the 2015 district population. 5.2 Data 201 Lake Chad Regional Economic Memorandum  |  Development for Peace 5.3 Empirical Strategy and Results 5.3.1 Ordinary Least Squares We run two other specifications. First we include a specification to study labor participation where the The first empirical strategy uses panel ordinary least dependent variable is not the normalized sectoral share squares (OLS) regressions,  which includes year and but the share of non-working individuals. Second, country fixed effects and a battery of initial district-level we use variables of usage rather than presence of hard controls. The OLS specification is: infrastructure to compare the results. Sectori,c,t = α + βR Paced Roadi,c,t + βE Electricityi,c,t Employmenti,c,t = α + βE Electricityi,c,t + βC Cari,c,t + βI Interneti,c,t + ΥRE Paved Roadi,c,t * + βM Motorcyclei,c,t + βLP LandPhonei,c,t + Electricityi,c,t + ΥRI Paved Roadi,c,t * Interneti,c,t βMP MobilePhonei,c,t + βI Useofinterneti,c,t + + Controlsi,c,t + FE + εi,c,t  (5.3.1) Controlsi,c,t + FE + εi,c,t  (5.3.2) Sectori,c,t is the share of employment in agriculture,  ith Electricityi,c,t the percentage of respondents in w manufacturing or services for district i in country c, at location i with access to electricity as indicated by year t. The shares are normalized and equal to one. Paved the household, Cari,c,t the percentage owning a car, Roadi,c,t is a dummy variable that takes a value of one if Motorcyclei,c,t the percentage owning a motorcycle, location i in country c contains a paved road at year t. LandPhonei,c,t the percentage owning a land phone line, Electricityi,c,t is a dummy variable that takes a value of Mobilephonei,c,t the percentage owning a mobile phone, one if location i in country c has more than 50 percent of and Useofinterneti,c,t the percentage using internet. its population with lights at night at year t. Interneti,c,t Employmenti,c,t can be either the sectoral share or the is a dummy variable that takes a value of one if location labor force participation share. i in country c has a node on the internet backbone fiber network at year t. Paved Roadi,c,t * Electricityi,c,t captures There are several identification challenges that we the interaction of the road and electricity infrastructure, have identified. Infrastructure investments are likely and Paved Roadi,c,t * Interneti,c,t the interaction of the endogenously allocated with respect to the outcomes of road and internet infrastructures. We add interaction interest. Given the high cost of infrastructure investments, effects between the dummies to better understand the conscious allocation decisions are to be expected, for complementarities between infrastructures. We do not example by targeting high growth potential locations or include an interaction effect for electricity and internet as politically demanded locations. Finally, measurement error access to internet is assumed to rely on electricity access. in the right-hand side variables may lead to attenuation Controlsi,c,t represents the additional location-specific bias, for example due to inaccurate timing information of controls, which include initial district population, access infrastructure expansion or imprecise historic road and to a main city, land size, distance to the coast, distance grid maps. Measurement errors which are expected to be to the border, access to a city of more than 50,000 large in this case, lead to an OLS estimate biased towards inhabitants, temperature, and elevation. FE is the year zero. In the next section, we present results from OLS and country fixed-effects. The coefficients β capture the regressions, keeping in mind that they do not represent correlation between access to a type of infrastructure causality. We explore an instrumental variable (IV) on the different sectoral employment shares, while the strategy in section 3.2 using instruments for paved roads coefficients Υ capture the infrastructure interaction terms. and electricity access only. 202 5.3 Empirical Strategy and Results Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region 5.3.1.1 OLS Results All countries. Access to electricity is associated with a transformation away from agriculture in the Lake Chad This section reports results for the unbalanced region. Table 5.4 reports the results of pooling together panel of districts that include at least one EA. We data from Nigeria, Cameroon, and Chad. Having access first estimate local average associations of the three to electricity at the district level is associated with a infrastructure investments—roads, electricity, and 20.8 percentage-point reduction in the employment internet—and the interaction between these investments share of agriculture, a 8.5 percentage-point increase on sectoral employment at the district-year level. Then in the employment share of manufacturing, and a we analyze the within-country heterogeneity in structural 12 percentage-point increase in the employment transformation across districts. share of services. Access to internet is associated with a decrease in the share of employment in agriculture and an increase in the share of services, but the coefficients 5.3.1.2 Average Effects are not significant. Access to a paved road is significantly associated with a transformation away from agriculture. We start with a regression that includes all countries Having access to a paved road at the district or commune from the Lake Chad. Then, we compare the results level is associated with 4 percentage-point increase in the by country revealing some heterogeneity in responses. employment share of manufacturing, and a 2 percentage- Throughout, standard errors are clustered at the district- point increase in the employment share of services. level, which is the level of the treatment. Agriculture, Table A5.2 in the appendix presents the results using the Manufacturing and Services represent the sector shares dummy variable that captures electricity access based on that add up to one. The ’Non-Working’ column represent the grid expansion for comparison. the share of active population reporting not working at the moment of the interview. Table 5.4: Nigeria, Cameroon, Chad Agriculture Manufacturing Services Not working -0.0580** 0.0387** 0.0207+ 0.0422** Paved road (-3.43) (4.96) (1.70) (5.82) -0.0791 0.00545 0.0674 0.0134 Internet (-1.58) (0.34) (1.63) (0.68) -0.208** 0.0849** 0.126** 0.0699** Electricity (>50p) (-6.68) (4.70) (5.70) (5.21) 0.0288 -0.0122 -0.00946 -0.0285 Road + Internet (0.57) (-0.73) (-0.22) (-1.39) -0.00994 -0.0126 0.0199 -0.0811** Road + Electricity (-0.32) (-0.68) (0.89) (-5.88) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.403 0.346 0.420 0.157 N. of observations 3,041 3,041 3,041 3,041 Notes: t statistics in parentheses; + p < 0.10; * p < 0.05; ** p < 0.01 5.3 Empirical Strategy and Results 203 Lake Chad Regional Economic Memorandum  |  Development for Peace Table 5.5 shows the results of the analysis when have strong impacts, with a 10 percentage-point increase including usages of infrastructure instead of the in the number of respondents using a mobile phone presence of infrastructure in the district. Both analysis (land phone) associated with a 1.9 (3.3) percentage-point are complementary. The presence of infrastructure does decrease in agricultural employment shares. not ensure its full usage by households and represents therefore a low bound of the impacts of infrastructure.  verall the impacts on employment and labor market O The usage represents the higher bound of the estimate of participation as reported on the last column ’Not impacts when individuals acquire the goods or services working’ are mixed. to make the best of these infrastructure. Estimated coefficients are therefore higher when considering Around the Lake Chad. We restrict our analysis to the the usage rather than the presence of infrastructure. regions neighboring the Lake Chad at the intersection of Usage variables in Table 5.5 represent the percentage of Chad, Cameroon and Nigeria as shown in Appendix on respondents that own a certain good or use the services Figure A5.20. These regions are poorer, less developed allowed by certain infrastructures. and more prone to conflicts. Table A5.5 shows that infrastructure investments there are associated with much A 10 percentage-point increase in the number of larger effects on sectoral shares, especially for electricity respondents using electricity is associated with and paved roads. Access to electricity and a paved road a 3.6 percentage-point decrease in the share of is associated with a large transformation away from agricultural employment,  a 1 percentage point agriculture in the districts directly located around the increase in manufacturing and a 2.5 percentage-point Lake. Having access to a paved road at the district level increase in services. Having a car, and to a lower extent, is associated with a 13 percentage-point reduction in the having a motorcycle has a large impact on structural employment share of agriculture, an 8 percentage-point transformation. Both the use of mobile and land phones increase in the employment share of manufacturing, and Table 5.5: Nigeria, Cameroon, Chad: the usage perspective, 2008–2018 Agriculture Manufacturing Services Not working -0.359** 0.109** 0.252** -0.0257** Electricity (share) (-23.64) (15.44) (19.92) (-3.47) -0.0177 -0.00197 0.0286 0.0140 Motorcycle (share) (-0.81) (-0.17) (1.46) (1.11) -0.362** -0.0517* 0.411** -0.0498* Car (share) (-8.19) (-2.11) (11.03) (-2.08) Use of internet -0.0393 -0.00883 0.0723+ -0.117** (share) (-0.80) (-0.31) (1.67) (-3.79) Mobile phone -0.189** 0.0909** 0.0942** 0.0667** (share) (-6.49) (6.35) (3.69) (3.08) -0.334** -0.000578 0.345** 0.179** Land phone (share) (-3.22) (-0.01) (3.55) (3.11) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.605 0.358 0.561 0.157 N. of observations 2,369 2,369 2,369 2,369 Notes: t statistics in parentheses; + p < 0.10; * p < 0.05; ** p < 0.01. All explanatory variables are shares of population between 0 and 1. 204 5.3 Empirical Strategy and Results Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region a 5 percentage-point increase in the employment share By country. In Cameroon, access to electricity is associated of services. Access to electricity is also associated with with a significant reduction in agriculture employment large impacts, with a decrease of 14 percentage points for and increase in employment in manufacturing and services agriculture employment. (Table 5.7). Having access to electricity at the district level is associated with a 26 percentage-point reduction in the agriculture employment share, a 21 percentage-point Table 5.6: Around Lake Chad Agriculture Manufacturing Services Not working -0.124* 0.0753* 0.0443 -0.0147 Paved road (-2.05) (2.39) (1.11) (-0.74) -0.0583 0.0551* 0.0269 0.0463* Internet (-1.02) (2.03) (0.65) (2.16) -0.240** 0.0450 0.177** 0.0247 Electricity (>50p) (-3.13) (1.66) (2.75) (1.00) -0.136 -0.0598 0.192+ -0.00410 Road + Internet (-1.22) (-1.31) (1.97) (-0.13) -0.154 0.143** 0.0348 0.0313 Road + Electricity (-1.32) (2.74) (0.42) (0.73) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.415 0.468 0.540 0.638 N. of observations 123 123 123 123 Notes: t statistics in parentheses; + p < 0.10; * p < 0.05; ** p < 0.01. All explanatory variables are shares of population between 0 and 1. Table 5.7: Cameroon Agriculture Manufacturing Services Not working -0.117** 0.0584** 0.0583** 0.0290** Paved road (-4.77) (4.68) (3.66) (2.59) -0.134* 0.0112 0.122** 0.0201 Internet (-2.46) (0.52) (2.63) (0.60) -0.261** 0.212** 0.0489 0.0538+ Electricity (>50p) (-3.21) (4.53) (1.16) (1.95) 0.0810 -0.0125 -0.0685 -0.0269 Road + Internet (1.37) (-0.53) (-1.38) (-0.75) 0.0135 -0.114* 0.100. -0.00549 Road + Electricity (0.16) (-2.41) (2.25) (-0.19) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.410 0.379 0.462 0.103 N. of observations 661 661 661 661 Notes: t statistics in parentheses; + p < 0.10; * p < 0.05; ** p < 0.01. All explanatory variables are shares of population between 0 and 1. 5.3 Empirical Strategy and Results 205 Lake Chad Regional Economic Memorandum  |  Development for Peace increase in the manufacturing employment share and a 5.3.2 Instrumental Variables 5 percentage-point increase in the employment share of services. Access to a paved road is also associated with In this section we use an instrumental variables a significant impact, from agriculture with a reduction identification strategy to deal with the potential of 12 percentage points to manufacturing and services. endogeneity in the placement of the infrastructure. We Finally access to internet is associated with a large impact instrument both roads and electricity and the interaction too, mostly towards the services sector. In terms of terms. combined investments, the combination of both paved roads and electricity seems to support employment in services at the expense of the manufacturing sector. 5.3.2.1 IV strategy Reduced-form estimates for the usage of infrastructure are reported in Appendix in Table A5.6. We instrument access to the national electricity grid and access to a paved road. Regarding electrification, Table 5.8 reports the results of the regression for the instrumental variable relies on four assumptions. Nigeria only. Similar to Table 5.4, access to electricity Electricity generation must be connected to demand, is associated to a large impact. The combined accesses which comes mostly from the main cities. The sources to internet and paved roads as well as paved roads and of energy generation that are identified are the main electricity are associated with a large reduction in the sources of electricity generation. Third the locations share of agriculture employment, and an increase in the of the supply sources are exogenous to economic services sector. Reduced-form estimates for the usage of geographic development. Finally, the locations between infrastructure are reported in Appendix in Tables A5.7. the generation sources and the main demand centers are Table A5.8 reports the results of the regression for Tchad. more likely to be electrified. Table 5.8: Nigeria Agriculture Manufacturing Services Not working 0.0137 0.00325 -0.0170 0.0113 Paved road (0.53) (0.31) (-0.82) (1.01) 0.0122 0.00139 -0.0136 0.0157 Internet (0.16) (0.07) (-0.19) (0.58) -0.151** 0.0319 0.119** 0.00496 Electricity (>50p) (-3.51) (1.60) (3.53) (0.29) -0.0757 -0.00170 0.0773 -0.0378 Road + Internet (-0.98) (-0.09) (1.08) (-1.36) -0.0431 0.0232 0.0199 -0.0113 Road + Electricity (-1.00) (1.14) (0.59) (-0.63) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.357 0.321 0.282 0.161 N. of observations 2,137 2,137 2,137 2,137 Notes: t statistics in parentheses; + p < 0.10; * p < 0.05; ** p < 0.01. All explanatory variables are shares of population between 0 and 1. 206 5.3 Empirical Strategy and Results Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region We identify two sources of energy generation that can economic activity in the area. The timing of opening can be used for the IV strategy: dams for hydroelectricity be considered as exogenous as years of delay are common and wind farms. These sources of energy cover most of for such projects. The random assignment assumption the sources of electricity generation as reported in Table of the IV would imply that a district’s inclusion along a A14. The main sources of energy supply are hydropower straight line corridor is spatially and temporally as good in Cameroon, and gas in Nigeria. Similar to Moneke as randomly assigned. (2020), we develop an IV which yields a hypothetical electrification status based on a location’s proximity to To instrument for the timing of a district’s paved road a straight line corridor from electricity generators to the connection, we find the optimal network to connect main cities. First, we identify the locations of the electricity all cities with more than 50,000 inhabitants in a generators using two databases, one on dams opening year least-cost fashion by employing common minimum and another on power plants locations (Platts database). spanning tree algorithms such as Kruskal’s and For Cameroon, we use the geolocalized database Boruvka’s algorithms. The list of cities with more than including all dams in Africa and their year of opening. 50,000 inhabitants varies over time because of changes For Nigeria, we use the global power plant database that in population, which creates a panel of roads for each includes all power plants per type of energy (hydro, wind, country. gas, and geothermal) with their capacity and year of commissioning. From the year of dam opening or power W  e now run two-stage least squares (2SLS) on the plant commissioning onwards, all districts lying along the following specifications, with province and year fixed straight lines connecting the dams or power plants to the effects and district-level initial values as controls:356 main demand centers are considered as having access to electricity. For Nigeria and Cameroon, the main sources Roadi,t #Electricityi,t = of demand vary across time. At the beginning of our  α + βR (RoadIVi,t = 1 & ElectricityIVi,t = 0) panel, all dams in Cameroon have been opened therefore  + βE (RoadIVi,t = 0 & ElectricityIVi,t = 1) a panel IV is created by varying the sources of demand  + γRE (RoadIVi,t = 1 & ElectricityIVi,t = 1) rather than the supply sources. For Cameroon, we set the  + βI Interneti,t + Controlsi + FE + εi,t  (5.3.3) threshold of 500,000 inhabitants for a city to be included as a main source of demand for the hydropower supply  ith Roadi,t #Electricityi,t being one of the interactions w sources. In 1990, only Douala and Yaounde are included. terms between the dummies Roadi,t and Electricityi,t. In 2000, Garoua in the North is included, and in 2015, Maroua is also included. For each year, all districts lying The second stage equation is given by:  along the straight lines connecting the dams to the cities of more than half million inhabitants will be considered Sectori,t = hypothetically electrified. We then identify the main  α + βR,2SLS (RoadIVi,t = 1 & ElectricityIVi,t = 0) sources of demand in each country. For Nigeria, we  + βE,2SLS (RoadIVi,t = 0 & ElectricityIVi,t = 1) include the cities with more than a million inhabitants.  + γRE,2SLS (RoadIVi,t = 1 & ElectricityIVi,t = 1)  + βI,2SLS Interneti,t + Controlsi,t + YearFE + εi,t Our IV satisfies the main assumptions of an IV (5.3.4) strategy. The choice of location of hydro, gas wind  ith Sectori,t being the share of employment in w generators can be assumed to be driven by geographic agriculture, manufacturing, or services in district i in year and climatic characteristics of the locations and not by t. 356 District level controls variables are interacted with the country dummy such that the effects of distances can only be compared within countries. 5.3 Empirical Strategy and Results 207 Lake Chad Regional Economic Memorandum  |  Development for Peace 5.3.2.2 IV Results countries, the combined effects of electricity and roads are dominant to lead to a reduction of agriculture Table 5.9 reports the results for the 2SLS method for employment and an increase in services and to a lesser Cameroon and Nigeria, that can be compared with the extent in manufacturing. The presence of a paved road OLS method regression for Nigeria and Cameroon in Cameroon has led to a larger impact than in Nigeria only in Table A5.10 in Appendix. First stages results where the change in access to paved roads has been very and weak instrument tests are available on demand. small over the period. Access to internet has had a larger The IV methodology shows stronger effects, especially impact in Nigeria with respect to Cameroon. Table A5.11 for combined investments, than the equivalent OLS in Appendix reports the results of the IV specification for regression. First access to infrastructure leads to a sharp the Lake Chad region. reduction in the share of non-working individuals, especially when access to electricity is secured. Both access to paved roads and to electricity lead to structural transformation with jobs moving from agriculture to manufacturing and services. The impact of infrastructure investments is even larger when combining investments in roads and electricity, with an additional reduction of 19 percentage points in agriculture employment, mostly redistributed towards services. Access to the fiber backbone also has a significant impact and supports employment in the services sector. Table A5.11 reports the results of the IV specification for Nigeria only, and Table A5.12 reports the results of the IV specification for Cameroon only. The effects are similar to Table 5.9 when doing the analysis at the country level for Cameroon and Nigeria. For both Table 5.9: Nigeria and Cameroon: 2SLS method Agriculture Manufacturing Services Not working Paved road=0 × -0.117 0.0499 0.0671 -0.334** Electricity (>50p)=1 (-0.64) (0.84) (0.39) (-3.76) Paved road=1 × -0.0521 0.0412+ 0.0110 -0.0376 Electricity (>50p)=0 (-1.03) (1.73) (0.26) (-1.44) Paved road=1 × -0.191** 0.0486+ 0.142** -0.0923** Electricity (>50p)=1 (-3.32) (1.86) (2.86) (-3.15) -0.0335** -0.00308 0.0365** -0.00654 Internet (-3.10) (-0.59) (3.85) (-1.02) Year + province FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.268 0.130 0.200 -0.122 N. of observations 2,798 2,798 2,798 2,798 Notes: t statistics in parentheses; + p < 0.10; * p < 0.05; ** p < 0.01. 208 5.3 Empirical Strategy and Results Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region 5.4 Welfare Impacts of Infrastructure We use a model to be able to assess the general land, h. Utility of a representative household in location equilibrium effects of infrastructure investments and n is assumed to follow an upper tier Cobb-Douglas produce counterfactuals for future investments. First, functional form over goods and land consumption, scaled we present the general equilibrium model. Second, we by a location-specific amenity shock Vn: calibrate the model based on the current distribution of α 1–α population and economic activities. Third we change the Un = VnCn hn (5.4.1) trade costs to produce counterfactuals.  ith 0 < α < 1. The goods consumption index is defined w over consumption of each tradeable sector’s composite 5.4.1 The Model good and services: T T ρ M M ρ S S ρ We now present a spatial general equilibrium model Cn = [ψ (Cn) + ψ (C n ) + ψ (Cn) ] (5.4.2) based on Moneke (2020). It is characterized by the following broad features. First, locations differ in their  ssuming consumption of sectoral composite goods to be a productivity, geography and trade links with each complementary, i.e. other. Second, road investments are assumed to have general equilibrium effects via changes in trade costs Consumers exhibit love of variety for both tradeable and the resulting reallocation of labor across space as sectors’ goods,CT and CM, which we model in the standard in Allen and Arkolakis (2014) and Redding (2016). CES fashion, where n denotes the consumer’s location Third, electrification investments are assumed to have and i the producer’s location, whereas j is a measure of general equilibrium effects via productivity, like models varieties. Consumption of each tradeable sector’s good is of differential productivity shocks across space such as defined over a fixed continuum of varieties j ϵ [0,1]: Bustos et al. (2016). Lastly, we assume the economy T 1 T v 1/v to consist of multiple sectors of production such that Cn = [∑ ∫0 (cni (j)) d j] (5.4.3) i N changes in sectoral employment across locations (i.e. spatial structural transformation) capture an outcome of  ith v an elasticity of substitution across varieties such w interest as in Michaels et al. (2011) and Eckert and Peters that varieties within each sector are substitutes for each (2018). Compared to Moneke (2020), we consider a other σ = 1/(1−v) > 1. An equivalent formulation is geography that includes several countries which can trade used for CMn. The following equation provides the classic with each other, where additional trade barriers apply for Dixit-Stiglitz price index over traditional sector goods: cross-border trade. Workers can move across locations T 1 T __ 1–σ 1 within country but not across countries. Pn = [∑ ∫0 (pni (j)) d j]1– σ (5.4.4) i N On the production side, there are two tradeable sectors 5.4.1.1 Setup from which firms produce varieties that can be traded across many other locations. Production uses labor and The whole geography consists of many locations, n ε N, land as inputs under constant returns to scale subject to of varying land size (Hn) and endogenous population stochastic location. (Ln). Consumers value consumption of agriculture goods, CT, manufacturing goods, CM, services, CS, and 5.4 Welfare Impacts of Infrastructure 209 Lake Chad Regional Economic Memorandum  |  Development for Peace i i service. We assume agriculture to be the most and services i i Ln μi hn 1–μ i (5.4.5) Yn = ğ ( ) ( ) i=T,M the least land-intensive sector μT < μM < μS. μi 1–μi  here 0 < μi < 1 and, zK denotes the sector-location- w W  ithin each location, the expenditure share on each specific realisation of productivity z for a variety in sector tradeable sector’s varieties and services depends on the i and location n. Following Eaton and Kortum (2002), relative (local) price of each sector’s (composite) good: locations draw sector specific idiosyncratic productivities K κ (ψ ) (P n) K 1–κ K for each variety j from a Frechet distribution: ξn = M κ M 1–κ T κ T 1–κ S κ S 1–κ K {T,M,S} (5.4.10) (ψ ) (Pn) +(ψ ) (Pn) +(ψ ) (Pn ) i i –0 i Fn (ɩ i) = e(–A z ) n i=T,M (5.4.6) G  iven the properties of the Frechet distribution of productivities, tradeable sectoral price indices can be  ith Ain the average sectoral productivity in location n. w further simplified: The shape parameter, θ, determines the variability of i μi 1–μ i –θ –1/θ T –1/θ productivity draws across varieties in a given location n. Pni = γ[kΣN Ak (ωk rk dnk) ] =γ(ɸn ) (5.4.11) Trade in both sectors’ final goods is costly and trade To arrive at a spatial equilibrium, we provide conditions costs are assumed to follow an iceberg structure. Trade for land market clearing, labor market clearing and costs between locations n and m are denoted as dnm, such a labor mobility condition. For an equilibrium in the that quantity dnm > 1 has to be produced in m for one land market, total income from land must equal total unit to arrive in n. We assume that trade costs are the expenditure on land, where the latter summarizes land same across sectors and are symmetric. expenditure by consumers, M-sector firms, T-sector firms and S-sector firms. Similarly, labor market clearing Given perfect competition in both production sectors, requires that total labor income earned in one location the price of a given i-sector variety equals marginal cost must equal total labor payments across sectors on goods inclusive of trade costs: purchased from that location everywhere. Finally, we i μi ωm rm dnm 1–μi assume that workers can freely move across locations p nm = i (5.4.7) within a country but cannot move across countries. zm Therefore, free mobility of workers across locations within with ωm the wage of a worker and rm the price of land.  country implies that the wage earned by workers in a given location after correcting for land and goods prices,  ach location n will buy a given variety from its minimum- E as well as a location’s amenity value, must be equalized cost supplier location m: across locations of a same country. The welfare in each i i location of a same country c is given by: pnm = min {pm , m N} (5.4.8) α 1–α α (1–α) Vn,cωn,c A Vn,c = Vc = α/(1–κ) 1–α , n country c (5.4.12) [Pn,c] rn,c Th  e share of expenditure that the destination location n spends on agricultural sector (and equivalently where Pn,c = (ɸM)k(PMn,c)1−k + (ɸT )k(PTn,c)1−k + (ɸS)  manufacturing sector) goods produced in origin m is k(PSn,c)1−k. We follow the specification in Moneke (2020) given by: and Michaels et al. (2011) to include the district specific Am i (ωmμi r1– μi m dnm) –0 parameter nn;c in the wage so that the welfare can be i πnm = i μ 1–μi –0 i (5.4.9) interpreted as the real income in each location. Σk N Ak (ωm rk dnk) Production of non-tradeable services also uses labor but output is a single homogeneous and land as inputs,  210 5.4 Welfare Impacts of Infrastructure Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region 5.4.2 Calibration of the Model 5.4.3 Counterfactuals We calibrate the model by using some parameters from We calibrate the model to assess the welfare and the literature and by recovering the key productivity spatial impacts of new transport investments. The parameters and wages to obtain an equilibrium for counterfactual exercise is done in three steps. First, we the current situation. Table A5.15 in the appendix calibrate the model to obtain the underlying parameters reports the parameters from the literature to calibrate of the model for the baseline situation, without the new the model which are like the ones in Moneke (2020) investments as explained in the previous section. Second, applied to Ethiopia. We use the sectoral labor shares from we update the trade costs based on the new assumptions. Table A5.15. To recover the productivity parameters, Third, we use the model to obtain the new employment we use the labor market clearing conditions, the land shares given the new transport costs, the wage per location, market conditions, and the labor mobility conditions. and therefore the real wage given the new equilibrium For each location, the model admits three equations goods and housing prices. for the three endogenous variables in each location— land market clearing, labor market clearing, and labor We use the available road networks for each country, mobility condition—which allows to solve for a general with assumptions on speed along the networks given equilibrium of the model in terms of its core endogenous the type and condition of roads that are registered. We variables: wages, land rental rates, and population. rely on additional features such as the type of surface and Moneke (2020) shows the uniqueness of the equilibrium the condition of the roads. Investments are assumed to based on a similar work by Michaels et al. (2011). We increase the speed at which vehicles can travel along the obtain a series of {ATn ,AMn ,ASn}nϵN for which the segments that are improved or build new links between distribution of population, employment, and land is an locations. We assume trade costs to be iceberg costs such equilibrium given the current trade costs. that the costs between location o and destination d are Map 5.1: Descriptive statistics for the 24 regions in the Lake Chad a. Share of employment in agricultural sector b. Total population per subregion Source: Authors’ calculations. 5.4 Welfare Impacts of Infrastructure 211 Lake Chad Regional Economic Memorandum  |  Development for Peace given by dod = max(1,timeτ). Border costs are also added 5.4.3.1 New transport infrastructure in Cameroon to trace costs as detailed in the following sections. and Chad We calibrate the model using spatial data for land, We investigate the impact of several transport and population, and sectoral shares from the sources trade facilitation projects listed in Table 5.10 and previously used. Because of the complexity of a 3-sector represented on Map 5.2 for the part potentially financed model to converge in order to recover the initial sectoral by the World Bank and on Figures A5.22 for all the productivities, we reduce the spatial disaggregation segments financed by different donors assessed in the to fewer locations. Such aggregation also smoothes next section. These projects are part of a comprehensive measurement issues of sectoral employment based on the and continued approach to provide a long-term, reliable, DHS data. For Lake Chad, we have a total of 24 regions, 8 safe and efficient multimodal corridor over the entire for Cameroon (adm1 level), 6 for Nigeria, and 8 for Chad. 1,800km long stretch between Douala-Ngaoundéré- Map 5.1 shows the share of agricultural employment and Koutéré-Moundou-Ndjamena (forthcoming World Bank population for each of these subnational regions. PAD document). The corridor contributes to improve domestic, regional as well as international connectivity for both countries. Map 5.2: World Bank investments in rail and road projects in Cameroon and Chad 212 5.4 Welfare Impacts of Infrastructure Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region Table 5.10: Summary of counterfactual scenarios Baseline Scenario Country Infrastructure Policies Rail line that is less and less Platform in Ngadoundere to move 1 Cameroon competitive with the road: speed from rail to road 40km per hour Corridor (Njamena-Moundou) in bad Land border with Cameroon: 30 conditions: speed 30km/h followed hours per each border point + Chad by the segment Moundou to the administrative costs to trade border in good condition Transport infrastructure investments Cameroon 2.1 Chad Upgraded road line Cameroon Upgraded rail line 2.2 Chad Upgraded road line Border investments 3 Cameroon and Chad Baseline from 30 hours to half time: 15 hours Transport + Border investments Cameroon Baseline Half border time 4.1 Chad Upgraded road corridor Half border time Cameroon Upgraded rail line Half border time 4.2 Chad Upgraded road corridor Half border time Rehabilitation of the rail line in Cameroon. The Tensions in the Extreme North have closed the corridors renovation of the rail line between Ngaoundere, Yaounde passing by the Northern part of Cameroon and opened and Douala in Cameroon is going through several steps. the possibilities for other corridors to develop. From The World Bank participates in the financing of the Ngaoundére to Ndjamena, the traditional road corridor Southern part of the project for the segment between crosses the region of North Cameroun which is under the Douala and Yaounde, while Agence Française de threat of Boko Haram, for this reason this road section is Développement (AFD) and European Investment Bank today considered as unsafe and unreliable. An alternative (EIB) are planning to finance in 2022 the rehabilitation of road corridor from Ngaoundére to Ndjamena crosses the the section from Belabo up to Ngaoundere in the North border near Moundou (second largest city in Chad) and Cameroun. We assume that the two rehabilitations will then connect Moundou with Ndjamena (about 600 km). happen at the same time so we consider both segments. The World Bank is currently assessing a project aiming at The government is currently planning to renovate the rehabilitating this section of the corridor with other donors most used segment between Yaounde and Douala, co-financing as today the road is totally dilapidated while whose condition has deteriorated in the last years. After 100% is paved. This project would improve connectivity these projects are completed, the whole existing railway to the port of Douala, increases domestic connectivity network will be rehabilitated, increasing capacity safety, between the main two cities of Chad and improves the speed, reliability and efficiency of rail traffic and therefore regional/ international connectivity of Moudoun. The improving performance of the corridors. We assume very proposed project covers the whole corridor between low speed on the whole line in the baseline. Koutere-Moundou-Ndjamena under a phased 10-year long-term Output Performance Based approach that Rehabilitation of road corridors in Chad. There are entails rehabilitation works, reinforcement, maintenance several historical corridors between Cameroon and Chad. and axle load monitoring facilities management. 5.4 Welfare Impacts of Infrastructure 213 Lake Chad Regional Economic Memorandum  |  Development for Peace Complementary policies: Border frictions. We assume Second we compute the welfare impacts in each that trade across locations from a same country only face counterfactual and compare it to the baseline welfare. transport costs while traders across countries have to wait Because we do not allow for mobility across countries, an additional 30 hours to be able to cross the borders. welfare is equalized within each country but not across Given the lack of data, we assume a level of 30 hours countries. by default. In the forthcoming counterfactuals, we add a reduction of half border time to the transport investments. ΔWelfarec = ΔPopulationn,c × Vn,c (5.4.14)  ith Vn,c the welfare in location n of country c defined in w 5.4.3.2 Calibration of the new counterfactual equation 4.12. We use the most recent transport networks from each Table 5.11 reports the change in employment share country as the baseline. We assume a new speed of in non-agricultural sectors at the national level from 70km/h for the new road corridor and reduce the time at the combined road and rail investments and with the the border in some of the scenarios. For the rail corridor, additional border friction reduction. The proposed we keep the road network as it is and assume a new transport investments are expected to have a marginal direct transport line between Ngaoundere, Yaounde and impact on structural change away from agriculture at the Douala. New transport times are computed assuming that national level in most countries. When reducing border the previous roads can be used as well as the new rail line, frictions, Nigeria benefits from a better access to its which is more efficient. Stops between the main cities neighbors to specialize slightly more in non-agricultural are not permitted. The first section between Ngaoundere activities. Chad tends to specialize more in the primary and Yaounde is assumed to be 627kms with an average sector. However most changes will happen between speed of 70km/h.357 The second section between Yaounde regions within country. and Douala is assumed to be 261kms with a new average speed of 70km/h. Tables 5.12 and 5.14 show the change in nominal income and welfare (real income) for the three countries when only considering the new road 5.4.3.3 Welfare impacts corridor in Chad. Tables 5.13 and 5.15 show the change in nominal income and welfare (real income) for the We look at the impacts on country GDP and welfare. three countries when considering both the rail and road Welfare differs from income as it also includes differences corridors in Chad and Cameroon. Overall Chad gains in prices for goods and housing across locations as well the most from the projected investments as it benefits as an amenity from living in different places. First we from its own corridor and the corridor in Cameroon. compute the nominal GDP impacts measured as the total The new transport corridor in Cameroon benefits the nominal incomes. most Cameroon when it is accompanied by a significant reduction in border costs. Nigeria benefits from welfare ΔGDPc = ΔPopulationn,c × NominalIncomen,c (5.4.13) gains through lower prices in the two other countries, but does not benefit from income gains. Tables 5.12 and 5.13  ith Incomen,c the total nominal income in location n w show that most gains come increasing purchasing power of country c. due to lower prices. 357 Distance assumptions come from the website rome2rio.com which reports distance per transport mode. 214 5.4 Welfare Impacts of Infrastructure Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region  oad and rail corridors: change of Table 5.11: R transport investments, especially the road corridor in employment share in non-agricultural sectors (in percentage points) Chad, and from indirect better connectivity to the rest of Cameroon as well as from the reduction in border delays. Scenarios Cameroon Chad Nigeria Transport -0.3 0.05 0.03 only 5.4.3.4 Spatial impacts of the road and rail Transport + corridors 0.1 -0.6 1.8 Border Table 5.12: R  oad corridor in Chad: Percentage Change Maps below report the spatial impacts in terms of in nominal GDP change in agricultural employment and welfare Scenarios Cameroon Chad Nigeria changes at the aggregated local level for two scenarios: only with the rail and road transport investments and Transport 0 0.1 0 with reduction in border time. only Transport + 0 0.7 0 Change in non-agricultural agricultural employment Border shares. Map 5.3 reports the change in employment at the  oad and rail corridors: Percentage Change Table 5.13: R subnational level. The specialization patterns differ when in nominal GDP reducing border frictions or not. Combined transport and Scenarios Cameroon Chad Nigeria border investments increase the specialization of Nigerian regions and the South-Eastern part of Cameroon towards Transport 0 0.3 0 non-agricultural activities. The rest of Cameroon and only Transport + most Chadian regions specialize more into tradable 0 0.6 0 Border agricultural activities. Map 5.4 show that it is mostly the South-Eastern parts of Cameroon that will specialize  oad corridor in Chad: Percentage Change Table 5.14: R more in manufacturing activities. in Welfare (real income) Nigeria Lake Chad Overall, welfare gains, measured as gains in real Scenarios Cameroon Chad region subnational income, are positive at the aggregate level Transport for all countries but not for all subnational regions. 0.0 0.23 0.02 0.1 only Regional real income, i.e. the sum of real incomes for Transport 2.25 3.3 2.3 4 the whole population of the region, increases for the + Border regions that benefit from new corridors and lower border Table 5.15: R  oad and rail corridors: Percentage Change costs first. When only investing in corridors, some in Welfare (real income) regions lose in terms of regional income, while others gain. When adding border time reduction, the large Nigeria Lake Chad Scenarios Cameroon Chad majority gains from lower trade costs and new regional region Transport trade opportunities. As seen from Tables 5.12 and 5.13, 0.5 0.7 0.03 0.8 only most gains come from lower prices and therefore higher Transport purchasing power. 2.8 3.7 2.3 4.8 + Border Tables 5.14 and 5.15 show that the welfare gains in the limited Lake Chad region would be relatively larger than at the country level. The region benefits from direct 5.4 Welfare Impacts of Infrastructure 215 Lake Chad Regional Economic Memorandum  |  Development for Peace  hange in share of employment in non-agricultural sectors from transport corridor investments (left) Map 5.3: C with additional border reduction (right) compared to baseline - in percentage points  hange in the share of employment in manufacturing sectors from transport corridor investments Map 5.4: C (left) with additional border reduction (right) compared to baseline - change in percentage points 216 5.4 Welfare Impacts of Infrastructure Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region  egional welfare impacts from transport corridor investments (left) with additional border reduction Map 5.5: R (right) - percentage change in regional welfare 5.4 Welfare Impacts of Infrastructure 217 Lake Chad Regional Economic Memorandum  |  Development for Peace 5.5 Conclusion This paper investigates how infrastructure—transport, electricity and internet—affects economic development through the channels of sectoral employment and structural change. First, the paper provides estimates of the impacts of past transport, electricity and internet investments in Nigeria, Cameroon and Chad on sectoral employment. Using a series of instruments, we estimate a large impact of infrastructure investments, especially from the combination of paved roads and electricity. The paper then uses a spatial general equilibrium model to provide estimates of the potential impacts of proposed regional transport corridor projects in the Lake Chad, with a focus on the structural transformation at the regional, county and subnational levels. The analysis also looks at the impact of complementary trade facilitation measures. The analysis shows the importance of such complementary interventions to facilitate regional trade and enhance the benefits of transport corridors. The spatial general equilibrium model developed in the paper does not consider investments in electricity and internet. The plan for future research is to include those infrastructure sectors in the model and link it with the empirical analysis. 218 5.5 Conclusion Technical Paper 4. Infrastructure and Structural Change in the Lake Chad Region References Adukia, Anjali, Sam Asher, and Paul Novosad, “Educational Investment Responses to Economic Opportunity: Evidence from Indian Road Construction,” American Economic Journal: Applied Economics, January 2020, 12 (1), 348–376. Ali, Rubaba, A. Federico Barra, Claudia Berg, Richard Damania, John Nash, and Jason Russ, Highways to Success or Byways to Waste number 22551. In ‘World Bank Publications.’, The World Bank, November 2015. Allen, Treb and Costas Arkolakis, “Trade and the Topography of the Spatial Economy,” The Quarterly Journal of Economics, 2014, 129 (3), 1085–1140. Arderne, C., C. Zorn, C. Nicolas, and E. Koks, “Predictive mapping of the global power system using open data.,” Sci Data 7, 19, 2020. Asher, Sam and Paul Novosad, “Rural roads and local economic development,” Policy Research Working Paper Series 8466, The World Bank June 2018. Bird, Julia, Mathilde Lebrand, and Anthony J. Venables, “The Belt and Road Initiative: Reshaping economic geography in Central Asia?,” Journal of Development Economics, 2020, 144 (C). Bustos, Paula, Bruno Caprettini, and Jacopo Ponticelli, “Agricultural Productivity and Structural Transformation: Evidence from Brazil,” American Economic Review, June 2016, 106 (6), 1320–1365. Calderon, Cesar, Enrique Moral-Benito, and Luis Serven, “Is infrastructure capital productive? A dynamic heterogeneous approach,” Journal of Applied Econometrics, March 2015, 30 (2), 177–198. Donaldson, Dave, “Railroads of the Raj: Estimating the Impact of Transportation Infrastructure,” American Economic Review, April 2018, 108 (4-5), 899–934. Eckert, Fabian and Michael Peters, “Spatial Structural Change,” 2018 Meeting Papers 98, Society for Economic Dynamics 2018. Foster, Vivien and Cecilia Briceno-Garmendia, Africa’s Infrastructure : A Time for Transformation [Infrastructures africaines] number 2692. In ‘World Bank Publications.’, The World Bank, June 2010. Fried, Stephie and David Lagakos, “Electricity and Firm Productivity: A General-Equilibrium Approach,” Working Paper 27081, National Bureau of Economic Research May 2020. Gertler, Paul J., Marco Gonzalez-Navarro, Tadeja Gravcner, and Alexander D. Rothenberg, “Road Quality, Local Economic Activity, and Welfare: Evidence from Indonesia’s Highways,” 2016. Herrera Dappe, Matías and Mathilde Lebrand, “The Spatial Effects of Logistics Interventions on the Economic Geography of Bangladesh,” 2019. _ and _ , “Infrastructure and Structural Change in the Horn of Africa,” 2021. _ , _ , and Diana Van Patten, “Bridging India and Bangladesh: Cross-border Trade and the BBIN Motor Vehicles Agreement,” 2021. Jedwab, Remi and Adam Storeygard, “The Average and Heterogeneous Effects of Transportation Investments: Evidence from Sub-Saharan Africa 1960-2010,” NBER Working Papers 27670, National Bureau of Economic Research, Inc August 2020. Lall, Somik V. and Mathilde Lebrand, “Who wins, who loses? Understanding the spatially differentiated effects of the belt and road initiative,” Journal of Development Economics, 2020, 146 (C). McMillan, Margaret, Dani Rodrik, and Claudia Sepulveda, “Structural Change, Fundamentals and Growth: A Framework and Case Studies,” Working Paper 23378, National Bureau of Economic Research May 2017. References 219 Lake Chad Regional Economic Memorandum  |  Development for Peace Michaels, Guy, Ferdinand Rauch, and Steve Redding, “Technical note: an Eaton and Kortum (2002) model of urbanization and structural transformation.,” 2011. Moneke, Niclas, “Can Big Push Infrastructure Unlock Development? Evidence from Ethiopia,” Mimeo 2020. Ngai, L. Rachel and Christopher A. Pissarides, “Structural Change in a Multisector Model of Growth,” American Economic Review, March 2007, 97 (1), 429–443. Redding, Stephen J., “Goods trade, factor mobility and welfare,” Journal of International Economics, 2016, 101 (C), 148–167. 220 References Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Appendix A5.1 Employment Figure A5.1:  Employment in agriculture in the Lake Chad countries Appendix 221 Lake Chad Regional Economic Memorandum  |  Development for Peace A5.2 Mapping Infrastructure in the Lake Chad A5.2.1 Measuring electricity access based on Night-Time Lights (2016)  ercentage of population with access to electricity (2016) Figure A5.2: P Percent of population with electricity Percent of population with electricity Percent of population with electricity 222 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region A5.2.2 Nigeria Figure A5.3: Access to paved roads a. All b. In fair or good condition Note: Data from Foster and Briceno-Garmendia (2010) and Ali et al. (2015). The map represents the year at which access to a paved road is observed. 0 means that no paved road is reported in the latest observed year. 2013 refers to districts with a paved road when observed in 2013 only. 2018 refers to additional districts with a paved road when observed in 2018 compared to 2013. Figure A5.4: Access to electricity Note: Authors’ calculations using Night Time Lights. The mapap represents the year at which at least 50 percent of the population has access to electricity, measured by lights at night. 0 means that no access to electricity is reported in the latest observed year. The earliest year refers to districts with access when observed in that year only. Successive years refer to additional districts which gained access when compared to previous years. Appendix 223 Lake Chad Regional Economic Memorandum  |  Development for Peace Figure A5.5: Access to internet fibre network a. From 2009 to 2019 b. Years of interest Note: Authors’ calculations using Africa Bandwidth Maps. The maps represent access to the fiber network as measured with a node being present in the district. 0 means that no access is reported in the latest observed year. The earliest year refers to districts with access when observed in that year only. Successive years refer to additional districts which gained access when compared to previous years.  ercentage of districts and population with access to a paved road Figure A5.6: P a. All b. In fair or good condition Note: Data from Foster and Briceno-Garmendia (2010) and Ali et al. (2015). The population used for weighted average of access is from GHS 2015. 224 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Figure A5.7:  Percentage of districts and population with access to electricity for different thresholds a. Threshold=10% b. Threshold=50% c. Threshold=90% Note: Authors’ calculations. Figure A5.8: Access to internet broadband Note: Authors’ calculations using Africa Bandwidth Maps. Appendix 225 Lake Chad Regional Economic Memorandum  |  Development for Peace  ercentage of districts and population with access to combined infrastructures Figure A5.9: P a. Roads and electricity b. Roads and internet c. Electricity and internet Note: Authors’ calculations. 226 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region A5.3 Cameroon Figure A5.10: Access to paved roads a. All b. In fair or good condition Note: Data from Foster and Briceno-Garmendia (2010) and government sources. The map represents the year at which access to a paved road is observed. 0 means that no paved road is reported in the latest observed year. 2013 refers to districts with a paved road when observed in 2013 only. 2018 refers to additional districts with a paved road when observed in 2018 compared to 2013. Appendix 227 Lake Chad Regional Economic Memorandum  |  Development for Peace Figure A5.11: Access to electricity Note: Authors’ calculations using Night Time Lights. The map represents the year at which at least 50 percent of the population has access to electricity, measured by lights at night. 0 means that no access to electricity is reported in the latest observed year. The earliest year refers to districts with access when observed in that year only. Successive years refer to additional districts which gained access when compared to previous years. 228 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Figure A5.12: Access to internet fibre network a. From 2009 to 2019 b. Years of interest Note: Authors’ calculations using Africa Bandwidth Maps. The maps represent access to the fiber network as measured with a node being present in the district. 0 means that no access is reported in the latest observed year. The earliest year refers to districts with access when observed in that year only. Successive years refer to additional districts which gained access when compared to previous years. Appendix 229 Lake Chad Regional Economic Memorandum  |  Development for Peace Figure A5.13:  Percentage of districts and population with access to a paved road a. All b. In fair or good condition Note: Data from Foster and Briceno-Garmendia (2010) and government sources. The population used for weighted average of access is from GHS 2015.  ercentage of districts and population with access to electricity for different thresholds Figure A5.14: P a. Threshold=10% b. Threshold=50% c. Threshold=90% Note: Authors’ calculations. 230 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Figure A5.15: Access to internet Note: Authors’ calculations using Africa Bandwidth Maps.  ercentage of districts and population with access to combined infrastructures Figure A5.16: P a. Roads and electricity b. Roads and internet c. Electricity and internet Note: Authors’ calculations. Appendix 231 Lake Chad Regional Economic Memorandum  |  Development for Peace A5.4 Chad Figure A5.17: Access to paved roads and electricity a. Paved roads b. Electricity (2014) Note: Data using government sources and Night Time Lights. The map (b) represents the year at which at least 50 percent of the population has access to electricity, measured by lights at night in 2014. The map (a) represents the year at which at the population has access to a paved road. The map represents the year at which access is observed. 0 means that no access is reported in the latest observed year. Successive years refer to additional districts which gained access when compared to previous years. 232 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Figure A5.18: Access to internet fibre network a. From 2009 to 2019 b. In 2014 Note: Authors’ calculations using Africa Bandwidth Maps. The maps represent access to the fiber network as measured with a node being present in the district. 0 means that no access is reported in the latest observed year. The earliest year refers to districts with access when observed in that year only. Successive years refer to additional districts which gained access when compared to previous years. Appendix 233 Lake Chad Regional Economic Memorandum  |  Development for Peace  ercentage of districts and population with access to infrastructure (2014) Figure A5.19: P a. All b. Combined c. Electricity thresholds Note: Authors’ calculations. The population used for weighted average of access is from GHS 2015. A5.5 Additional Data Table A5.1: GADM administrative levels Figure A5.20: Regions around the Lake Chad Level NGA NER TCD CMR adm1 37 8 23 10 adm2 775 36 55 58 adm3 NA 132 348 360 234 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Figure A5.21: Electricity grid A5.6 Regression tables Table A5.2: Lake Chad (electricity grid) Agriculture Manufacturing Services -0.0598** 0.0383** 0.0143 Paved road (-2.82) (3.94) (0.91) -0.0852+ 0.0142 0.0582 Internet (-1.81) (0.97) (1.53) -0.0451+ 0.0243+ 0.0244 Electricity grid (-1.83) (1.95) (1.44) -0.00883 -0.00867 0.0290 Road + Internet (-0.18) (-0.56) (0.73) -0.0106 -0.000294 0.0176 Road + Electricity grid (-0.38) (-0.02) (0.88) Year + Country FE Yes Yes Yes Controls Yes Yes Yes R-squared 0.296 0.342 0.347 N. of observations 3,041 3,041 3,041 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Appendix 235 Lake Chad Regional Economic Memorandum  |  Development for Peace  eterogenous effects on agriculture employment, by the initial share of agricultural employment, in Table A5.3: H Cameroon Agriculture q0.25 Agriculture q0.5 Agriculture q0.75 main -0.134** -0.128** -0.107** Paved road (-3.34) (-6.59) (-4.95) -0.0598 -0.0726 -0.141+ Internet (-0.52) (-0.71) (-1.75) -0.342** -0.269+ -0.192+ Electricity (>50p) (-3.13) (-1.87) (-1.71) 0.0282 0.0551 0.105 Road + Internet (0.26) (0.51) (1.30) 0.105 0.0117 -0.0520 Road + Electricity (1.06) (0.08) (-0.41) Year FE Yes Yes Yes Controls Yes Yes Yes R-squared N. of observations 661 661 661 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Table A5.4: Nigeria, Cameroon, Chad in the period 2008–2018 Agriculture Manufacturing Services Not working -0.0944** 0.0377** 0.0611** 0.0583** Paved road (-4.47) (4.36) (3.68) (6.11) -0.0706 0.00874 0.0680 0.0162 Internet (-1.29) (0.42) (1.56) (0.67) -0.182** 0.0661** 0.117** 0.0778** Electricity (>50p) (-5.38) (3.37) (4.60) (4.81) 0.0268 -0.0110 -0.0204 -0.0293 Road + Internet (0.48) (-0.51) (-0.46) (-1.17) -0.0122 0.00351 0.00647 -0.111** Road + Electricity (-0.35) (0.18) (0.25) (-6.67) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.458 0.341 0.417 0.162 N. of observations 1,817 1,817 1,817 1,817 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 236 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Table A5.5: Around Lake Chad: the usage perspective, 2008–2018 Agriculture Manufacturing Services Not working -0.227 0.155 0.165 0.0238 Electricity (share) (-1.01) (1.44) (1.13) (0.50) -0.0123 -0.0223 0.0820 -0.0885 Motorcycle (share) (-0.10) (-0.37) (0.85) (-1.64) -1.799** 0.432+ 0.975+ -0.0441 Car (share) (-3.02) (1.94) (1.97) (-0.14) Use of internet -0.712 -0.442 0.983 -0.380 (share) (-0.70) (-0.77) (1.52) (-1.32) Mobile phone -0.526* 0.113 0.324* 0.278** (share) (-2.52) (1.22) (2.10) (3.03) 0.258 -0.0435 -0.217 0.383 Land phone (share) (0.30) (-0.13) (-0.34) (0.93) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.552 0.466 0.540 0.749 N. of observations 96 96 96 96 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Table A5.6: Cameroon: from the usage perspective, after 2008 Agriculture Manufacturing Services Not working -0.413** 0.157** 0.256** 0.0727** Electricity (share) (-11.60) (7.69) (9.42) (3.98) -0.125* 0.0456 0.0799 -0.0116 Motorcycle (share) (-2.10) (1.36) (1.65) (-0.37) -0.369** 0.0479 0.321** 0.00608 Car (share) (-2.88) (0.63) (3.18) (0.07) Use of internet -0.284** 0.123* 0.161* -0.0947 (share) (-3.11) (2.18) (2.13) (-1.64) Mobile phone -0.220** -0.0159 0.236** 0.176** (share) (-3.13) (-0.42) (4.21) (3.99) -0.244 -0.197 0.441* -0.140 Land phone (share) (-1.09) (-1.08) (2.29) (-1.00) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.661 0.427 0.604 0.178 N. of observations 442 442 442 442 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Appendix 237 Lake Chad Regional Economic Memorandum  |  Development for Peace Table A5.7: Nigeria: from the usage perspective, after 2008 Agriculture Manufacturing Services Not working -0.323** 0.0827** 0.241** -0.0458** Electricity (share) (-18.64) (10.53) (16.55) (-5.59) 0.0161 -0.00839 -0.00770 -0.00806 Motorcycle (share) (0.64) (-0.61) (-0.34) (-0.59) -0.382** -0.0461+ 0.428** -0.0144 Car (share) (-8.56) (-1.79) (11.20) (-0.61) Use of internet 0.111* -0.0852** -0.0259 -0.244** (share) (2.14) (-2.86) (-0.56) (-7.65) Mobile phone -0.198** 0.103** 0.0951** 0.0551* (share) (-6.61) (6.85) (3.48) (2.34) -0.271* -0.0236 0.295* 0.166** Land phone (share) (-2.20) (-0.35) (2.40) (2.67) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.572 0.315 0.514 0.176 N. of observations 1,684 1,684 1,684 1,684 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Table A5.8: Chad Agriculture Manufacturing Services 0.0220 0.00467 0.0122 Paved road (0.28) (0.29) (0.29) -0.113 0.0273 0.0403 Internet (-0.94) (1.14) (1.31) 0.0595 -0.0145 0.00794 Electricity (>50p) (0.82) (-0.86) (0.31) 0.111 -0.0590* -0.0184 Road + Internet (0.78) (-2.10) (-0.37) 0 0 0 Road + Electricity (.) (.) (.) Year FE Yes Yes Yes Controls Yes Yes Yes R-squared 0.426 0.339 0.551 N. of observations 243 243 243 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 238 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region A5.7 Regression tables: heterogenous impacts  eterogenous effects on agriculture employment, by the initial share of agricultural employment, in Table A5.9: H Nigeria Agriculture q0.25 Agriculture q0.5 Agriculture q0.75 main 0.00808 0.0171 0.0122 Paved road (0.26) (0.64) (0.22) 0.000394 0.0777 0.0696 Internet (0.00) (0.96) (0.66) -0.0709 -0.134** -0.245** Electricity (>50p) (-1.06) (-2.61) (-3.08) -0.0203 -0.109 -0.149 Road + Internet (-0.18) (-1.33) (-1.31) -0.0795 -0.0739 -0.00163 Road + Electricity (-1.14) (-1.59) (-0.02) Year FE Yes Yes Yes Controls Yes Yes Yes R-squared N. of observations 2,137 2,137 2,137 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Appendix 239 Lake Chad Regional Economic Memorandum  |  Development for Peace A5.8 Regression tables: IV strategy Table A5.10: Nigeria and Cameroon: comparison table (OLS) Agriculture Manufacturing Services Not working -0.0683** 0.0406** 0.0292* 0.0337** Paved road (-3.77) (4.89) (2.25) (4.41) -0.0869+ 0.0110 0.0682 0.0121 Internet (-1.68) (0.65) (1.58) (0.51) -0.200** 0.0870** 0.117** 0.0481** Electricity (>50p) (-5.30) (3.82) (4.49) (3.24) 0.0277 -0.0132 -0.00571 -0.0292 Road + Internet (0.53) (-0.76) (-0.13) (-1.21) -0.00481 -0.0238 0.0253 -0.0580** Road + Electricity (-0.13) (-1.04) (1.00) (-3.89) Year + Country FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.396 0.325 0.431 0.114 N. of observations 2,798 2,798 2,798 2,798 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Table A5.11: Nigeria - 2SLS method Agriculture Manufacturing Services Paved road=0 × Electricity -0.142 -0.0299 0.172 (>50p)=1 (-0.77) (-0.48) (0.95) Paved road=1 × Electricity -0.0273 0.0202 0.00709 (>50p)=0 (-0.47) (0.76) (0.14) Paved road=1 × Electricity -0.221** 0.0478+ 0.173** (>50p)=1 (-3.24) (1.67) (2.86) -0.0319** -0.00275 0.0346** Internet (-2.63) (-0.49) (3.13) Year + province FE Yes Yes Yes Controls Yes Yes Yes R-squared 0.224 0.076 0.177 N. of observations 2,137 2,137 2,137 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 240 Appendix Chapter 5. Infrastructure and Structural Change in the Lake Chad Region Table A5.12: Cameroon - 2SLS method Agriculture Manufacturing Services Paved road=0 × Electricity 0.479 0.0795 -0.559 (>50p)=1 (0.93) (0.35) (-1.31) Paved road=1 × Electricity -0.207* 0.130** 0.0773 (>50p)=0 (-2.42) (2.60) (1.22) Paved road=1 × Electricity -0.321** 0.0478 0.273** (>50p)=1 (-2.95) (0.75) (3.37) 0.0000678 -0.0264+ 0.0264 Internet (0.00) (-1.75) (1.01) Year + province FE Yes Yes Yes Controls Yes Yes Yes R-squared 0.252 0.145 0.055 N. of observations 661 661 661 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Table A5.13: Lake Chad area: 2SLS method Agriculture Manufacturing Services Not working Paved road=1 × -0.133 0.271 -0.138 -0.166 Electricity (>50p)=0 (-0.48) (1.21) (-0.60) (-1.17) Paved road=1 × -0.482 0.518* -0.0363 0.0923 Electricity (>50p)=1 (-1.58) (2.11) (-0.17) (0.61) -0.221** 0.0257 0.195* 0.0225 Internet (-2.59) (0.42) (2.41) (0.46) Year + province FE Yes Yes Yes Yes Controls Yes Yes Yes Yes R-squared 0.528 0.099 0.284 0.063 N. of observations 91 91 91 91 t statistics in parentheses; + p < 0.10, * p < 0.05, ** p < 0.01 Table A5.14: Energy sources in electricity production (in %) 1990 2015 Cameroon Nigeria Cameroon Nigeria Hydro 99% 42% 75% 18% Renewable, excluding hydro NA NA 1% NA Oil, gas and coal 1% 58% 25% 82% Appendix 241 Lake Chad Regional Economic Memorandum  |  Development for Peace Figure A5.22: Rail and road investments in Cameroon (left) and Chad (right) A5.9 Calibration of the Model: Parameters Table A5.15: Parameters for Structural Estimation Parameter Value Source Description σ 4 Bernard et al. (2003) Elasticity of substitution between varieties 1−α 0.25 Data for Ethiopia (HCES) Expenditure share on land/housing κ 0.5 Ngai and Pissarides (2007) Elasticity of substitution across sectors μM 0.82 Moneke (2020) for Ethiopia Labor share in M-production μT 0.78 Moneke (2020) for Ethiopia Labor share in T-production μS 0.84 Moneke (2020) for Ethiopia Labor share in S-production τ 0.3 Moneke (2020) for Ethiopia Elasticity of trade cost with respect to distance θ Shape parameter of productivity distribution 4 Donaldson (2018) across varieties & locations 242 Appendix