Report No. 115122 BURKINA FASO POVERTY AND VULNERABILITY ANALYSIS June 2016 Poverty Global Practice Africa Region Document of the World Bank For Official Use Only Acknowledgements This report was prepared by a core team consisting of Prospere Backiny-Yetna (TTL), Yele Maweki Batana (Senior Economist), Cheikh Ibrahima Nokho (Consultant), Herve Guene (Consultant) and Theophile Bougna (Consultant). The report was prepared in close collaboration with Victoria Monchuk (Senior economist, GSP07) and Pascale Schnitzer (Economist, GSP07) who provide inputs on the structure of the report and on many aspects related to vulnerability analysis. The report was prepared under the guidance of Pablo Fajnzylber (Practice Manager). The Team benefited from comments provided by the peer reviewers: Mariam Diop (Senior Economist, GMF01) and Carlo Del Ninno (Senior Economist, GSP07) and Jacques Morisset (Program leader, AFCF2. The Team would also like to thank Zakaria Koncobo of INSD for his helpful suggestions. Funding for the report was in part provided by the Adaptive Social Protection Program multi-donor trust fund, currently supported by the Department for International Development (DFID). 2 ABBREVIATIONS AND ACRONYMS AAP Africa Adaptation Program CFA Franc de la Communauté Financière Africaine EICVM Enquête Intégrale sur les Conditions de Vie des Ménages EMC Enquête Multisectorielle continue FAO Food and Agriculture Organization FIES Food Insecurity Experience Scale GDP Gross domestic product ILO International Labor Organization INSD Institut National de la Statistique et de la Démographie PPP Parité du pouvoir d’achat QUIBB Questionnaire sur les Indicateurs de Base et de Bien-être RDA Recommended Dietary Allowances FGT Foster-Greer-Thorbecke SCADD Strategy de Croissance Accélérée et de Développement Durable SCD Systematic Country Diagnostic SDGs Sustainable Development Goals SSA Sub-Saharan Africa WAEMU West African Economic and Monetary Union Vice President : Makhtar Diop Country Director : Pierre Frank Laporte Poverty GP Senior Director : Ana Revenga Poverty GP Practice Manager : Pablo Fajnzylber Task Manager(s) : Prospere Backiny-Yetna 3 Table of Contents Acknowledgements ....................................................................................................................................... 2 ABBREVIATIONS AND ACRONYMS ..................................................................................................... 3 List of Tables ................................................................................................................................................ 5 List of Boxes ................................................................................................................................................. 7 EXECUTIVE SUMMARY .......................................................................................................................... 8 1. COUNTRY CONTEXT...................................................................................................................... 15 2. PROGRESS IN POVERTY REDUCTION IN BURKINA FASO .................................................... 18 2.1 Poverty trends ........................................................................................................................... 18 2.2 Poverty profile and determinants in 2014 ............................................................................... 37 3. FOOD INSECURITY IN BURKINA FASO ..................................................................................... 45 3.1 Characteristics of food insecurity ............................................................................................ 45 3.2 Food insecurity and vulnerability to shocks ........................................................................... 55 4. RURAL INCOME AND POVERTY ................................................................................................. 62 4.1 Profile of rural households ....................................................................................................... 62 4.2 Stylized facts on labor market and income source in rural areas ........................................ 63 4.3 Agricultural sector .................................................................................................................... 70 4.4 Non-farm enterprise income .................................................................................................... 76 4.5 Private transfers ........................................................................................................................ 79 5. CONCLUSIONS................................................................................................................................. 85 ANNEXES .................................................................................................................................................. 87 4 List of Tables Table 2.1 Poverty indicators by area of residence ...................................................................................... 24 Table 2.2: Actual and alternative poverty lines .......................................................................................... 26 Table 2.3: Comparison of national accounts and household survey growth rates ...................................... 28 Table 2.4: Average GDP growth rates ........................................................................................................ 29 Table 2.5: Primary employment by year and type of industry .................................................................... 33 Table 2.6: Inequality indicators by area of residence ................................................................................. 33 Table 2.7: GDP per sector (2009-2014) and projections (2015-2017) ....................................................... 35 Table 2.8: Poverty Projections for 2016-2030 ............................................................................................ 37 Table 2.9: Basic poverty indicators by household characteristic ................................................................ 40 Table 2.10: Poverty indicators using alternative poverty measures ............................................................ 42 Table 3.1: Average annual cereals production for the last 20 years (*) ...................................................... 48 Table 3.2: Comparison of the different forms of food insecurity ............................................................... 51 Table 3.3: Characteristics of food insecurity .............................................................................................. 54 Table 4.1: Characteristics of rural households ............................................................................................ 63 Table 4.2: Total household income by source and welfare quintile ........................................................... 67 Table 4.3: Distribution of agricultural production (by value) by area of residence and type of crop ......... 71 Table 4.4: Characteristics of rural agriculture by welfare quintile ............................................................. 74 Table A1. Growth Inequality Decomposition ............................................................................................. 87 Table A2. Sectoral Decomposition of a Change in Poverty Headcount ..................................................... 87 Table A3. Ranking of regions using the poverty headcount (from the least poor to the most poor region) .................................................................................................................................................................... 88 Table A4. Regression of the logarithm of per capita consumption............................................................. 88 Table A5. Probit model of FIES food insecurity ........................................................................................ 92 Table A6. Regression on calories consumption .......................................................................................... 94 Table A7. Regression on the dynamic food insecurity ............................................................................... 96 Table A8. Regression on the agricultural productivity ............................................................................... 97 Table A9. Regression on the the transfers .................................................................................................. 98 Figure 1.1: GDP per capita in selected African countries in 2014.............................................................. 16 Figure 2.2: First order dominance curves by survey year .......................................................................... 19 Figure 2.3: Percentage of population benefiting from the service .............................................................. 20 Figure 2.4: Comparison of between Burkina Faso and African countries on selected indicators .............. 21 Figure 2.5: Resources to eradicate poverty, in percentage of GDP ........................................................... 23 Figure 2.6: Poverty headcount by area of residence in selected African countries .................................... 25 Figure 2.7: Poverty Headcount and Number of Poor in Burkina Faso using alternative Poverty lines ...... 26 Figure 2.8: Growth incidence curve 2003-2014 ......................................................................................... 27 Figure 2.9: Index of per capita expenditure and share of total consumption .............................................. 28 Figure 2.10: Growth Poverty elasticity using alternative growth measures ............................................... 29 Figure 2.11: Agricultural Production for Main Crops (tons) 2003-2014 .................................................... 31 5 Figure 2.12: Evolution of some weather characteristics in Burkina Faso................................................... 31 Figure 2.13: Main-crop yields (Kilograms per hectare) 2003-2014 ........................................................... 32 Figure 2.14: Poverty Headcount, Percentage of the population and Percentage of poor per region .......... 38 Figure 2.15: Per Capita Consumption in 2014 by Percentile ...................................................................... 39 Figure 3.1: Geographic map of food insecurity (FIES approach) ............................................................... 46 Figure 3.2: Food insecurity incidence (FIES approach) by household characteristics ............................... 47 Figure 3.3: Balance of international cereals transactions (*) ...................................................................... 49 Figure 3.4: Annual average cereal prices (FCFA per kilogram) in some main cities in Burkina Faso ...... 50 Figure 3.5: Food insecurity (calorie intake approach) incidence in Burkina by region .............................. 51 Figure 3.6: Food insecurity incidence (calorie) by some socioeconomic and demographic characteristics .................................................................................................................................................................... 52 Figure 3.7: Annual per capita consumption of food items by food security status (calorie) ...................... 53 Figure 3.8: Incidence of shocks by place of residence................................................................................ 57 Figure 3.9: Coping strategies in Burkina Faso, 2014 .................................................................................. 61 Figure 4.1. Working population (age 15 and older) by area of residence ................................................... 64 Figure 4.2: Active population 15 years and older by main occupation and welfare quintile ...................... 65 Figure 4.3: Active population 15 years and older by secondary employment and welfare quintile ........... 65 Figure 4.4: Active population 15 years and older by main and secondary jobs and welfare quintile......... 66 Figure 4.5: Percentage of households by type of specialization and welfare quintile (*) ........................... 69 Figure 4.6: Percentage of households by type of specialization by region (*) ........................................... 69 Figure 4.7: Distribution of rural agricultural production by welfare quintile and type of crop .................. 71 Figure 4.8: Distribution of rural agricultural production by region and type of crop ................................. 72 Figure 4.9: Kernel density of logarithm of agricultural production per hectare (in FCFA/Ha) .................. 73 Figure 4.10: Share of transfers by their origin ............................................................................................ 80 Figure 4.11: Share of transfers (in the total) by motive .............................................................................. 81 Figure 4.12: Probability of receiving transfers and annual transfers by log of pre-transfers per capital annual income ............................................................................................................................................. 82 Figure 4.13: Probability of receiving transfers by head of household age ................................................. 83 Figure 4.15: Amount of transfers and total income by head of household age........................................... 84 6 List of Boxes Box 2.1: Data for poverty analysis ............................................................................................... 23 Box 3.1: Categorization of shocks affecting households .............................................................. 56 Box 4.1. Income aggregate ........................................................................................................... 69 7 EXECUTIVE SUMMARY Burkina Faso is a West African Sahelian landlocked country covering 274,200 square kilometers. In January 2015, the population was estimated at just over 17.9 million. The capital city is Ouagadougou. The country has a tropical climate with two very distinct seasons: dry and rainy. In the rainy season which lasts from May/June to September, the country receives between 600 and 900 mm of rainfall in the south, but less than 600 mm in the Sahel in the north. Despite the hard climate the country has agricultural and livestock-breeding potential that represents around a quarter of GDP (2010-2014) and provide a living for more than 80 percent of the population. Burkina Faso is the top cotton producer in Africa. The principal subsistence crops are sorghum, millet, corn, and rice. The secondary sector accounts for one-fifth of GDP, and mining in particular plays an important role in the Burkina Faso economy. The tertiary sector, comprising many micro-enterprises, accounts for 45 percent of GDP. While Burkina Faso has been successful in reducing poverty, this phenomenon is still high in the country. The objective of this report is to review the state of knowledge of the profile and dynamics of poverty and to assess the tangible achievements of Burkina Faso in the fight against poverty over the past decade, highlighting the major issues and obstacles in the march towards the twin goals. A substantial drop in poverty, but an increase in the number of poor… Burkina Faso has enjoyed real economic progress during the last 15 years. Except in 2014 and 2015, the country has been politically stable and the GDP has grown at an average annual rate of 6 percent thanks to the boost in mining, particularly gold and the cotton sector. The solid economic growth translated into a substantial drop in poverty between 2003 and 2014. Despite this sharp decline in poverty, the number of poor has not decreased. The percentage of people living below the poverty line declined from 52.7 percent in 2003 to 40.1 percent in 2014, a drop of 13 percentage points over 11 years. In addition, the poverty gap and the squared poverty gap declined as well, confirming the robustness of this trend. But Burkina Faso is a country with high demographic growth, 3.1 percent per year. This strong population growth is the consequence of high fertility (6 children per woman in 2010), while mortality is declining. The decline in poverty was not strong enough to stop the increase in the number of poor which rose from 7,012,000 to 7,473,000 between 2003 and 2019, and then slightly dropped in 2014 to 7,171,000, though still above its 2003 level. The trend in non-monetary indicators mirrors the trend in poverty. Health indicators have improved substantially with the decrease on infant mortality and maternal mortality. There is also an improved in education indicators, with girls catching-up to boys. In 2003 one third of the kids age 7 to 12 were enrolled in school, they are 55 percent in 2010; and the ratio between the enrollment rate of girls to boys has increased from 0.77 to 0.99. And the living conditions of the population is also better, even though the country is lagging in many dimensions. In 2014, half of the population lives in a house with an improved floor (cement or tile) against one third in 2003; 8 80 percent have access to safe water, but only 20 percent use electricity as a source of lighting and 5 percent use a clean source of energy (electricity or gas) for cooking. … explained by growth, less inequality and labor market mobility Changes in poverty come from solid performances in economic growth and less inequality. Over the entire period 2003-2014, the 13-point decline in poverty incidence was due 50:50 to economic growth and a decline in inequality. Burkina Faso has recorded strong growth (based on GDP) over the past 15 years, with an annual average rate of 6 percent. The performance of the primary sector, which employs most of the labor force, is modest compared to the other. Agriculture is organized around small family farms, is highly dependent on weather conditions and capital accumulation is low because. For these reasons, productivity measured by yield are modest. But the secondary and tertiary sectors got better results. In the mining sector, the boom in the production of gold and other mineral resources resulted in excellent performances of extractive industries. In addition to mining, the construction industry which benefits from increased investments from households, as well as energy, also showed good performance during the past decade. With regard to the tertiary sector, the most dynamic branches are communication and finance. As in other African countries, the communications branch benefits from the penetration of mobile phones even in remote areas. These good results led to job creation and consequently to poverty reduction. In non- agricultural sectors, the most prolific branches in terms of number of new jobs are services other than real estate and business services, trade and manufacturing industries. Communications, construction and mining, which are dynamic sectors in terms of growth, also created jobs. However, in absolute terms the number of jobs created in these branches is relatively modest, as is their share in the labor market. While jobs in business services and trade sectors show higher productivity than agricultural jobs, they are paid less than in the finance and telecommunications sectors. Indeed, new jobs in the trade sector and in other services and in manufacturing are relatively low-productivity as they are created in urban informal sector micro-enterprises. Half of the workers in non-agricultural sectors are self-employed. Thus, even in non-agricultural sectors, the majority of jobs offer modest pay. Although these jobs have helped improve peopl e’s living conditions, they have not improved as much as it would take to achieve the goal of strong poverty reduction. Another important factor underlying poverty trends is the decline of inequality. It is interesting to note that the various inequality indicators move in the same direction. The Gini index, which is the most often used, varies between zero and one, and the more it is close to one, the more inequality is high. This index decreased by 7 percentage points between 2003 and 2014. Similarly, the ratio of the consumption share of the richest 20 percent of the population to that of the bottom 20 percent of the population declines significantly from 7.8 to 5.3. The drop in inequality can be explained either by structural factors or by the result of short term economic policies. Among the structural factors there is education, which can help poor children to move to the middle class when they become adults, and better opportunities to access physical capital (credit, land, etc.). Assessing how those factors have impacted inequality in the past decade is beyond the scope of this report. But in the short term, the fact that growth has been pro-poor is consistent with policies in favor of the poorest of the population. 9 The drop of poverty is also partly explained by migration and labor market mobility. The evolution of the structure of Burkina Faso population shows a growing trend of rural migration. The proportion of rural population decreased from 84 percent in 2003 to 78 percent in 2014. While the urbanization rate in the country is lower than in other African countries, it is nevertheless growing. This urbanization benefits the largest cities, including the capital, whose population represented 9 percent in 2003 and 14 percent in 2014. One of the consequences of migration is the shift in the labor market structure. During the past 15 years, rural migration resulted in a decline in the share of population living in households whose head is a farmer in favor of households where the head works in trade or construction in urban areas. The result of the breakdown shows that migration and the labor market mobility that accompanies it account for 3 of the 13-point drop recorded in poverty between 2003 and 2014, or one-quarter of the decline. Despite the progress, it is a real challenge for Burkina Faso to achieve the SDG of eradicating poverty The developing countries have committed to achieve the Sustainable Development Goals (SDGs) and one of these goals is to eradicate extreme poverty by 2030. In the case of Burkina, this goal will be difficult to achieve. Poverty projections have been made using a number of assumptions about economic growth, transmission of this growth in terms of poverty reduction, and population growth. The key finding is that even with the most optimistic scenario of high economic growth and high poverty/growth elasticity, Burkina Faso is far from reaching the goal of eliminating poverty by 2030. In the best scenario there is still 15 percent of the population living below the poverty line while the objective is less than 3 percent. In any case the simulations highlight the efforts that Burkina Faso needs to make to significantly reduce poverty and to reverse the trend in terms of number of poor people. Indeed, it would be necessary to fulfill two conditions, one of which was not fulfilled during the past decade. First, there is a need of strong and sustained growth. The country is on the right path in this regard. Second, it is important that growth be more pro-poor. While poverty has declined, it is still widespread in the country. Remember that 40 percent of the Burkinabé still live below the poverty line. Poverty incidence in rural areas is 3.5 times higher than in urban areas. Poverty is also very high in four of the thirteen regions (Nord, Boucle du Mouhoun, Centre-Ouest, Est), regions in which at least half of the population lives below the poverty line, this proportion even climbing to 7 out of 10 people in the Nord region. Poverty also varies with the sociodemographic characteristics of the household and its head. The profile of the poor household in Burkina Faso is classic. The poor live in large households in rural areas, particularly in one of the regions (Nord, Boucle du Mouhoun, Est, Plateau Central, and Centre Nord). The head of the household works in agriculture, has no education and is a man in his 50s or older. This profile is robust when using alternative monetary poverty measures. Food insecurity is also another dimension of poverty According to the FAO, food security is assured when all people at all times have economic, social and physical access to sufficient, safe, nutritious food that meets their dietary needs as 10 well as their preferences and allows them to maintain a healthy and active life. If even one of these conditions is not met, people suffer food insecurity. This, therefore, involves many factors. The food must physically exist. People must be able to physically reach it and afford to buy it. The food must be nutritious to maintain a healthy and active life. It must offer a balanced diet. And the food must be continually available. The first of these issues—supply and shortages—can usually be gleaned from annual agricultural surveys. In this section, we address the other three aspects. The FAO uses the FIES (Food Insecurity Experience Scale) approach to measure food insecurity relative to a limited access to food and in 2014, this form of food insecurity affected nearly 38% of individuals. Individuals experiencing food insecurity according to this approach are either in a moderate situation, insofar as they were led to reduce the amounts normally consumed by skipping meals, or in a severe situation, i.e., facing famine. More than 15%, i.e., one person out of seven, is affected by a severe form and faces a virtual lack of food at certain times. This form of food insecurity is higher in the eastern regions (Est, Sahel) and in rural areas. The second approach assesses food insecurity based on the household’s calorie intake and under this definition, 43% of the people were food-insecure in Burkina Faso in 2014, with one-fourth of the urban and nearly half of the rural. This is, hence, a nutritional approach that determines how well needs are being met based on the number of calories drawn from the consumption of food products. A household is experiencing food insecurity if consumption is below 2,283 kcal per adult equivalent and per day. Anyone living in a food-insecure household is also in this situation. This form of food insecurity has a positive correlation with poverty. Food insecurity decreases with the household’s standard of living (measured by the household’s consumption per capita). It affects almost all of the poorest households in the first quintile, nearly three-quarters of those in the second quintile, and is virtually nonexistent among well-off households in the fifth quintile. Moreover, among the subpopulation that has not reached the minimum calorie level, seven out of ten people are poor. Furthermore, because rural areas are more affected by this than urban areas, and most of the population is rural, nearly nine out of ten people suffering a calorie deficit live in a rural area. The third approach of food insecurity is a dynamic one using the same definition as the previous one. Food insecurity is characterized by strong seasonal variations that most often translate into a worsening of the households’ situation. Some households are vulnerable in the sense that they may be affected by food insecurity at certain times of the year. For example, farmers have an excess of provisions right after harvest, and the situation may become difficult as time passes. One-third of people live in a situation of food insecurity in the first quarter; this figure rises to 45% in the second quarter, nearly 42% in the third, and nearly 47% in the last quarter. In fact a significant proportion of households undergo a change in status. Between the first two quarters, more than one-fourth live in households that have undergone a change in status; 18% of those experiencing food security in the first round find their situation changing for the worse, and just 7% find their situation improving. These changes in situation occur in all periods, thus revealing the level of vulnerability of Burkinabé households. 11 Food insecurity is more of a transient rather than a chronic phenomenon. The 2014 results show that only one-third of these people are not experiencing food insecurity at any given time of the year. For the two-thirds who experience this difficulty, 18% are in this situation chronically, and nearly half transitorily (i.e., once, twice or three times during the year). The chronic nature of this phenomenon is the result of extreme poverty. Of those living with chronic food insecurity, 80% are in the fifth quintile, most of whom come from the poorest households. On the other hand, the transitory nature of the phenomenon has the result of a combination of multiple factors. As noted above, agricultural production does not always meet the needs of these populations. Furthermore, price variations during the year explain the variations in real income, which may decline at certain times of the year and cause temporary food insecurity. Food insecurity is correlated with shocks, making households more vulnerable The impact of shocks is very significant; they affect poor populations the most. Burkinabé households are often hit by idiosyncratic and covariant shocks. Idiosyncratic shocks are those affecting a household (loss of job, divorce, crime, separation, etc.) in particular, and covariant shocks affect a group of households (price variations, drought, flooding, etc.), for example a village, region or even the entire country. More than two-thirds of households reported they had suffered at least one shock, most frequently of natural origin (43% of households), caused by price fluctuations (25%) or by the death or serious illness of a member of the household (17%). Other shocks are less frequent and affect less than 5% of households. Shocks affect rural populations more than urban populations. Rural households suffer more from problems associated with weather and plant diseases, meaning poor harvests. There are also events associated with price fluctuations that can be correlated with natural shocks. Since these rural households live mainly from agriculture, they are more exposed to shocks of this kind. Moreover, because the health system is poorly developed in rural areas, we also find that the incidence of shocks relating to a serious illness or death of a household member is greater there. On the other hand, events associated with the loss of a non-agricultural job or income naturally affects city households more. Shocks have a generally negative impact on all forms of food security. Price shocks have the most negative impact on household food security. The price effect lowers the calorie intake per adult-equivalent by more than 19% in cities and 18% in the countryside. In urban areas, where consumption comes from the market, an increase in food prices contributes to a reduction in real income that forces households to reduce the amount of food consumed. In rural areas, on the other hand, some households are net producers, and for them a price increase can be beneficial, but others are net consumers, and for them the situation is like that of urban residents. In any case, the weakness of the country’s agricultural production makes rural households dependent on the market, because they produce little in the way of surplus. The other type of shock that has a negative impact on a household’s food security relates to issues affecting the household such as divorce, a separation or the end of transfers sent home by a family member. Shocks of this kind have an impact mainly in urban areas where they cause consumption to drop by 18%. Better income can improve the wellbeing of rural population, but productivity of most activities is too low 12 Agriculture is the most important income source. It represents nearly 61 percent of the total rural household income. Less than half of a percentage point of this income comes from wages, so the total agricultural income is from farming. In Burkina Faso farms are small and the production is for self-consumption, so the biggest part of this income is in nature, not so much cash except for those who grow cash crops. Non-agricultural activities account for 36 percent of income, two-thirds of this income derived from self-enterprises. The low level of wage income (less than 7 percent) reflects the rarity of wage earners in countryside. Other income represents just 3 percent of total income, most of it from private transfers. In particular, it is interesting to note the scarcity of public transfers in a country where households are vulnerable to many hazards (climate, shocks, etc.). Just for comparison, the distribution of income at national level shows that 41 percent of total national income comes from agriculture and 53 percent from non- agricultural activities. Low productivity in agriculture is explained by the weak capital access to capital and the low human capital. Agriculture is a high risky activity in Burkina Faso because of weather adverse conditions and other multiple shocks. In addition rural households face several poverty traps which hamper their ability to improve productivity. First, agriculture is not mechanized whereas equipment has a real impact on productivity. And the imperfection of the credit market makes it difficult to borrow and acquire equipment. Only 11 percent of households have a bank account and the poorest households are more penalized. The second poverty trap is the low use of fertilizer and pesticide, which also has a negative impact on agricultural productivity. Access to labor input is better, but even that is not optimally used by households. The third point is the specialization in households. Most of the areas cultivated are mainly used for dry cereals, crops with a negative impact on productivity. Cotton, rice and tubers have a better impact on productivity and are probably a pathway to improving it; of course it can be worth exploring other potential high productivity crops like fruit and vegetables. Finally households have limited access to market. Half of households have to walk more than an hour to find transportation and 38 percent are more than an hour from the nearest road. In such conditions, even if farmers were able to produce a surplus, they would have difficulty getting it to market and selling it at a fair price. Productivity in rural non-farm enterprises is also low, because of the small size of the firms. While productivity is positively correlated with the size of the firm, non-farm enterprises also operate at a small level, and working conditions are precarious. The main place of business is outdoors, either a specific spot by the side of the road or a market place or as a street vendor. One third of the enterprises operate at home and only 7 percent own a specific business premises. In addition to the absence of a business premises and basic commodities, the start-up capital of the average enterprise is 80,000 FCFA (less than $150) and consists essentially of tools and basic equipment. Less than 3 percent of enterprises have machines, less than 6 percent have motorbikes and automobiles, and less than 1 percent have furniture. At this low level of business, it is difficult to achieve good productivity and a decent income. Burkina Faso faces multiple challenges to reduce poverty. Poverty projections show that, with the current trend, the country will not be able to reach one of the twin goals, which is eradicating poverty by the year 2030. The first challenge in poverty reduction is demography. Burkina Faso has very rapid population growth, around 3 percent a year. High fertility rates are a 13 real challenge for growth and poverty reduction and getting a better understanding of the determinants of fertility and the channels by which it can be reduced is a path for better results on poverty reduction. The second challenge is education. Education improves human capital and has a positive impact on income and on poverty reduction. Education and in particular women’s education has a positive impact on many other phenomena, including the use of contraceptives and fertility, under-nutrition, etc. The third challenge for poverty reduction in Burkina Faso is improving productivity, in agriculture in particular. This sector is the main income source for the vast majority of the population. But performance in the sector is poor. The whole challenge is how to reach the poorest farmers who are usually in remote areas and are even less productive. Improving productivity might need to improve irrigation, access to credit, implement modern agricultural techniques, diversify crops toward value-added ones, improve access to market and invest on research. The fourth challenge is to increase the resilience of the population by implementing safety nets. In the absence of social security mechanisms, households rely on their own resources to cope with adverse situations. A system of safety nets would prevent people to fall into deep poverty and keep their dignity. 14 1. COUNTRY CONTEXT The purpose of this report is to review the state, and trends, of poverty and vulnerability in Burkina Faso and to evaluate the possible effects of development strategy policies (SCADD - Strategy de Croissance Accélérée et de Développement Durable) on poverty and social development. This Poverty Review provides detailed analyses of the "micro" environment of poor households in Burkina Faso and how they are affected by specific economic policies. The report has two complementary objectives: (i) review the state of knowledge of the profile and dynamics of poverty in Burkina Faso; (ii) assess the tangible achievements of Burkina Faso in the fight against poverty over the past decade, highlighting the major issues and obstacles in the march towards the twin goals. This will inform the authorities of the potential benefits of specific policies and how to improve the targeting of some others. The report will also provide input to the ongoing Systematic Country Diagnostic (SCD). Burkina Faso is a West African Sahelian country covering 274,200 square kilometers. It is bordered by Mali in the north; Niger in the east; Benin in the southeast; Togo and Ghana in the south; and Côte d’Ivoire in the southwest. The country is diverse with national languages spoken including Moore, Dioula and Foufouldé, but French is the official language. The capital city is Ouagadougou. In January 2015, the population was estimated at just over 17.9 million. Three climate zones can be defined, the Sahel in the north, the Sudan-Sahel in the middle, and the Sudan-Guinea in the south. The country has a tropical climate with two very distinct seasons: dry and rainy. In the rainy season which lasts from May/June to September, the country receives between 600 and 900 mm of rainfall in the south, but less than 600 mm in the Sahel in the north. Burkina Faso is part of the West African Monetary and Economic Union (UMEOA) and has adopted the CFA Franc. Politically, Burkina Faso has just ended a difficult period of transition and a new President, Roch Christian Kaboré has been elected. The country was ruled by President Blaise Compaoré from 1987 to 2014. During the 27 years of his rule, elections were regularly organized and the President was reelected. Even though the results of the elections were somewhat disputed, the country enjoyed stability and economic progress. Because of his desire to amend the constitution and seek re-election, the President was ousted from power by a popular youth uprising on October 31, 2014 and Michel Kafando was appointed Acting President for a transitional period. Presidential and parliamentary elections were organized at the end of this period and the new President took office on the eve of 2016. Despite the hard climate the country has agricultural and livestock-breeding potential that represents around a quarter of GDP (2010-2014) and provide a living for more than 80 percent of the population. Agriculture is essentially rain-fed, with irrigation representing less than 0.5 percent of the 3.3 million hectares of cultivated land. Burkina Faso is the largest cotton producer in Africa. The principal subsistence crops are sorghum, millet, corn, and rice. The secondary sector accounts for one-fifth of GDP, and mining in particular plays an important role in the Burkina Faso economy. The country’s natural resources include manganese, limestone, marble, phosphates, and gold. Gold production increased from 23 tons in 2010 to 32 tons in 2011, making Burkina Faso the fourth-largest gold producer in Africa, after South Africa, Mali and Ghana. The tertiary sector, comprising many micro-enterprises, accounts for 45 percent of GDP. This sector consists mainly of trade, telecommunications, transport and some other services. 15 But the Burkina Faso economy suffers two major constraints. First, the country is landlocked and is far from any seaport. The capital city, Ouagadougou is 930 kilometers from Tema in Ghana, 960 kilometers from Lome in Togo, 1040 kilometers from Cotonou in Benin and 1130 kilometers from Abidjan in Côte d’Ivoire. Transaction costs are high and they naturally impact prices, particularly prices of inputs. Second, the climate is dry and the country is poorly-watered, making agriculture, the principal activity of the population, difficult. Figure 1.1: GDP per capita in selected African countries in 2014 7000 6000 5000 4000 3000 2000 1000 0 South Sudan South Africa Rwanda Tanzania STP Senegal Botswana CAR Niger Burkina Faso Benin Zambia DRC Togo Mali Uganda Sierra Leone Chad Cameroon Burundi Comoros Mauritania Sudan Lesotho Namibia Malawi Guinea Guinea-Bissau Ethiopia Madagascar SSA Liberia Zimbabwe Kenya Ghana Cote d'Ivoire Mozambique Source: WDR The poor performance of agriculture, which feeds the vast majority of the population, makes Burkina Faso a poor country. In 2014, per capita GDP was $690.4 in PPP, making it one of the poorest countries in the world when using this indicator. The poverty rate, using the international poverty line of $1.90 (in 2011 PPP), was 45.4 percent in 2014. The country’s population lacks basic commodities like electricity. Social indicators, even though they have improved during the last decade, are still not good. This low level of welfare negatively impacts social outcomes, the life expectancy at birth being less than 57 years. To make the situation worse, the country is subject to multiple shocks, including climate disasters. According to the AAP, as a low-income, landlocked country with limited natural resources, Burkina Faso will experience some of the worst impacts of climate change. It faces very challenging changes in temperature, rainfall, storms and extreme weather events, which will compound the low agricultural productivity that continues to constrain the country’s growth.1 This negative impact on growth affects primarily the poorest segment of the population making them more vulnerable in a country with no social protection mechanisms. 1 Africa Adaptation Program (AAP). 16 The report is structured as follows. Section 2 presents the state of poverty, the trend in the period 2003-2014, and the poverty profile and determinants in 2014. In section 3, we provide an analysis of food security and vulnerability. In section 4, we discuss the correlation between income diversification and poverty. The last section concludes. 17 2. PROGRESS IN POVERTY REDUCTION IN BURKINA FASO 2.1 Poverty trends 2.1.1 Poverty trends 2003-2014 Burkina Faso has enjoyed real economic progress during the last 15 years. Except in 2014 and 2015, the country has been politically stable and the GDP has grown at an average annual rate of 6 percent thanks to the boost in mining, particularly gold and the cotton sector. The solid economic growth translated into a substantial drop in poverty between 2003 and 2014. The percentage of people living below the poverty line declined from 52.7 percent in 2003 to 48 percent in 2009, and 40.1 percent in 2014. This corresponds to a drop of 13 percentage points over 11 years. The poverty headcount alone does not show the whole poverty picture and the poverty situation is better presented when adding other indicators. These indicators are the poverty gap (which measures the difference between the average consumption of the poor and the poverty line) and the squared poverty gap (which measures the degree of inequality among the poor) and are used to supplement the analysis of poverty trends. These indicators are consistent with the poverty headcount, which shows that the decline in poverty in Burkina Faso has been robust over the period. Moreover since the poverty gap and the squared poverty gap put more weight to the very poor, their decrease at the national level indicate that to some extent the poor have benefited from the well-being improvement. Despite this sharp decline in poverty, the number of poor has not decreased. Burkina Faso is a country with high demographic growth (3.1 percent according to the 1996 and 2006 population censuses). This strong growth is mainly due to high fertility and has been steady for 40 years (with 6 children per woman in 2010, Burkina Faso fertility rate is one of the highest in the world), while mortality is declining.2 The decline in poverty was not strong enough to stop the increase in the number of poor which rose from 7,012,000 to 7,473,000 between 2003 and 2009, and then slightly dropped in 2014 to 7,171,000, though still above its 2003 level. Figure 2.1: Poverty headcount and number of poor per year and area of residence 80.0 8000 60.0 6000 40.0 4000 20.0 2000 0.0 0 Urban Rural Total Urban Rural Total Urban Rural Total 2003 2009 2014 Number of poor (1000) Poverty headcount Source: Author’s calculations using INSD surveys QUIBB -2003, EICVM-2009, EMC-2014 2 Zakaliyat Bonkoungou, Firmin Nana, Boureiman Zongo (2011), Should we fear population growth in Burkina Faso? Sixth International Conference on Population: African Population: Past, Present and Future, 5-9 December 2011, Ouagadougou - Burkina Faso 18 The use of an alternative approach to assess poverty trends confirms the decline in poverty. Poverty comparisons are done using different assumptions when constructing poverty lines. The dominance approach provides more robust results because the comparison is valid regardless of poverty line. The application of this technique to the 2003, 2009 and 2014 surveys confirms the decline in poverty. The graph of the first-order dominance shows that the curve relating to 2014 is below that of 2009 which is below that of 2003. This means that the poverty level for 2003 is the highest and that of 20143 is the lowest. Furthermore, for all other FGT poverty indicators, there was a decline between 2003 and 2014. Figure 2.2: First order dominance curves by survey year FGT Curves (alpha=0) 1 .8 .6 .4 .2 0 3735.953 202988.8 402241.6 601494.4 800747.2 1000000 Poverty line (z) 2003 2009 2014 Source: Author’s calculations using INSD surveys QUIBB -2003, EICVM-2009, EMC-2014 The trend in non-monetary indicators mirrors the trend in poverty. Health indicators have improved substantially. Using the most recent DHS surveys for 2003 and 2010, infant mortality has decreased from 91 per 1000 to 65 per 1000 and mortality of children less than five from 168 per 1000 to 125 per 1000. Maternal mortality follows the same trend, from a level of 440 for 100,000 births in 1998 to 341 for 100,000 in 2010. These trends are somehow linked with a better access to health services. For example the vaccination rate of children between 12 and 23 months has improved from a low 39 percent in 2003 to 81 percent in 2010. There is also an improved in education indicators, with girls catching-up to boys. In 2003 one third of the kids age 7 to 12 were enrolled in school, they are 55 percent in 2010; and the ratio between the enrollment rate of girls to boys has increased from 0.77 to 0.99. The literacy rate of the adult population (age 15 and higher) has increased from one fifth to one third during the same period, due to a better achievement of the young generation. But the probability of being at school after the completion of primary school is still low, only a quarter of the 13-16 years old are still enrolled at school and only 5 percent of those kids aged 17 to 19; the school life expectancy is still less than 6 years. This low level of achievement is a real constraint to 3 A first-order dominance implies that whatever additively separable poverty indicator is considered, poverty measured with this indicator declines. Thus, poverty does not decline only according to incidence, but also according to the other FGT indicators such as the poverty gap and the squared poverty gap. 19 improve human capital. Most of the kids dropping from school before the age of 16 years of aged will barely acquire the necessary skills to compete in the labor market. The living conditions of the population is also better, even though the country is lagging in many dimensions. In 2014, half of the population lives in a house with an improved floor (cement or tile) against one third in 2003; 80 percent have access to safe water, but only 20 percent use electricity as a source of lighting and 5 percent use a clean source of energy (electricity or gas) for cooking. One fifth of the individuals live in a household where a television set is present in in 2014 against one tenth in 2003; half of the individuals live in a household where there is a motorbike against one fourth in 2003, but less than 3 percent live in a household where there is a car. Figure 2.3: Percentage of population benefiting from the service Selected indicators of weel-being (percentage of population) 100.0 80.0 60.0 40.0 20.0 - 1 2 3 4 5 All 1 2 3 4 5 All 2003 2014 Dwelling with an improved floor Using a source of safe water Using electricity for lighting Using electricity or gas for cooking Owning a TV set Owning a cell phone Enrollment rate (aged 7-12) Enrollment rate (aged 13-16) Literacy rate (15 and older) Source: Author’s calculations using INSD surveys QUIBB-2003EMC-2014 Although there is improvement in non-monetary dimension of wellbeing, the country still faces widespread deprivation as it can be seen by comparing with other countries. Data from the World Development Indicators (WDI) and appropriate charts are used for this comparison (Figure 2.4). Each of the graphs shows the level of the indicator function of the Gross National Income (GNI). In the six dimensions of well-being considered but access to safe water, Burkina Faso is less efficient as it should according to its GNI. Access to electricity, improved sanitation and literacy are among the lowest in Africa, far behind the regional average. Regarding education, 60 percent of the current generation is enrolled in primary school and among them 60 percent achieved it, meaning that only one third of a generation achieved primary school. The combination of high monetary poverty and non-monetary poverty makes life very difficult for Burkina population. 20 Figure 2.4: Comparison of between Burkina Faso and African countries on selected indicators Source: Author’s calculations using WDI data base But the improvement in many living conditions dimensions benefits both poor and non- poor, reinforcing the quality of the progress achieved so far in the country. For example while 15 percent of the poorest (20 percent) lived in a house with an improved floor at the beginning of this century, this percentage has doubled. In 2014, 7 percent of the poorest (20 21 percent) live in a household with electricity against 1 percent 11 years earlier. Also 5 percent of the poorest (20 percent) enjoy to live if a household with a TV set, 11 years back none in this category of the population was able to afford this commodity. It is however important to note that the gap between the poorest and the wealthiest is still very important and as stated before, the country and particularly the poor are lagging in many dimensions of living conditions. The low access to services might come from the absence of the service (lack of supply), the high prices or other regulations preventing households to use a service even if it exists (lack of demand) or a combination of both factors. In the case of Burkina, rural households do not have access to many services like electricity, tap water, pointing out the weakness of supply. The situation is the same in the education sector with secondary schools being located far from many villages. In this situation, a kid without a family member in the neighbor city where the school is located might drop even if he has the potential to continue. That being said, even if urban areas where some of those services are available, there are not always affordable for the poorest, showing that the demand is also a concern4. Besides, the amount of resources needed to eradicate poverty, measured as a percentage of the GDP is declining sharply. The calculation of the resources needed to eradicate poverty is interesting by itself because it produces a benchmark to assess the efforts made in this regard. The results show that in 2003, an average 30,408 CFA F was needed to get a poor person out of poverty. This amount was 42,575 CFA F in 2009 and 37,053 CFA F in 2014.5 Given the number of poor in 2014, the amount of extra resources required to free Burkina Faso of poverty was 265.7 billion CFA Francs, or 14.5 percent of the national budget. The required effort, if measured as a percentage of GDP, decreased by half between 2003 and 2014, from more than 8 percent to less than 4 percent. This significant decline was not due to a decrease in the number of poor, but to economic growth. In a very poor country like Burkina Faso, social safety nets is one of the channel which can be used to improve the well-being of the poorest and this benchmark provides some food for thought in this regard. Spending in social safety is low in Burkina Faso. Public transfers come usually to mitigate a specific shock, in particular weather shocks or as recently to address refugees’ issues. There is a couple of recent social safety net programs, but they target only 40000 households, less than 2 percent of the population. Part of poverty can be reduced by allowing some funds to intelligent-targeted social safety nets programs. 4 See for example Komives, K., V. Foster, J. Halpern, and Q. Wodon, with support from R. Abdullah, 2005, Water, Electricity, and the Poor: Who Benefits from Utility Subsidies? , World Bank, Directions in Development, Washington, DC. The authors show in this study that bringing infrastructures to the poor goes beyond simple access; without a clear demand policy, infrastructures might exist, but the poor won’t use them because they are not affordable. 5 Calculations are made in current CFA F. 22 Box 2.1: Data for poverty analysis Three household surveys are used for this study: the 2003 QUIBB, the 2009 EICVM and the 2014 EMC. The surveys were conducted by the National Institute of Statistics and Demography (Institut National de la Statistique et de la Démographie - INSD) with the objective of producing sound data for poverty analyses. The sample size of each of the surveys (8,500 households in 2003, 8,404 in 2009 and 10,511 in 2014) are representative at regional level. Concepts and definitions are very similar across all the surveys, making most of the indicators comparable. However, consumption data (the main ingredient of welfare) collection methods were somewhat different, making it difficult to compare poverty indicators over time.6 Statistical techniques were used to obtain better comparability; nevertheless, there are still problems, especially with trends at regional level. Based on these three surveys, a person is poor if he/she lives in a poor household. A household is poor if the annual per capita consumption is below 87,837.3 CFA F in 2003, 130,735.3 CFA F in 2009 and 153,530 CFA F in 2014. In per capita consumption per day, the values are respectively 241 CFA F, 358 CFA F and 421 CFA F. Figure 2.5: Resources to eradicate poverty, in percentage of GDP 10 9 8 Pecentage of GDP 7 6 5 4 3 2 1 0 2003 2009 2014 Source: Author’s calculations using INSD surveys QUIBB-2003, EICVM-2009, EMC-2014 Poverty trends highlight a dichotomy between urban and rural areas, with a steady decline in rural areas and contrasting trends in urban areas. In cities, poverty indicators increased somewhat between 2003 and 2009 before significantly declining between 2009 and 2014. For example, the incidence of poverty increased by three percentage points in 2009 (compared to 2003) to 27.9 percent, before declining by half in 2014 to 13.7 percent. This decline in poverty should nevertheless be read cautiously, given the differences in survey methodologies as noted above. However in rural areas, the poverty incidence showed a monotonic decrease from 57.9 percent to 47.5 percent over the period 2003-2014. The combination of a sharp decline in poverty in urban areas and a low urbanization rate results in a higher concentration of poor 6 Consumption (food and non-food) is the main ingredient in constructing a welfare aggregate for poverty measurement. Consumption data in the three surveys have been collected in different periods of the year, with different numbers of visits to households (one for the first two and four for the last one), and different recall periods (usual month for the first, three days recall for the second, 7-days recall for the third); each one of these issues has an impact on consumption measurement. 23 people in rural areas. In 2014, nearly 80 percent of the population were still rural, with more than 90 percent of the poor living in rural areas. Thus, the potential for poverty reduction in the country relies on rural-oriented policies. Table 2.1 Poverty indicators by area of residence Poverty Poverty Squared poverty Percentage Percentage headcount gap gap population poor 2003 Urban 24.6 6.7 2.6 15.5 7.2 Rural 57.9 20.4 9.5 84.2 92.8 Total 52.7 18.3 8.4 100.0 100.0 2009 Urban 27.9 7.8 3.2 18.7 10.9 Rural 52.6 17.4 7.8 81.4 89.1 Total 48.0 15.6 7.0 100.0 100.0 2014 Urban 13.7 2.9 0.9 21.8 7.5 Rural 47.5 11.6 4.0 78.2 92.5 Total 40.1 9.7 3.3 100.0 100.0 Source: Author’s calculations using INSD surveys QUIBB -2003, EICVM-2009, EMC-2014 The low level of poverty in urban area is questionable, because of problems of comparability between the surveys. Poverty in urban area is three times less than in rural areas and it is unusual in the Africa region. In a sample of 11 countries, this happens only in Cameroon, Niger and to a lesser extent Mauritania and the Republic of Congo. In Benin and DRC, poverty numbers in urban and rural areas are very close. In Mali, Senegal and Togo, poverty is higher in the countryside but the difference with urban areas is of a magnitude of one to two. To show how low the level of poverty in urban Burkina Faso is, one can for example compare it with urban Côte d’Ivoire where poverty headcount is twice higher. In fact the comparison between countries is difficult, because of differences in measurement tools (survey methodology, welfare aggregate, or poverty line). But even without comparing with other countries, a couple of specific factors can explain the low level of poverty in urban Burkina Faso, namely the survey design, the construction of the consumption aggregate and that of the poverty line. A number of studies have shown that the survey design can have a large impact on poverty numbers7. Survey design includes sampling technique, questionnaire design and in particular the list of consumption items, method of data capture (diary or recall), period of the year the data are collected, etc. A work has been done to improve sampling resulting to a light revision of 2003 and 2009 poverty numbers. The other issue might be a low poverty line and we explore that. 7 See for example Beegle, Kathleen, Joachim De Weerdt, Jed Friedman and John Gibson. 2010. “Methods of Household Consumption Measurement through Surveys: Experimental Results from Tanzania.� World Bank, Policy Research Working Paper, No. 5501. Washington, DC 24 Figure 2.6: Poverty headcount by area of residence in selected African countries 80 Poverty headcount by area of residence in selected countries 70 60 50 40 30 20 10 0 Benin 2015 Burkina Cameroon Congo Cote DRC 2012 Mali 2011 Mauritanie Niger 2014 Senegal Togo 2015 2014 2014 2011 D'ivoire 2014 2011 2015 Urban Rural All Source: Poverty Reports of different countries Alternative poverty lines are used to assess the sensitivity of urban poverty numbers. In addition to the original poverty line for each of the years (2003, 2009, 2014), two alternatives are considered. The first one is the original poverty line scaled-up by 10 percent. Since the 2014 Burkina Faso poverty line is very close to the World Bank extreme poverty line ($1.9 a day in 2011 PPP), increasing this line by 10 percent seems a reasonable hypothesis. The second poverty line is choose to be equal to the Niger poverty line. In 2014 Niger poverty line was 23 percent lower than the one in Burkina Faso, however Niger GDP per capita is lower. In general poverty lines are correlated with the level of income in the country, so the second alternative is to use the Niger national poverty line in Burkina Faso in 2014. On this second alternative, the lines of the two other years (2003, 2009) are adjusted accordingly, by 23 percent. The poverty lines are in the table below. Using these new poverty lines provide higher poverty indicators, but the differences between urban and rural areas remain important. When using the first alternative, the decrease in poverty is 11 percentage points between 2003 and 2014, compared to the original 13 percentage points. But the decrease in urban poverty is still substantial, 16 percentage points between 2003 and 2014 (but starting at a higher poverty level in urban areas). As for the third alternative, it provides similar results with a higher poverty headcount in particular in urban areas, 24 percent. With this last hypothesis when the poverty line is adjusted to be equal to the one of Niger, the number of poor exceeds 10 million, with nearly one million in urban areas. This analysis helps draw two conclusions. First, the Burkina Faso poverty line is somehow low and future work needs to fix this problem. Second, the fact that poverty numbers increase nearly in the same proportion nationally and in urban areas show that the welfare distribution is very similar around these different poverty line in urban and rural areas. The problem of the high drop of poverty between 2009 and 2014 might also come from the design of the survey. The 2003 survey used the usual month approach to capture data, the 2009 used the diary (one day recall) and the 2014 7-day recall8. It has been shown that the diary is less comparable to the two other 8 Prospere Backiny-Yetna, Diane Steele, Ismael Yacoubou Djima, “The Impact of Household Food Consumption Data Collection Methods on Poverty and Inequality Measures in Niger �, 25 approaches, and this can also explain why 2009 seems so especial, to keep a minimum of consistency, the comparison is done more between 2003 and 2014. Table 2.2: Actual and alternative poverty lines H2: Actual scaled-up H3:Niger PL H1: Actual PL 10% 2003 87837 96621 108130 2009 130735 143809 160939 2014 153530 168883 189000 Source: Author’s calculations Figure 2.7: Poverty Headcount and Number of Poor in Burkina Faso using alternative Poverty lines Poverty Headcount in Burkina Faso using Number of Poor in Burkina Faso using alternative Poverty lines alternative Poverty lines 80.0 12000 70.0 10000 60.0 50.0 8000 40.0 6000 30.0 4000 20.0 10.0 2000 0.0 0 H1: Actual PL H1: Actual PL H2:Actual PL scale-up 10% H2:Actual PL scale-up 10% H3:Niger PL H3:Niger PL Source: Author’s calculations using INSD surveys QUIBB-2003, EICVM-2009, EMC-2014 2.1.2 Understanding poverty trends 2.1.2.1 Growth and Inequality Changes in poverty come from two sources: growth and inequality. Growth is the consequence of wealth creation, either through new investment or gains in the productivity of existing production factors. Inequality is the result of redistribution policies (tax and subsidies, transfers, etc.). Any one of these elements can contribute to a decrease or an increase in poverty, where inclusive growth with a decrease in inequality stands out as the best combination to reduce poverty. 26 Over the entire period 2003-2014, the 13-point decline in poverty incidence was due 50:50 to economic growth and a decline in inequality. Table A1 (in Annexes) shows the contribution of each of the two component to the evolution of poverty for each sub-period 2003-2009 and 2009-2014, and the entire 2003-2014 period. Growth and inequality follow the same trend and contribute to the decline in poverty. The contribution of growth to poverty reduction is graphically illustrated by the growth incidence curves. The 2003-2014 curve shows that the growth rate of annual per capita consumption is positive at all points in the distribution. Furthermore, this growth is progressive insofar as growth rate is a monotonically decreasing function of welfare (again measured as annual per capita consumption). Understanding the contribution of growth and inequality can be refined when we consider the place of residence. In both urban and rural areas, growth has contributed to poverty reduction. The profile of rural areas is the same as the national profile, as growth and inequality equally contribute to poverty reduction. However, in cities, the decline in poverty is dominated by inequality, with growth playing a smaller role. Figure 2.8: Growth incidence curve 2003-2014 Growth incidence curve 2003-2014 - National 100 Growth rate 50 0 0 20 40 60 80 100 Centiles Source: Author’s calculations using INSD surveys QUIBB -2003, EICVM-2009, EMC-2014 The question as to how growth leads to poverty reduction is important for public policy. The idea is to identify pro-poor policies. However, growth as measured by national accounts does not always coincide with that of household surveys. At least two factors explain the discrepancies between these two sources. First, the concept of household consumption does not correspond to the exact same reality in both cases (for example, durable goods are not treated similarly)9. Second, there are measurement errors in national accounts (issues related to coverage of informal sector just to cite an example) as well as in household surveys (sampling and data collection errors, etc.). That is why it is important to first examine the consistency of growth from the two sources. 9 For national accounts, the purchase of a durable good (like a car) is considered consumption, while surveys treat it as investment. 27 Burkina Faso has recorded strong growth (based on GDP) over the past 15 years, with an annual average rate of 6 percent. This rate was 5.4 percent for the sub-period 2003-2009 and 6.4 percent for the sub-period 2009-2014. Private consumption (as defined in national accounts) growth is not as high as GDP growth. The average annual growth rate of private consumption was 4.4 percent over the period; i.e.1.3 percent growth per capita. The growth of per capita consumption as measured by household surveys is lower (0.7 percent annual on average between 2003 and 2014). Despite this difference, the national accounts and surveys reflect the same reality: the situation of households improved in the period 2003-2014, but not by the same magnitude. Table 2.3: Comparison of national accounts and household survey growth rates 2003/2014 Average annual growth rate using national accounts GDP 5.85 GDP per capita 2.75 Private consumption 4.41 Private per capita consumption 1.31 Average annual growth rate using household surveys Total household consumption 2.17 Per capita household consumption 0.74 Source: Author’s calculations using INSD data Figure 2.9: Index of per capita expenditure and share of total consumption 150 100 50 0 2003 2009 2014 2003 2009 2014 Index of Per capita expenditure (100 in 2003) Share of total consumption (%) Bottom 40 percent Top 60 percent Source: Author’s calculations using INSD survey EMC-2014 The growth recorded in Burkina Faso during this period improved shared prosperity. The per capita consumption of the bottom 40 percent of the population increased in real terms by 32 percent between 2003 and 2014 (on average 2.5 percent a year), from 47,000 CFA F to 63,000 CFA F. This increase has been both for urban populations and rural ones, being even more important in rural areas. Also consumption grew faster among the bottom 40 percent than in the rest of the population, resulting in a catch-up effect in the distribution. The consumption share of the bottom 40 percent was 17 percent in 2003 and 20 percent in 2014. 28 Figure 2.10: Growth Poverty elasticity using alternative growth measures 0 2003/2009 2009/2014 2003/2014 -1 -2 -3 -4 GDP per capita Private consumption Household Survey Source: Author’s calculations Furthermore, poverty/growth elasticity confirms the positive impact of growth on poverty. However, this effect varies significantly depending on how growth is measured. If using GDP, the average elasticity for the period 2003/2014 is -0.7; which means that one percent growth helps reduce poverty by 0.6%. This elasticity is around -1.6 if growth is measured by national private consumption accounts, and is close to -3 if we consider consumption from household surveys. In the first case, poverty responds positively to growth to a rather modest extent, while in the latter case, poverty is very sensitive to growth. This reflects a statistical difficulty that can only be resolved with more comparable surveys. The economic growth recorded between 2003 and 2014 was mainly driven by the tertiary and secondary sectors. The tertiary sector accounted for less than 40 percent of GDP before the 2000s, and for the recent decade this sector accounts for 45 percent of the wealth created in the country. The weight of the secondary sector is stable, around one-fifth, and the re-composition of the GDP structure was done at the expense of the primary sector, which accounted for more than 30 percent before 2000, but has been reduced to a quarter in recent years. Table 2.4: Average GDP growth rates 2003-2009 2009-2014 2003-2014 Primary sector 2.39 5.76 3.91 Subsistence farming 0.73 3.92 2.17 Commercial agriculture 0.01 6.27 2.80 Livestock breeding 4.08 1.18 2.75 Forestry, fishing and hunting 6.27 19.51 12.10 Secondary sector 4.25 6.10 5.09 Extractive industries 35.81 20.92 28.82 Textile industries -1.19 9.70 3.62 Other manufacturing -0.19 -0.92 -0.52 Energy (electricity, gas, water) 8.58 2.55 5.79 Building and civil industry 6.23 8.49 7.25 Tertiary sector 7.22 6.66 6.96 Transport network -0.16 10.93 4.74 Post and telecommunications 16.62 22.81 19.39 Trade 9.52 1.31 5.71 29 Banks and insurance 8.85 12.24 10.38 Other market services 8.18 -3.68 2.62 Government and NPIs 5.39 5.79 5.57 Other non-market services 3.08 5.12 4.00 FISIM 10.54 14.07 12.13 Duties and taxes 9.30 12.36 10.68 GDP 5.36 6.44 5.85 Source: Author’s calculations using INSD data The performance of the primary sector, which employs most of the labor force, is modest making it difficult for the country to get better result in poverty reduction. Agriculture is organized around small family farms and is highly dependent on weather conditions. The capital accumulation is low because this agriculture is not mechanized and farmers have limited access to fertilizers and modern input. In addition, human capital is low as well and most farmers use rudimentary techniques. Burkina Faso’s agriculture is mainly food-crop oriented including sorghum, maize, millet, cowpeas, rice and peanuts. However, cotton and more recently sesame are relatively important cash crops. While sesame and maize have good performance, the production of other crops is somewhat stagnant in the best case. For example, Burkina Faso ranks among the world's top cotton producers and is the leading African producer, with production peaks exceeding seven hundred thousand tons in 2006 or 2014. However, the production also happens to fall below three hundred thousand tons in 2010 with producers turning to alternative crops because of low prices. As high levels of production do not always correspond to high prices on the global market, the sector struggles to maintain high performance in the long term. The performance of agriculture depends highly on the weather. As it can be seen on the figure below describing weather trends, years like 2003 and 2014 were good. The average number of raining days and rainfall are high. On the contrary, 2009 was a year of bad weather. These weather patterns are in parallel with the performance of the agriculture sector. The production was relatively low in 2009 but high in 2003 and 2014. For example cotton production increased by more than 40 percent between 2009 and 2014, and this is not due only to the extension of cultivated areas. Cereal production (particularly maize) show a similar trend during this time period. Part of the decrease in poverty during this period is explained by this trend, with 2009 being a year of low production thus low income, and 2003 and 2014 being better years in this regard. The mixed performance of agriculture in Burkina Faso and its inability to play a key role in reducing poverty and improving households’ living conditions can be viewed in parallel with productivity trends. The level of agricultural yields highlights the low dynamism of the agricultural sector. Figure 2.9 shows that changes in the yields of major crops are mixed, with good years such as 2006, 2008 and 2014 and less good years such as 2003, 2007 and 2011. In the absence of productivity gains, increase in agricultural production is driven either by population growth, or by more areas being cultivated. The absence of productivity gains in agriculture results in a low increase in per capita income and therefore a relatively lesser decline in poverty than in rural areas. 30 Figure 2.11: Agricultural Production for Main Crops (tons) 2003-2014 1800000 1600000 1400000 1200000 1000000 800000 600000 400000 200000 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 White Sorghum Red Sorghum Maize Millet Paddy Rice Cotton Groundnut Sesame Cowpea Tubers Source: Authors calculations using the 2012 “Agricultural Statistical Yearbook� and the 2014 Annual Survey report Figure 2.12: Evolution of some weather characteristics in Burkina Faso Average number of raining days in 10 Average rainfall in10 stations (in stations by year milimeters) by year 80 1200 78 1000 76 800 74 600 72 400 70 68 200 66 0 2003 2005 2007 2009 2011 2013 2015 2000 2005 2010 2015 Source: Authors calculations using the 2012 “Agricultural Statistical Yearbook� and the 2014 Annual Survey report 31 Figure 2.13: Main-crop yields (Kilograms per hectare) 2003-2014 2500 2000 1500 1000 500 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 White Sorghum Red Sorghum Maize Millet Paddy Rice Cotton Groundnut Sesame Cowpea Source: Author’s calculations using the 2012 “Agricultural Statistical Yearbook� and the 2014 Annual Survey report The significant poverty reduction in urban areas can also be viewed in parallel with the strong economic performance of the secondary and tertiary sectors. In the mining sector, the boom in the production of gold and other mineral resources resulted in excellent performances of extractive industries. Gold production, which was virtually non-existent before 2008, exceeded ten thousand tons in 2009, twenty thousand tons in 2010 and has exceeded thirty thousand tons since 2011. Besides, world prices were more favorable. Between 2009 and 2012, world prices increased by 50 percent, resulting in improved export revenues for the country. In addition to mining, the construction industry which benefits from increased investments from households, as well as energy, also showed good performance during the past decade. With regard to the tertiary sector, the most dynamic branches are communication and finance (banking and insurance). As in other African countries, the communications branch benefits from the penetration of mobile phones even in remote areas. In Burkina Faso, more than four fifths of households reported having a cell phone in 2014. Public services and the distribution sector also showed good performance. These good results led to job creation and consequently to poverty reduction. In non- agricultural sectors, the most prolific branches in terms of number of new jobs are services other than real estate and business services, trade and manufacturing industries. Communications, construction and mining, which are dynamic sectors in terms of growth, also created jobs. However, in absolute terms the number of jobs created in these branches is relatively modest, as is their share in the labor market. While jobs in business services and trade sectors show higher productivity than agricultural jobs, they are paid less than in the finance and telecommunications sectors. Indeed, new jobs in the trade sector and in other services and in manufacturing are relatively low-productivity as they are created in urban informal sector micro-enterprises. Half of the workers in non-agricultural sectors are self-employed. Thus, even in non-agricultural sectors, the majority of jobs offer modest pay. Although these jobs have helped improve people’s living conditions, they have not improved as much as it would take to achieve the goal of strong poverty reduction. 32 Table 2.5: Primary employment by year and type of industry 2003 2009 2014 Primary sector 1539316 70.2 2311838 66.8 2154911 58.7 Agriculture, hunting and forestry 1537213 70.1 2307832 66.7 2153459 58.7 Fishing, aquaculture 2103 0.1 4006 0.1 1452 0.0 Secondary sector 157106 7.2 304838 8.8 353138 9.6 Mining and quarrying 21904 1.0 23277 0.7 41637 1.1 Manufacturing 88560 4.0 215332 6.2 211781 5.8 Electricity, gas and water 14354 0.7 8479 0.2 5154 0.1 Construction 32288 1.5 57750 1.7 94567 2.6 Tertiary 497437 22.7 842960 24.4 1160790 31.6 Trade and repairs 281561 12.8 436821 12.6 581574 15.9 Hotels and restaurants 15134 0.7 52205 1.5 85831 2.3 Transport and communications 29490 1.3 44769 1.3 75515 2.1 Financial activities 10485 0.5 11526 0.3 12905 0.4 Real estate and business services 7146 0.3 19474 0.6 22926 0.6 Public administration and other services 153622 7.0 278166 8.0 382040 10.4 Total 2193859 100.0 3459636 100.0 3668839 100.0 Source: Author’s calculations using INSD data Another important factor underlying poverty trends is the evolution of inequality. This phenomenon is measured by various indicators. The reason is that inequality indicators have different properties. Some of them, such as the Gini index, are sensitive to changes that occur in the middle of the distribution (that is to say, households with a living standard around the average), while other indicators, such as Theil indices tend to be much more driven by changes among the poorest and the richest. Four inequality indices are considered: Gini, Theil(0) Theil(1) and the ratio of the consumption share of the wealthiest 20 percent of the population to that of the poorest 20 percent of the population. This set of indicators can reveal robust results about the evolution of inequality. Inequality is declining, confirming the previous results that half of the drop in poverty is due to a better redistribution of wealth. It is interesting to note that the various indicators move in the same direction. The Gini index, which is the most often used, varies between zero and one, and the more it is close to one, the more inequality is high. This index decreased by 7 percentage points between 2003 and 2014. Similarly, the ratio of the consumption share of the richest 20 percent of the population to that of the bottom 20 percent of the population declines significantly from 7.8 to 5.3. The drop is inequality can be explained either by structural factors or by the result of short term economic policies. Among the structural factors there is education, which can help poor children to move to the middle class when they become adults, and better opportunities to access physical capital (credit, land, etc.). Assessing how those factors have impacted inequality in the past decade is beyond the scope of this report. But in the short term, the fact that growth has been pro-poor is consistent with policies in favor of the poorest of the population. Table 2.6: Inequality indicators by area of residence 2003 2009 2014 33 Urban Rural Total Urban Rural Total Urban Rural Total Gini 46.2 36.3 42.3 43.6 35.0 39.8 38.4 27.3 35.3 Theil(0) 35.7 22.0 29.8 31.9 20.3 26.3 24.1 12.0 20.2 Theil(1) 41.2 26.0 36.9 37.3 22.7 31.5 26.2 13.4 24.2 Q5/Q1 9.7 6.0 7.8 8.5 5.7 7.0 6.7 3.8 5.3 Source: Author’s calculations using INSD surveys QUIBB -2003, EICVM-2009, EMC-2014 2.1.2.2 Migration and labor market mobility Migration is one of the strategies adopted by individuals and households to improve their living conditions. People who live in communities with few opportunities, especially in terms of jobs, will tend to migrate to places where more opportunities exist. However, migration has a cost, at least two in fact. On one hand there are direct costs (transportation, basic needs costs before getting a job, etc.); and on the other hand opportunity costs related to leaving an existing job or activities. The evolution of the structure of Burkina Faso population shows a growing trend of rural migration as everywhere in Africa. The proportion of rural population decreased from 84 percent in 2003 to 78 percent in 2014. While the urbanization rate in the country is lower than in other African countries, it is nevertheless growing. This urbanization benefits the largest cities, including the capital, whose population represented 9 percent in 2003, but 14 percent in 2014. One of the consequences of migration is the shift in the labor market structure. The analysis which follows is based on eight categories, four in urban areas and four in rural areas. In each of the areas, individuals are considered in terms of whether the household head: (i) is a farmer; (ii) works in manufacturing or construction industry; (iii) works in trade or service sector; (iv) is unemployed. During the past 15 years, rural migration resulted in a decline in the share of population living in households whose head is a farmer. In rural areas this share declined from 78 percent in 2003 to 67 percent in 2014, while urban agricultural population increased slightly from 3.7 percent to 4.4 percent over the same period. While the percentage of people living in agricultural households remains very high, this trend contributes mainly to increasing the percentage of people living in a household where the head works in trade or construction in urban areas; and as a rather negative signal also people living in households with unemployed heads in rural areas. In general, migration from rural areas to urban areas is from high poverty areas to low poverty areas; and the consequence is a decrease in poverty. A breakdown can be done to identify the contribution of each of the two following factors in the change in poverty: the intra- sectoral effect resulting from the poverty decline in the sector of activity where people are involved; and the decline resulting from migration to other sectors. The result of the breakdown shows that migration and the labor market mobility that accompanies it account for 3 of the 13- point drop recorded in poverty between 2003 and 2014, or one-quarter of the decline. The comparison of the labor market structure in this time period shows that the labor force has moved from agriculture to the tertiary, mainly small trade activities and other services. While those activities are not necessary highly productive either, they are better than the small rural agriculture for some populations, in particular those lack land to cultivate. 34 2.1.3 Poverty Projections (2016-2030) Sustainable Development Goals (SDGs) cover the period 2015-2030 and one of these goals is to eradicate extreme poverty by 2030. A person is considered extreme poor when he/she lives on less than 1.90 USD a day10. The assessment of this goal requires adequate poverty monitoring. For each country, an obvious question is whether or not this goal can be achieved. In cases where the goal can be achieved with the growth projections considered, it is necessary to maintain efforts to this end. In cases where the objective would be hard to achieve, it is necessary to identify constraints impeding growth and address them. An interesting feature of the Burkina Faso 2014 poverty line is that it is very close to this international poverty line, so assessing this goal using the national poverty line is the same as assessing this SDG goal. The following exercise deals with poverty projections. These projections are based on a number of assumptions about economic growth, transmission of this growth in terms of poverty reduction, and population growth. Three simulations are proposed, based on the EMC-2014 data. The population growth rates of the last three censuses are used in the three simulations (1985, 1996 and 2006)11. According to the results of the three censuses, the population is growing at an average rate of 2.73 percent per year, 5.61 percent in urban areas and 2.13 percent in rural areas. Based on these assumptions, one-quarter of the population will be urban in 2020, and this proportion will increase to one-third in 2030, indicating that the majority of the Burkina Faso population will still be rural. Economic growth assumptions are applied to each household. Four household categories are considered, based on the sector of activity of the household head (primary, secondary, tertiary and unemployed). For households with an unemployed household head, we apply the average growth rate of GDP. Table 2.7: GDP per sector (2009-2014) and projections (2015-2017) Average Average 2009/14 2015 2016 2017 2015/17 Primary 3.0 4.8 4.8 5.6 5.0 Secondary 6.8 7.4 10.7 10.2 9.3 Tertiary 6.6 4.3 6.5 6.0 5.5 All 5.8 5.3 6.9 7.1 6.4 Source: INSD With regard to the other assumptions, for the first simulation we assume that GDP grows at the same annual average rate as in the period 2009-2014. The growth rate is 5.8% annually, or around 3.1% for per capita GDP. This growth rate is applied each year from 2016 to 2030. Since growth as measured by national accounts does not always match growth as measured by household surveys, it is assumed that only half of this growth is translated into poverty reduction. This first assumption has the lowest growth rate and the lowest pass-through between growth and poverty, so it is the less optimistic assumption. 10 Strictly speaking, we should say 1.90 USD a day in 2011 purchasing power parity. 11 We prefer to use the long-term population growth over the last 20 years, rather than that of the last 10 years. The former seems more robust as it is based on three censuses. 35 For the second simulation, the growth assumptions are based on INSD projections for the period 2015-2017. As in the previous simulation, we apply this average growth rate over the whole period 2016-2030. The projected growth for this period is stronger, especially in primary and secondary sectors, though the tertiary declined somewhat. With regard to the relationship between economic growth and poverty, we adopt the same assumption as above, which is that half of this growth translates into poverty reduction. Since the growth is stronger and we apply the growth/poverty elasticity, this is an intermediate assumption. The third simulation uses the same growth assumption as the previous one, but we adopt a more favorable assumption that 80 percent of the growth is translated into poverty reduction. This is the more optimistic assumption. The results of the simulations are presented in Table 2.7. The simulations are done for the years 2016-2030. The table also exhibits the baseline situation, the year 2014. According to the first simulation, poverty will decline steadily, but slightly, between 2014 and 2030. The poverty headcount is expected to be 37 percent in 2020, 34 percent in 2026 and 32 percent by 2030. Thus, one third of the population will still live below the poverty line in 2030. The decline in poverty is constrained by two factors. The first factor, which is common for the three simulations is that despite solid demographic urban growth, most people will still live in poor rural areas by 2030. As long as this part of the country offers few opportunities, it would be difficult to significantly reduce poverty. The other factor, which is specific to this simulation, is the low poverty/growth elasticity which results in growth translating (relatively) little into poverty reduction. With this simulation Burkina Faso is far from reaching the SDG of eradicating extreme poverty by 2030. Furthermore, the number of poor will continue to increase, to exceed 8 million in 2022 and 9 million people in 2028. However, given the dynamic economic situation, the resources to eradicate poverty (when measured as a percentage of GDP and not in absolute terms) are in continuous decline. As for the second simulation, poverty drop is more significant because of higher economic growth in general and, in particular, better results for the primary sector which provides livelihoods for most of poor households. One third of the people would be poor in 2020 and one fourth would still be in this situation in 2030. Based on this simulation, the number of poor people would decline steadily to 6 million people in 2030. So the growth differential between the two assumptions (keeping the same poverty/growth elasticity) would result in 3 million fewer poor, showing the importance of strong growth in poverty reduction. The third simulation provides better results in terms of poverty reduction. Remember that this simulation uses the same GDP growth rate as the previous one, but a high percentage of growth translates into poverty reduction, 80 percent, which is clearly more pro-poor. For this last simulation, a quarter of the population would still be poor in 2022 and less than 15 percent in 2030. The number of poor people would then be 4 million, i.e., 2 million less than in the previous simulation. Furthermore, it would only take 0.7 percent of GDP in 2030 to eradicate poverty. Some interesting findings emerge from this exercise. Even with the most optimistic scenario of high economic growth and high poverty/growth elasticity, Burkina Faso is far from 36 reaching the goal of eliminating poverty by 2030. In the best scenario there is still 15 percent of the population living below the poverty line while the objective is less than 3 percent. In any case the simulations highlight the efforts that Burkina Faso needs to make to significantly reduce poverty and to reverse the trend in terms of number of poor people. Indeed, it would be necessary to fulfill two conditions, one of which was not fulfilled during the past decade. First, there is a need of strong and sustained growth. The country is on the right path in this regard. Second, it is important that growth be more pro-poor. The difference between the first and the second simulations lies in growth; a stronger growth results in poverty reduction and a reduction in the number of poor people. The difference between the second and the third simulations lies in how growth benefits the poor; it refers to the quality of growth. As poor people live in rural areas, support for small-scale agriculture and diversification of activities in rural areas is a route to explore. Table 2.8: Poverty Projections for 2016-2030 H1: Average GDP Growth rate H2: Forecast Average GDP Rate H3: Forecast Average GDP Rate 2009-14, Pass thru of 0.5 2015-17, Pass thru of 0.5 2015-17, Pass thru of 0.8 Population Poverty # Poor % GDP to Poverty # Poor % GDP to Poverty # Poor % GDP to (1000) headcount (1000) eradicate headcount (1000) eradicate headcount (1000) eradicate 2014 17900 40.1 7171 4.3 40.1 7171 4.3 40.1 7171 4.3 2016 18900 39.0 7391 3.9 37.6 7113 3.6 36.2 6861 3.5 2018 20100 38.0 7617 3.6 34.9 6998 3.1 32.8 6577 2.8 2020 21300 37.0 7868 3.3 32.7 6968 2.6 28.9 6143 2.2 2022 22600 36.1 8154 3.0 30.1 6802 2.2 25.7 5815 1.8 2024 24000 35.1 8415 2.8 27.7 6658 1.9 22.5 5401 1.4 2026 25500 34.1 8691 2.5 25.6 6532 1.6 19.4 4951 1.1 2028 27200 33.2 9026 2.3 23.4 6356 1.3 16.9 4604 0.9 2030 28900 32.3 9336 2.1 21.5 6212 1.1 14.6 4216 0.7 Source: Author’s calculations using INSD survey, EMC 2014 2.2 Poverty profile and determinants in 2014 In this section we try to answer two questions: what are the characteristics of poor households and why are they poor. As noted in previous sections, a household is poor if its annual per capita consumption is below 153,530 CFA F or 421 CFA F per day. Every individual living in a poor household is poor. 2.2.1 Basic poverty profile Poverty is, first, a geographic issue. As noted previously, poverty incidence in rural areas is 3.5 times higher than in urban areas. The high level of poverty in rural areas is correlated with low incomes in the agricultural sector, the main economic activity. The geography of poverty in Burkina Faso is complex. In some countries, there is a correlation between the poverty level of regions and their distance to the capital city. Demand, particularly in food products, is strong in big cities which are potential markets for surrounding rural areas. This does not seem to be the case in Burkina Faso where the areas surrounding the capital are not necessarily the less poor. 37 The regions of Burkina Faso can be grouped into four broad categories based on their 2014 poverty level. Poverty is very high in four of the thirteen regions (Nord, Boucle du Mouhoun, Centre-Ouest, Est), regions in which at least half of the population lives below the poverty line, this proportion even climbing to 7 out of 10 people in the Nord region. These four regions share one-third of the population of the country, but half of the poor population. They represent two poles of concentration of high poverty: one in the east with the Est region and the other in the west, with the other three regions. The second category has three regions (Centre-Nord, Plateau Central, Sud-Ouest) with moderately high poverty. Poverty headcount is above the national average. The two regions of Centre-Est (which lies in the southeast of the country and has borders with Togo and Ghana) and Hauts-Bassins in the west (region which includes Bobo- Dioulasso, the second city) have low moderate poverty incidence around 35 per cent. The last group of three regions (Cascades, Sahel, Centre) is the group with relatively low poverty levels. The Centre region includes the capital city (Ouagadougou), which has the lowest poverty incidence (below 10 per cent). An interesting aspect of regional poverty is the correlation between the degree of urbanization of a region and its level of poverty. In fact, the urbanization rate is too low in Burkina Faso. While the Centre region where Ouagadougou the Capital city is located is more than 80 percent urban; only two other regions have urbanization rates above 20 percent: the Cascades and Hauts Bassins regions. These regions are part of low or moderate poverty regions. We can presume for other regions, although this requires further analyses, that the level of urbanization is too low for small cities to play a catalytic role in poverty reduction in the regions, including providing opportunities for the rural population. Figure 2.14: Poverty Headcount, Percentage of the population and Percentage of poor per region 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Poverty Headcount Percentage of Population Percentage of poor Source: Author’s calculations using INSD surveys QUIBB -2003, EICVM-2009, EMC-2014 Households are vulnerable in Burkina Faso in the sense there is a high concentration of individuals around the basic needs threshold. The Burkina Faso 2014 poverty line is very 38 close to the international extreme poverty line which is 1.9 dollars a day in 2011 PPP;12 so we can consider this national poverty line as an extreme poverty line. For instance, a 10 percent increase in the poverty line would lead to an 8 percentage-point increase in the poverty headcount and would add 1.3 million (two hundred thousand in urban areas and 1.1 million in rural areas) more people living in poverty. The significant number of people clustering around the poverty line suggests that an important proportion of moderately poor people are positioned to move out of poverty, but also that an important proportion of non-poor people are vulnerable to falling into poverty. Figure 2.15: Per Capita Consumption in 2014 by Percentile FGT Curve (alpha=0) 1 .8 .6 .4 .2 0 0 300000 600000 900000 1200000 1500000 Poverty line (z) Source. Author’s calculations using the EMC 2014 Poverty also varies with the sociodemographic characteristics of the household and its head. The poverty incidence is higher in households that have a man as a head. Poverty headcount is 41 percent in these households against 30 percent in households with a female head. As a result, 94 percent of the poor live in a household having a man as a head because of higher poverty rates in these households and also the fact that households with men as a head are by far more significant in number. The key factor behind this result is household size. Households headed by a woman are smaller on average, almost half the size of those with a male household head. The poverty incidence is also an increasing function of the age of the head of the household: 26 percent for households whose head is between 15 and 29 years of age to nearly 50 percent for those whose head is aged 60 years or more. As previously, household size largely explains the difference in poverty incidence since there is a positive correlation between the age of the head of the household and household size. Household size for heads under 30 years of age 12 The most recent Burkina Faso national poverty line in 2014 was 153,530 FCFA per capita per year or 421 FCFA per capita per day. This poverty line is very close to the international poverty line which is $1.90 per day in 2011 PPP. Indeed using the 2011 PPP and the inflation between 2011 and 2014, the international poverty line corresponds to 426.8 FCFA per capita per day in Burkina Faso; and the $3.1 per day per capita corresponds to 626.3 FCFA per capita per day in Burkina Faso in 2014. 39 (4.6 people) is half that of heads aged 60 or above (8.6 people). Households with heads older than 50 account for 47 percent of the population, but they have 54 percent of the poor. The fact that poverty becomes higher when the head of the household is older can make households with older heads more vulnerable, most of them being forced to continue working at an age that they would expect to retire. Unlike age, the level of education of the head of the household is negatively correlated with the level of poverty. While half of the population living in a household with a head having no education are poor, only 3 percent of those in households whose head has reached upper secondary education level are in the same situation. Households whose head has no education account for more than 9 poor people out of 10. The economic sector of the head of the household is also a discriminating factor of poverty. When the main economic activity of the head is agriculture, half of the population is poor; and agricultural households account for 9 poor out of 10. Compared to agriculture, the other economic sectors have relatively low levels of poverty. The poverty profile in Burkina Faso is rather classic. The poor live in large households in rural areas, particularly in some regions (Nord, Boucle du Mouhoun, Est, Plateau Central, and Centre Nord). The head of the household works in agriculture, has no education and is a man in his 50s or older. Table 2.9: Basic poverty indicators by household characteristic Poverty Squared headcount Poverty gap poverty gap % population % poor Gender of head Male 41.0 9.9 3.4 91.6 93.8 Female 30.4 7.5 2.6 8.2 6.2 Total 40.1 9.7 3.3 100.0 100.0 Age of head Less than 30 25.9 5.4 1.6 8.0 5.2 30-39 31.9 7.2 2.3 21.1 16.8 40-49 40.9 9.9 3.3 24.0 24.5 50-59 41.7 10.6 3.8 22.1 23.0 60+ 49.6 12.2 4.2 24.7 30.6 Total 40.1 9.7 3.3 100.0 100.0 None 45.3 11.0 3.8 81.0 91.8 Primary 26.1 5.6 1.8 10.6 6.9 Lower secondary 10.7 2.5 0.8 4.1 1.1 Upper secondary 3.2 0.6 0.1 2.4 0.2 University 0.1 0.0 0.0 1.6 0.0 Type of industry Agriculture 48.2 11.6 4.0 71.5 86.2 Industry 13.7 2.4 0.7 3.3 1.1 Construction 6.7 1.5 0.4 1.6 0.3 Commerce 14.9 2.9 0.9 5.4 2.0 Restaurant/Hotel 2.6 0.3 0.0 0.6 0.0 Transport 5.6 1.1 0.3 1.2 0.2 40 Education/Health 1.7 0.4 0.1 1.2 0.1 Other services 6.2 1.3 0.4 5.0 0.8 No job 37.5 9.8 3.4 10.1 9.4 Employer Public administration 3.6 0.7 0.2 2.7 0.2 Public enterprise 1.8 0.2 0.0 0.4 0.0 Private enterprise 6.0 1.3 0.4 1.9 0.3 Individual enterprise 42.6 10.2 3.5 84.4 89.9 Household 17.9 4.3 1.3 0.3 0.2 No job 37.5 9.8 3.4 10.1 9.4 Total 40.1 9.7 3.3 100.0 100.0 Source: Author’s calculations using INSD surveys QUIBB-2003, EICVM-2009, EMC-2014 2.2.2 Assessing the robustness of the poverty profile It is important to get the poverty profile right for the poor to be targeted effectively. Three ingredients are needed for poverty comparisons: (i) a welfare indicator; (ii) a poverty line; and (iii) poverty indices, the best known being the FGT indices used in the previous sections. Since the targeting can change depending on the indicator used, in this section alternative poverty indicators are used to test the robustness of the poverty profile. For this purpose, five other monetary welfare indicators are calculated. The welfare indicator used so far is the per capita household consumption; i.e., the total household consumption divided by the household size. Alternative indicators use different definitions of adult-equivalent instead of the household size. There are at least two reasons to use an alternative to the household size. First, individual needs depend on age, sex and other biological factors. The needs of a five-year-old child are not the same as those of a thirty-year-old adult. Therefore, instead of dividing household consumption by its size, it is divided by a number of adult- equivalents, a quantity that takes into account not only the size of the household but also its composition. Second, there are economies of scale in a household. For example, a two-person household does not need twice as many refrigerators as a one-person household. One of the ways to account for economies of scale is to apply a factor between zero and one to the household size or the number of adult-equivalents. The five alternative indicators incorporate some of these assumptions. The first indicator introduces the concept of economies of scale in the basic indicator, applying a factor of 0.9 to household size. The second welfare indicator uses the Oxford scale to calculate the number of equivalent adults. The third and fourth, respectively, use the RDA scale and the FAO scale. The fifth uses the FAO scale and applies a factor of 0.9 economies of scale. There is a strong positive correlation between the six indicators. The original welfare indicator is called PC0 and the others are named PC1 to PC5. Poverty headcounts calculated with the indicators are very different. On the one hand, PC0 classifies two-fifths of the population as poor and at the other extreme PC2 classifies just one-tenth of the population as poor. However, there is a strong positive correlation between the six indicators, the linear correlation coefficient ranging from 0.97 to 0.99. The fact that there is a strong positive correlation between the six 41 indicators means that households are broadly ranked in the same way regardless of the indicator and that the poverty profiles derived from these indicators will be closed. Table 2.10: Poverty indicators using alternative poverty measures Urban Rural All Poverty % % Poverty % % Poverty % % Poor Headcount Population Poor Headcount Population Poor Headcount Population PC0 13.7 21.8 7.5 47.5 78.2 92.5 40.1 100.0 100.0 PC1 6.1 21.9 6.3 25.3 78.1 93.7 21.1 100.0 100.0 PC2 3.3 21.9 6.0 14.5 78.1 94.0 12.0 100.0 100.0 PC3 5.1 21.9 7.0 18.9 78.1 93.0 15.8 100.0 100.0 PC4 8.7 21.9 7.6 29.7 78.1 92.4 25.1 100.0 100.0 PC5 13.3 21.9 8.6 39.6 78.1 91.4 33.8 100.0 100.0 Source: Author’s calculations using INSD survey EMC -2014 For regions, which are an important dimension of the poverty profile, the six indicators provide a consistent ranking. Poverty comparison between regions is important because many projects use geographical targeting to start, even if this type of targeting is combined with others. The ranking of regions using the six poverty indicators shows that the four wealthiest regions (Centre, Sahel, Cascades, Hauts Bassins) are always the same and the two poorest regions (Boucle du Mouhoun, Nord) as well. Three other regions, the Centre-Sud, the Centre-Ouest and the Est have close rankings with the different indicators. So 9 of the 13 regions have similar rankings and only 4 have a greater variability in their rankings. Overall, a poverty profile developed using any of the five alternative welfare indicators is identical to the one developed with per capita consumption. When we consider sociodemographic characteristics, we note that households with male heads have higher poverty rates than those with female heads; the poverty headcount increases with the age of the household head; it decreases with the educational level of the household head and is higher for households whose head works in agriculture. 2.2.3 Poverty correlates The problem with a poverty profile is that while it gives information on who are the poor, it cannot be used to assess with any precision what are the determinants of poverty. For example, the fact that households in some regions have a lower probability of being poor than households in other regions may have nothing to do with the characteristics of the regions in which the household lives. The differences in poverty rates between regions may be due to differences in the characteristics of the households living in the various regions, rather than to differences in the characteristics of the regions themselves. To sort out the determinants of poverty and the impact of various variables on the probability of being poor while controlling for other variables, regressions are needed. The analysis is done using the 2014 data. The dependent variable is the logarithm of per capita annual expenditure divided by the poverty line. The explanatory variables fall into six broad categories: (1) Sociodemographic (Household composition, gender of household head, handicap, etc.); (2) Human capital (education of the head and the spouse, experience of the head and the spouse, etc.); (3) Labor market (institutional sector, type of industry, etc.); and (4) 42 Productive and social capital (land ownership, membership of an association, etc.); (5) Access to infrastructure (time to the nearest road, time to nearest market, etc.); (6) geographic (Area and region of residence). The results show that sociodemographic characteristics have a significant impact on household welfare. In particular, household size is negatively correlated with per capita consumption. Having a new member in the household, regardless of gender or age contributes to reducing it; for example, an additional teenager reduces per capita consumption by 14 percent in urban areas and 11 percent in rural areas. This result confirms that demography is important for poverty alleviation policies. Population growth in the country is extremely high, more than 3 per cent per year, mainly because of high fertility, with more than 6 children per woman. The issue might be sensitive, but it deserves at least to be discussed. Compared to a household with a male head, a household with a female head has a per capita consumption 36 percent lower in urban areas and 14 percent lower in rural areas. This result is consistent with the fact that on average women have a lower human capital and have less opportunities (land, credit, etc.). The result also clarifies two interesting points. First, contrary to what appears in simply descriptive statistics, households with a female head actually have a lower welfare than their male counterparts when all other factors are controlled. Second, the welfare gap between male and female households is even greater in urban areas than in rural areas. Other sociodemographic characteristics, such as a household having a disabled head and non-national from Burkina Faso, have a negative impact on household welfare. Human capital variables, including work experience and level of education, are also correlated with welfare. Age is a proxy of labor market experience. If experience is valued in the labor market, we would expect poverty to decline with higher age. In fact, the age of the head of the household has a positive impact on per capita consumption on urban households, not for rural ones. The fact that age does not have a positive impact of welfare in rural areas may be due to the fact that the majority of the population is in traditional agriculture where experience is not valued enough. However, the experience of the spouse of the household head is rather negatively correlated with welfare. In other words, the spouse (who is very often a woman) is more productive at young ages, and becomes less productive as she becomes older, maybe because of maternity. The level of education of the head of the household and that of his/her spouse contributes to significantly improving household welfare. In urban areas, the fact that a household head has primary education level improves per capita consumption by 13 per cent compared to a household whose head has never been at school. This figure increases to 84 percent for households whose head has a tertiary education level; and these effects are also important in rural areas. So education attracts a premium in the labor market, and giving as many young people as possible the opportunity to study is a path to reducing poverty for the next generation. With regard to labor market characteristics, the type of industry and the institutional sector are also key determinants of poverty. There is a discount for a household whose head is working in his own individual enterprise compared to other institutional sectors (public administration, public enterprises and formal private enterprises). As for the type of industry, the 43 results of the model confirm that there is a positive effect for households whose head is working in any branch of activity compared to those working in agriculture. The study also identifies production assets and social capital as correlates of household welfare. Ownership of a hectare of land helps improve the level of per capita consumption by 0.3 per cent. Similarly, the fact that a household has at least one member affiliated with any association helps improve its level of per capita consumption by 8 percent as compared to a household that has no member affiliated with any association. Indeed, associations play an important role in improving access to credit to finance income-generating activities. They also play an insurance role in case of negative shocks (illness, death, etc.). As for infrastructure, the results of the model indicate that the time necessary to reach the nearest basic infrastructure has an impact on household welfare. The longer the time to the nearest grocery market, pharmacy or police station, the lower the household welfare. When infrastructure is close, transaction costs are lower and this has a positive impact on household welfare. The study also shows regional differences, which can mirror the unobserved potential of the regions. The Nord region, which has the highest poverty rate, is considered the reference region for the econometric model. The results show no difference between this region and the Boucle du Mouhoun region, confirming that these two regions are the poorest and that the situation of the poor in those regions comes in part from the low potential they offer. When all other characteristics are controlled, the region with the highest potential is the Sahel region, a region offering opportunities for livestock breeding. Living in this region increases welfare by 44 percent compared to the poorest Nord region. The other regions with good potential are the Centre-Nord, Centre, Cascades and Centre-Est. 44 3. FOOD INSECURITY IN BURKINA FASO Food insecurity is one aspect of poverty. According to the FAO,13 food security is assured when all people at all times have economic, social and physical access to sufficient, safe, nutritious food that meets their dietary needs as well as their preferences and allows them to maintain a healthy and active life. If even one of these conditions is not met, people suffer food insecurity. This, therefore, involves many factors. The food must physically exist. People must be able to physically reach it and afford to buy it. The food must be nutritious to maintain a healthy and active life. It must offer a balanced diet. And the food must be continually available. The first of these issues—supply and shortages—can usually be gleaned from annual agricultural surveys. In this section, we address the other three aspects. 3.1 Characteristics of food insecurity 3.1.1 Food insecurity according to the food access approach The FAO uses the FIES (Food Insecurity Experience Scale) approach to measure food insecurity relative to a limited access to food14. The approach consists of calculating an indicator using a series of eight questions asked to an adult member of the household. The questions explore various situations: (i) has the household had to worry about not being able to meet its food needs; (ii) has the household had to reduce the quality or variety of its food; (iii) has the household had to reduce the amounts consumed by skipping meals; (iv) has the household had to deal with famine. The indicator calculated using the Rasch approach starts with the observation that situations (i) to (iv) reveal the seriousness of the food insecurity. A household dealing with the first situation is experiencing moderate food insecurity; and the closer one comes to point (iv) the worse the situation becomes. In 2014, this form of food insecurity affected nearly 38% of individuals. Individuals experiencing food insecurity according to this approach are either in a moderate situation, insofar as they were led to reduce the amounts normally consumed by skipping meals, or in a severe situation, i.e., facing famine. More than 15%, i.e., one person out of seven, is affected by a severe form and faces a virtual lack of food at certain times. 13 FAO (Food and Agriculture Organization of the United Nations). Rome declaration on world food security and world food summit plan of action. Rome: FAO; 1996. 14 Terri J. Ballard, Anne W. Kepple, Carlo Cafiero. The Food Insecurity Experience Scale. Development of a Global Standard for Monitoring Hunger Worldwide, FAO, 2013. 45 Figure 3.1: Geographic map of food insecurity (FIES approach) Source. Author’s, using FAO and INSD calculation from EMC 2014 survey Food insecurity is characterized by geographic disparity, particularly the area of residence and the region. While agriculture is the main activity in rural areas, this first form of food insecurity affects the countryside one-and-a-half times more than urban areas. In rural areas, two out of every five people are experiencing food insecurity (severe or moderate), and only one person in five is really experiencing genuine food security. The others (also two out of five people) are in an intermediate situation, neither experiencing genuine food security nor food insecurity. Rural areas are therefore vulnerable in terms of access to food. The reality is that agricultural productivity is rather weak, and agricultural operators do not produce enough to meet their needs. And because their incomes are low, they have difficult accessing products, perhaps because of the combination of distance to market and low purchasing power. In urban areas, most consumption comes from the market, and when there are resources, there are generally opportunities to find provisions. At the regional level, this form of food insecurity worsens as one moves from west to east in the country; indeed, it is very strong in the East and in the Sahel (nearly 60%); and it is also high in the Sud-Ouest, Centre-Sud, Centre-Nord and Centre-Est. Interestingly, this regional food insecurity map does not overlap the income-poverty map. The Nord and the Boucle du Mouhoun, two regions where income poverty is the highest, are instead spared from this form of food insecurity. Nonetheless, some regions have high levels of income poverty and of food insecurity, as is the case in particular of the Est, the Centre-Sud and the Centre-Nord. 46 Figure 3.2: Food insecurity incidence (FIES approach) by household characteristics Source. FAO and INSD calculation using EMC 2014 survey The incidence of this form of food insecurity varies somewhat with the household’s socioeconomic characteristics, particularly the gender of the household head and his or her educational level, but it is independent of their standard of living. The incidence of food insecurity forms an inverted U according to the size of the household, relatively low for single- person households, a maximum for average households with around five people; it drops for the largest households, but it does not drop to the level of one-person households. Conversely, the incidence of this phenomenon is U-shaped in relation to the size of the household, with the minimum being reached for households whose head is around 40 years old. We also find that households headed by a woman are more food-insecure than those headed by a man. The variable with the greatest impact on food insecurity is the household head’s educational level. The incidence rises to over 40% for households whose head has not received schooling, as against just 2% for those whose head has received higher education. It is also worthwhile noting that this form of food insecurity does not seem to be correlated with income poverty, which would be the case if the incidence of food insecurity decreased with the household’s standard of living. In fact, only households in the fifth quintile really stand out from the others, with the lowest incidence. This last result shows up in the level of food consumption; when this is analyzed as a function of the level of food security, this consumption does not differ among the various categories of households. Since food insecurity is measured by the fact of having more or less access to food products, it is natural to compare food consumption among households according to their level of expenditure. The various categories of households fall within 72,000 FCFA to 76,000 FCFA per person, per capita and per year. Households experiencing food insecurity do not consume less than the others, at least in terms of annual value. 47 Table 3.1: Average annual cereals production for the last 20 years (*) 1996-2000 2001-2005 2006-2010 2011-2014 Rice 97,523 92,497 172,392 303,285 Maize 386,075 641,081 892,436 1,412,893 Millet 811,762 1,064,374 1,103,013 989,556 Sorghum 1,118,862 1,461,474 1,681,935 1,754,357 Fonio 13,423 9,360 17,261 15,902 Total Production 2,427,645 3,268,786 3,867,037 4,475,993 Potential production available 2,024,489 2,741,469 3,218,025 3,683,280 Population (annual average) 10,979,729 12,667,258 14,721,026 16,842,854 Production available per capita 184 216 219 219 Source. FAO and author’s calculation (*) Cereal production is in tons, except production per capita which is in kilograms Food security is ensured, first, when foodstuffs are available and, second, when they are accessible to the populations and, in the case of supply, it is structurally weak in Burkina Faso. Cereals form the main food consumption item in the country. Cereal production, which was 2.5 million tons 20 years ago, has been 4.5 million tons in recent years. This production increased faster than the population, which led to an improvement in the overall supply of food products Available production, which was 184 kilograms of cereals per person between 1996 and 2000, was 219 kilograms per person between 2011 and 2014.15 One person’s needs are assessed at 203 kilograms per year, so in theory the country has a slight surplus. However, this weak surplus does not leave much room for maneuvering. First, there is a problem with the spatial distribution of this supply. The western provinces are regularly in a surplus situation, while those in the Northeast experience shortfalls. For example, for the most recent harvest (2015-16), the authorities estimate that of the country’s 45 provinces, one-third of them are experiencing a food deficit. In 2011/12, 16 of the 45 provinces were in this situation. Plus, when surpluses are weak, there are major opportunities for speculation. Indeed, when neighboring countries are experiencing a drop in supply, for example, farmers are tempted to sell outside the country, and this intensifies the shortfall at the national level. The fact that the country is a net importer of cereals confirms the fact that it is not entirely self-sufficient in this area. While the country manages to meet its needs for dry cereals (millet, sorghum, etc.), its needs for rice and wheat are met mostly by importers and, to a lesser extent, food aid, which once again highlights the insufficiency of the domestic production. Wheat imports are nearly 50,000 tons annually but reached 100,000 tons in 2014. As for rice, since 2012 imports and aid, net of exports (which are virtually non-existent), have exceeded 400,000 tons. This means that more than 10% of consumption comes from outside the country, most of that being rice consumption, which is eaten more in urban areas than in the countryside. 15 The available production is calculated by subtracting requirements for seeds and other losses from gross production (15% for all cereals except rice, where 45% is subtracted); Final results of the 2011/12 farming season; Ministry of Agriculture. 48 Figure 3.3: Balance of international cereals transactions (*) Source. Author’s calculation using 2014 Statistical Year Book (*) The balance is equal to the sum of food imports and aid minus exports Moreover, road conditions do not facilitate the circulation of goods to resolve problems of physical access to provisions. Roads are far from being the main mode of transport in this country: there is no road transport and no maritime transport; planes serve only a few large population centers and are not well suited to transporting goods. But road density is low, with 5.6 kilometers of roads per 100 square kilometers, as opposed to an average of 6.84 km for Africa, 12 kilometers for Latin America, and 18 kilometers for Asia. In addition, these roads are not always well maintained. The limited road network and the poor condition of the roads lead to high transaction costs that affect prices on food products for the final consumer. In addition to physical availability, economic availability is also impeded by the major price differences from one region to another. For the four centers used to check cereal prices, the difference between the price extremes is regularly on the order of 50%, while no surveys have been conducted in the countryside where prices are still lower. The city of Banfora, in western Comoe province, posts low prices. At the other extreme, Ouagadougou, the country’s capital, and Dori, the Sahel region capital where agricultural production is less significant, posts high prices. These price differences negatively impact the purchasing power of poor households in cities where the prices are high, particularly those in the capital, Ouagadougou, and heighten food insecurity in those regions 49 Figure 3.4: Annual average cereal prices (FCFA per kilogram) in some main cities in Burkina Faso Source. Author’s calculation using Statistical Year Book 3.1.2 Food insecurity based on calorie intake The second approach assesses food insecurity based on the household’s calorie intake. This is, hence, a nutritional approach that determines how well needs are being met based on the number of calories drawn from the consumption of food products. A household is experiencing food insecurity if consumption is below 2,283 kcal per adult equivalent and per day. Anyone living in a food-insecure household is also in this situation. Under this definition, 43% of the people were food-insecure in Burkina Faso in 2014, with one-fourth of the urban and nearly half of the rural. The level of food insecurity is very close to that of income poverty, and the two approaches (food insecurity according to calorie intake and income poverty) that directly measure household consumption; it is legitimate to wonder to what extent there is a correlation between these two aspects of poverty. Food insecurity decreases with the household’s standard of living (measured by the household’s consumption per capita). It affects almost all of the poorest households in the first quintile, nearly three-quarters of those in the second quintile, and is virtually nonexistent among well-off households in the fifth quintile. Moreover, among the subpopulation that has not reached the minimum calorie level, seven out of ten people are poor. Furthermore, because rural areas are more affected by this than 50 urban areas, and most of the population is rural, nearly nine out of ten people suffering a calorie deficit live in a rural area. Figure 3.5: Food insecurity (calorie intake approach) incidence in Burkina by region Source. Source. Author’s calculation using the 2014 EMC survey This second form of food insecurity has the same characteristics as income poverty and is different from food insecurity using the FIES approach. The regions where food insecurity is less prevalent are the Sahel, the Center and the Cascades regions (under 20%), which have less than 11% of the food-insecure population; these are the same regions with the lowest poverty rates. At the other extreme, nearly seven out of ten people suffered food insecurity in the North and the Boucle du Mouhoun, these two regions also being those most affected by income poverty. These two regions have 18% of the population and 28% of the people suffering a caloric deficit. Other regions with an equally high incidence (more than 50%) are the Center-West, the Central Plateau and the Center-South. In terms of other characteristics, the incidence of food insecurity is higher in households headed by a man than those headed by a woman; this incidence is an increasing function of the age of the household head and a decreasing function of his/her level of education. Moreover, among the 38% of people suffering moderate or severe food insecurity according to the FIES approach, 16% are also insecure according to the calorie-intake approach. And of the 38% of people suffering moderate or severe food insecurity according to the FIES approach, 35% are also insecure according to the second approach. So half the people form a hard core of people living in households that are experiencing food insecurity regardless of the form, i.e., always experiencing food insecurity, and the other half are experiencing a changing situation, and hence are potentially vulnerable. Table 3.2: Comparison of the different forms of food insecurity Calorie consumption 51 Food Food insecurity FIES safety All Severe food insecurity 6.3 8.8 15.1 Moderate food insecurity 10.0 12.4 22.4 Food safety 27.4 35.1 62.5 Total 43.7 56.3 100.0 Source. Source. Author’s calculation using the 2014 EMC survey Figure 3.6: Food insecurity incidence (calorie) by some socioeconomic and demographic characteristics Source. Author’s calculation using the 2014 EMC survey The consumption levels for the various household categories show that households experiencing food insecurity consume one-half less (by value) than those with food security. The average consumption of a food-insecure household is on the order of 46,000 FCFA per person per year, as opposed to nearly 97,000 FCFA for those experiencing food insecurity. Hence those in the second category simply do not have enough resources to meet their food needs, which basically explains their situation. An examination of the structure of household consumption makes it even easier to determine why some households are experiencing food insecurity. Burkinabé households spend half their food budget on consuming cereals, mainly millet, sorghum and maize. Much of this consumption of cereals is satisfied by self-production, especially with regard to maize, millet and sorghum. So behind this poor country’s consumption habits there is first and foremost a philosophy of feeding oneself. Meat and fish, more of a luxury food, account for 15% of food consumption. It is worth noting that vegetables and dairy products, foods considered good for the health, are also important, accounting for 6% and 4%, respectively, of the total. On the other hand, households consume little fruit, less than 1%. Food-secure households tend to consume relatively luxury foods, whereas those experiencing food insecurity consume necessary foods first. Hence cereal consumption is higher among households experiencing food insecurity (with a 52% budget share) than among food-secure households (45%). Among cereals, households experiencing food insecurity consume relatively more less-expensive cereals (millet, sorghum, maize) and other households consume relatively more rice. Likewise, the former consume two-and-a-half times more meat 52 than the latter, whereas fish consumption is similar in both groups. Those households that are better off in terms of food security consume more milk and dairy products, fruit and vegetable oils. Food-insecure households consume more sugar, nuts and condiments. These results on consumption habits tend to show that if matters of nutritional balance were also analyzed, a great many food-insecure households might also find themselves back in a situation of nutritional imbalance. Figure 3.7: Annual per capita consumption of food items by food security status (calorie) 25,000 20,000 15,000 10,000 5,000 - Food safety Moderate food insecurity Severe food insecurity Source. Author’s calculation using the 2014 EMC survey 3.1.3 Transient and chronic food insecurity Some households are vulnerable in the sense that they may be affected by food insecurity at certain times of the year. The preceding analyses reveal the overall food-insecurity situation for the year This snapshot is incomplete, however, since a household’s situation can change from one season to the next. For example, farmers have an excess of provisions right after harvest, and the situation may become difficult as time passes. Some households may be in a precarious situation at any given time of the year, and others are chronically in a precarious situation. It is worthwhile studying household mobility in regard to food insecurity: different policies need to be adopted depending on whether a household is chronically or transitorily food-insecure. Like the previous one, this subsection uses the calorie-intake approach. Since we know about food consumption for each of the four rounds of the EMC, we can study the transient and chronic nature of this phenomenon. Food insecurity is defined similarly for each of the four stages of the survey. A household is experiencing this situation if calorie intake per adult-equivalent is less than 2,283 kilocalories per day. Mobility between two periods is monitored using a transition matrix. Food insecurity is characterized by strong seasonal variations that most often translate into a worsening of the households’ situation. One-third of people live in a situation of food insecurity in the first quarter; this figure rises to 45% in the second quarter, nearly 42% in the 53 third, and nearly 47% in the last quarter.16 In fact a significant proportion of households undergo a change in status. Between the first two quarters, more than one-fourth live in households that have undergone a change in status; 18% of those experiencing food security in the first quarter find their situation changing for the worse, and just 7% find their situation improving. These changes in situation occur in all periods, thus revealing the level of vulnerability of Burkinabé households. Food insecurity is more of a transient rather than a chronic phenomenon. The 2014 results show that only one-third of these people are not experiencing food insecurity at any given time of the year. For the two-thirds who experience this difficulty, 18% are in this situation chronically, and nearly half transitorily (i.e., once, twice or three times during the year). The chronic nature of this phenomenon is the result of extreme poverty. Of those living with chronic food insecurity, 80% are in the first quintile, most of whom come from the poorest households. On the other hand, the transitory nature of the phenomenon has the result of a combination of multiple factors. As noted above, agricultural production does not always meet the needs of these populations. In 2012, the cereal shortfall was estimated at more than 4% of production. And, as stated previously, a surplus of provisions at the national level does not mean that there will be food balance throughout the country given the differences in production levels among regions and the difficulties associated with transportation. Furthermore, price variations during the year explain the variations in real income, which may decline at certain times of the year and cause temporary food insecurity. During 2012, the food-products price index varied by nearly 6% between the month when prices were the lowest and the month when they were highest; this figure was over 11% in 2013. The prices of some products, particularly fresh produce, show greater variations (16% in 2012 and 9% in 2013). The profile of chronically food-insecure households and that of households experiencing transient food insecurity are close to that for income poverty, with the profiles differing only by the fact that certain socioeconomic characteristics of these households are more or less acute. The first factor characterizing the food-insecurity profile is household composition. Chronically food-insecure households are larger (nearly 10.5 people) and have a larger number of young individuals (dependency ratio of 1.2). Households in a transient situation are also large, albeit less so (8.3 people), compared to fewer than 6 people in households that never experience food insecurity. Households experiencing chronic food insecurity are relatively more numerous in rural areas than those experiencing transient food insecurity. The household head of the first group is also older on average, and likely to be uneducated. Table 3.3: Characteristics of food insecurity % In food % Chronically % Never in insecurity at in food All populations food insecurity least once insecurity 16 It is more likely that the figure for the second round is overestimated. In our view, it should have been between the figure for the first round and that for the third. Harvests in fact begin in November (period preceding the first round) and continue on into January. Farmers have more provisions during this period, and normally the situation deteriorates the more time passes after the harvests. It appears that the second round did not receive as much supervision as the others, as the central team had been selected to produce the first results. 54 % in each situation 33.5 48.3 18.2 100 Household size 5.6 8.3 10.5 7.4 Dependency ratio 0.88 1.16 1.22 1.07 Masculinity ratio 0.93 0.95 0.97 0.94 % Head is female 15.3 12.2 10.6 13.4 Age of household head 43.1 48.5 51.6 46.5 % Head has no education 64.9 85.2 88.1 76.6 % Head has no education 13.6 10.0 9.2 11.5 % head not in agriculture 48.4 21.1 19.1 32.9 % in first welfare quintile 0.2 22.5 79.7 20.0 % in second welfare quintile 5.6 36.1 16.0 20.0 % in third welfare quintile 18.3 26.8 3.1 20.0 % in fourth welfare quintile 33.7 11.5 1.0 20.0 % in fifth welfare quintile 42.2 3.1 0.1 20.0 % in rural areas 58.3 83.6 85.9 72.7 Hts Bassins 12.5 10.0 12.9 11.5 Bcle Mouhoun 3.2 12.2 15.6 8.7 Sahel 8.8 4.6 0.4 5.9 East 4.6 11.6 7.3 8.0 Southwest 4.8 4.5 8.5 5.2 Center-North 7.9 7.1 4.2 7.1 Center-West 5.3 10.7 9.1 8.1 Central Plateau 3.4 4.7 5.4 4.2 North 2.7 7.2 13.4 6.0 Center-East 8.4 8.8 7.7 8.5 Center 28.4 9.5 7.9 17.7 Cascade 5.9 4.2 1.8 4.6 Center-South 3.8 4.9 5.7 4.5 % faced Natural shock 33.6 49.8 54.5 43.3 % faced Price Shock 18.7 28.2 33.5 24.6 % faced an employment shock 5.1 2.8 3.4 3.9 % faced death/Illness 14.9 18.6 16.2 16.7 % faced security shock 4.6 5.6 4.2 5.0 % faced household shock 2.1 2.3 2.7 2.3 % faced other shock 2.2 2.5 2.8 2.4 % faced any shock 56.0 70.8 76.1 64.9 Source. Author’s calculation using the 2014 EMC survey 3.2 Food insecurity and vulnerability to shocks 3.2.1 Main shocks suffered by households 55 Burkinabé households are often hit by idiosyncratic and covariant shocks. Idiosyncratic shocks are those affecting a household (loss of job, divorce, crime, separation, etc.) in particular, and covariant shocks affect a group of households (price variations, drought, flooding, etc.), for example a village, region or even the entire country. This is an important distinction as it better indicates the measures to be taken to mitigate or buffer the effects of shocks. In the event of an idiosyncratic shock, the household uses its resources and means to deal with it; in the case of a covariant shock, in addition to individual means, a larger-scale intervention may prove necessary. Food insecurity can be aggravated or even provoked by a shock. A drought affects harvests and the availability of provisions in households. The preceding analysis is based on the retrospective questions of the 2014 EMC. The impact of shocks is very significant; they affect poor populations the most. More than two-thirds of households reported they had suffered at least one shock, most frequently of natural origin (43% of households), caused by price fluctuations (25%) or by the death or serious illness of a member of the household (17%). Other shocks are less frequent and affect less than 5% of households. Shocks affect rural populations more than urban populations. Rural households suffer more from problems associated with weather and plant diseases, meaning poor harvests. There are also events associated with price fluctuations that can be correlated with natural shocks. Since these rural households live mainly from agriculture, they are more exposed to shocks of this kind. Moreover, because the health system is poorly developed in rural areas, we also find that the incidence of shocks relating to a serious illness or death of a household member is greater there. On the other hand, events associated with the loss of a non-agricultural job or income naturally affects city households more. We also see major regional variations. The prevalence of natural and job-related shocks is also closely associated with the area’s climate and urban development. Northern areas (Centre-Nord, Sahel), characterized by a Sahelian climate, experience higher incidences of natural and price-related shocks. On the other hand, the Centre, Hauts-Bassins and Cascades, located to the south, are more subject to shocks related to the job market. We also find that the regions on the south-southwest axis (Cascades, Sud-Ouest, Centre-Ouest) have a greater frequency of shocks associated with health issues. Box 3.1: Categorization of shocks affecting households Household Issues: Divorce, separation, End of regular transfers from other households Prices: significant drop in prices of agricultural products, high prices of agricultural inputs, high food prices Natural hazards: Droughts, floods, high rate of crop diseases, high rate of animal diseases Employment: Significant loss of nonfarm income, bankruptcy of a non-farm business, significant loss of wage income (other than due to an accident or illness), Loss of employment of a household member Health: Serious illness or accident of a household member, Death of an active member of the household, Death of another household member Crime & Safety: Theft of money, goods or harvest, Conflict / Violence / insecurity Other: Other issues non classified above 56 Figure 3.8: Incidence of shocks by place of residence A - Urban areas B - Rural areas 60 40 20 0 Natural hazards Prices Health Employment Household issues Crime & Safety Other Source. Author’s calculation using the 2014 EMC survey 3.2.2 The impact of shocks on household food security Shocks can result in a loss of real income and negatively affect household food security, particularly in the most vulnerable households. We examine the correlation between food insecurity and shocks by estimating an econometric model for each of the three approaches to food insecurity. For the first approach, this involves a probit model whose dependent variable is binary and takes the forms one or zero, depending on whether the household is food-insecure or not, respectively. For the second approach, the dependent variable is the number of calories per adult-equivalent consumed in the household,17 as part of a classic linear model. For the third approach, the explanatory variable is status in relation to food security; it takes three forms (chronic food insecurity, transitory food insecurity, never food insecurity) as part of a multinomial logit model. The explanatory variables are the same; the sociodemographic characteristics of the household and its head (composition of the household, human capital), standard-of-living variables (access to electricity, modern toilets, ownership of durable goods), and geographic variables that take into account unobservable effects and shock variables. Two models are estimated; the first considers shocks by type, and the second includes all shocks.18 Each of the models is estimated at the national level and for urban and rural areas. FIES Approach The probit model’s results show too weak a correlation between sociodemographic characteristics and this form of food insecurity. The only variables correlated with food insecurity in the FIES approach are the standard-of-living variables for the household and the region. When the household has electricity, the likelihood that the household is food-insecure diminishes. The same is true for a means of transportation (automobile, motorcycle) or a refrigerator. In the case of the region, compared to the region of the Hauts-bassins (chosen as 17 In the model, we standardize household calorie intake by dividing it by 2283, the food-insecurity threshold. 18 For the multinomial logit model, we estimate only the model with all shocks together. 57 reference) located in the southeast part of the country, the likelihood of being less insecure diminishes in the regions of the Boucle du Mouhoun and the Central Plateau, and it is neutral or increases in the other regions. This form of food insecurity is correlated with physical and economic access to food products. The main variable measuring physical access is the distance to transportation, which shows a likelihood of reducing food insecurity when transportation is closer. However, as we saw above, ownership of a means of transportation also reduces the likelihood of being food- insecure; indeed, all things being equal, having a means of transportation is not just a sign of material ease but also reduces the distance to markets and provides an opportunity to expand the geographic field of supply by going to more attractive markets. The same holds true for households with the fact of having access to electricity and owning a refrigerator. The former makes the latter possible, and the latter promotes food storage and hence the possibility of buying larger quantities at better prices. Calorie-intake approach Unlike the previous case, the results of econometric regressions show that sociodemographic characteristics are closely correlated with the household’s food-security situation. First, the place of residence has no significant effect on food insecurity. Even though the level of food insecurity is higher in rural areas, the phenomenon is not due to the fact of living there but rather to other factors. The variables of household composition have a significant impact on the level of food insecurity. When household size increases, calorie intake is reduced. One additional individual lowers the calorie intake level per capita by 23% in urban areas and by nearly 7% in the countryside. Moreover, a one-point increase in the household’s dependency ratio further accentuates this drop in calorie intake. The effects of the human capital are mitigated. Calorie intake is a decreasing function of age, but mainly in rural areas, age having no effect in urban areas. This means that the older a household head is, the more likely that the household will fall into food insecurity; in this way, households with an older head seem more vulnerable. On the other hand, educational level has a mainly positive impact on calorie intake; education also probably makes it possible to earn more income and to be better informed about proper nutrition. Variables of housing characteristics or ownership of household assets, which are proxies for a household’s level of welfare and permanent income, are negatively correlated with food insecurity. Hence the fact that a household is hooked up to electricity, has flush toilets, a car, motorcycle or refrigerator causes the household’s calorie intake to rise. Dynamic aspect of food insecurity These same demographic variables are correlated with chronic and transitory food insecurity; the difference between states of insecurity is to be found in its marginal effects, which are more accentuated for the chronic form of the phenomenon. First, the household’s size worsens all states of food insecurity but especially so the chronic form. As we saw, households experiencing chronic food insecurity have a very large household size, nearly 11 people on average. In addition, the dependency ratio aggravates the food insecurity. This ratio is also positively correlated with both states of the phenomenon. In others words, household size, and a large number of dependents, are two characteristics leading to food insecurity. Burkina 58 Faso’s demographics, characterized by a high composite fertility index and strong demographic growth, is a major factor that intensifies not just income poverty but also food insecurity. On the other hand, the household head’s gender is not a determining factor in food insecurity. The characteristics of human capital, professional experience and educational level are only weakly correlated with food insecurity. Professional experience, measured by age, is positively correlated with transitory food insecurity, but not with chronic food insecurity. The older a household head, the more likely that the household is insecure at any given time of the year; but, on average over a year, a household with an older head is not more exposed than the household with a younger head. This can be expressed by saying that households whose heads are older are more vulnerable in the sense that they do not always manage to have sufficiently calorie-rich food throughout the year. Educational level is also weakly but negatively correlated with transitory food insecurity: the more a household’s head has a good level of education (from secondary on), the less likely the household is food-insecure at any given time of the year. In contrast, the characteristics of household living standards are negatively correlated with food insecurity. The fact that a household has electricity, a motorcycle and a car reduces the likelihood that it is experiencing chronic or transitory food insecurity. These variables measure rather the household’s permanent income, and a household that has these assets is probably a well-off household. Shocks and three approaches to food insecurity Shocks have a generally negative impact on all forms of food security. When we look at the first form of food insecurity (the FIES approach), shocks aggravate the households’ situation, which consequently renders them vulnerable. Natural shocks (drought, flooding) negatively impact the food security of households in rural areas. These shocks have direct consequences for agricultural production and, therefore, for the availability of products. Shocks from the job market (layoffs, bankruptcy), death and safety issues (violence, rape) more negatively affect food security in urban areas. It is worthwhile noting that price shocks (strong variation in the prices of food products) are positively correlated with this form of food security in rural areas. The results of the model with the second approach to food insecurity point in the same direction. Price shocks have the most negative impact on household food security. The price effect lowers the calorie intake per adult-equivalent by more than 19% in cities and 18% in the countryside. In urban areas, where consumption comes from the market, an increase in food prices contributes to a reduction in real income that forces households to reduce the amount of food consumed. In rural areas, on the other hand, some households are net producers, and for them a price increase can be beneficial, but others are net consumers, and for them the situation is like that of urban residents. In any case, the weakness of the country’s agricultural production makes rural households dependent on the market, because they produce little in the way of surplus. The other type of shock that has a negative impact on a household’s food security relates to issues affecting the household such as divorce, a separation or the end of transfers sent home by a family member. Shocks of this kind have an impact mainly in urban areas where they cause consumption to drop by 18%. The third model yields similar results, with a more pronounced marginal effect for households experiencing chronic food insecurity. 59 These results also show that shocks heighten the vulnerability of households. To examine the vulnerability of households in the face of shocks, we assess the numerical impact on households that are just above the food-insecurity line (using the second approach), i.e., households in the fifth and sixth quintiles of calorie intake per equivalent-adult, since the incidence of food insecurity is 43%. When we include all shocks in a single variable, a household that experienced those shocks would see its calorie intake reduced by 35% in urban areas and 15% in rural areas. Thus a shock affecting a household in the fifth quintile with average calorie intake of 2,322 kilocalories per adult-equivalent, would see it drop to around 1,825 kilocalories per adult-equivalent, well below the food-insecurity threshold. For a household in the sixth decile, it would go from 2,687 to 2,042 and would also find itself in a situation of food insecurity. Hence households up to the sixth decile are vulnerable because of shock during the year plunges them into food insecurity. It is therefore important to guard against shocks and find solutions so they can be dealt with. 3.2.3 Food insecurity and anti-shock strategies The preceding analyses show that the profile of households affected by food security differs according to the approach considered; consequently the responses, and public policies, also differ. The first form of food insecurity (the FIES approach) measures economic access to food products. Food insecurity arises in part from a limited supply of products, difficulties in transporting products from the production centers to the places where they are consumed, and also high prices in consumption centers. Fluctuations in the supply of products in this country facing covariant shocks in terms of climate reinforce food insecurity. Moreover, as we have seen, problems with moving goods and merchandise aggravate the situation; all these things make households more vulnerable. The other two forms of food insecurity are closer to poverty. They are characterized by large households with low human capital and hence low income, living in rural areas primarily from agriculture. This form of food insecurity makes households vulnerable because of the idiosyncratic shocks that affect them, for example a divorce, or a halt in transfers from a household member living elsewhere. In all cases, shocks have a big impact on household vulnerability. The strategies adopted to deal with shocks reflect the lack of social-protection mechanisms and the country’s less-than-perfect insurance market. The insurance market is not well developed. Health insurance is available for a proportion of wage-earners in the modern sector in a country where a majority of individuals work as independent laborers in agriculture. Other types of insurance are non-existent, and, given the high risk particularly in agricultural activities and also high moral hazard, there are few insurers willing to venture into this sector. In addition, there is no organized social safety-net system (such as unemployment allocations or some specific assistance for poor households). Given the virtual nonexistence of modern social safety- net mechanisms, households try to mitigate the effects of shocks by adopting strategies based on their personal relations. Nearly half of them use their own savings to cope with them. And one- fourth of households hit by a shock turn to selling off part of their assets or take out a loan or get assistance from a relative or friend. Strategies originating in the modern social safety-net system 60 are rare. Only 1.5% of households say they receive government assistance, and less than 5% have obtained a loan in the formal system. However, 20% of households also admit that they do not adopt any strategy. Figure 3.9: Coping strategies in Burkina Faso, 2014 A - Urban areas B - Rural areas 50 40 30 20 10 0 Asset-Savings Asset-Sale of assets Behavior-Employment or migration Behavior-Reduction in food consumption Behavior-Reduction in non-food Credit/assistance-Credit Credit/assistance-Formal assistance Informal credit and assistance Other Did nothing Adaptation strategies to cope with shocks reflect the country’s poverty and increase household vulnerability. It was noted above that the main strategy of households is to use their own savings. In a poor country where the level of savings is necessarily low, recourse to savings as the main strategy illustrates the paucity of opportunities available to households in this area. In urban areas, the other two major strategies are recourse to loans from relatives and friends and a reduction in food consumption, which one-third and one-fourth of households, respectively, turn to. This assumes that these households depend either on the generosity of others or simply endure eating less, thereby directly jeopardizing their food security. In all cases, it is obvious that these households are fragile when coping with shocks. Rural households do not have better solutions. One-third of them adopt the strategy of relying on the meager assets they own, and one-fourth accept the generosity of relatives and friends. The broad guidelines of public policies to improve food security should include improving agricultural productivity, improving access to provisions, and expanding social safety-net mechanisms. The first two measures are very general and go beyond mere questions of food security. Agricultural productivity is low in the country, and farmers can get better results if it is improved. In addition, increasing agricultural productivity increases supply and reduces imports. Then this supply must be available to populations all over the country, which requires working on the issue of transportation, particularly the poor state of the road network, a structural weakness in Africa. This means the country is regularly faced with climate shocks, not to mention shocks at the individual level. Shocks make households vulnerable, and even in rich countries the insurance market is used to soften the impact of shocks. A minimum of social safety-net mechanisms would ease the negative impact of shocks. 61 4. RURAL INCOME AND POVERTY 4.1 Profile of rural households The previous analysis shows that the potential for poverty reduction lies in rural areas. Half of the rural population lives below the national poverty line and 9 poor out of 10 reside in the countryside. Half of the rural population is also food insecure and most of the people are vulnerable in the sense that two-thirds face a situation of food insecurity at some time of the year. The 2011 development strategy also known as SCADD19 intends to achieve the objective of a substantial poverty reduction through two main channels, a solid GDP growth rate (an average of 10 percent per year) and by promoting pro-poor growth policies. Agriculture was obviously one of the key sectors selected to promote solid growth. The strategy recognizes that significant physical, technical and socioeconomic constraints have been limiting the performance of the agricultural sector and it was important to remove these constraints in order to improve productivity and income. The most important constraints identified are land tenure security, agricultural mechanization, access to improved seeds and fertilizers, vulnerability to climate change and promoting the marketing of agricultural products. This section provides an analysis of source of income for a better understanding of rural poverty. Rural households are constrained in many aspects, making it difficult to use their potential. The average household size is high, 8 persons versus 6 in urban areas. The large number of individuals in households can be a potential for agricultural activities where labor is needed. But this happens only if most of these individuals are of productive age, which is not the case. The dependency ratio is 1.36, meaning that there are 36 percent more young (under the age of 15) and elderly (over the age of 64) than people of working age (age 15 to 64). Most of the households are headed by a married man with no education. The educational level of the population is important for the effective exercise of any activity. Better-educated people are more likely to benefit from the training offered (for example on extension services) and would be more open to new techniques and the use of modern equipment. But the human capital of heads of household is low. Nine out of ten heads of households in rural areas have no education. This low level of education cannot even be complemented by other members since two-thirds of the most educated adults in the household are in the same situation. This low level of education is also a major constraint on the modernization of this sector. The living conditions of rural households are precarious. On the positive side, 9 out of 10 households own their dwelling, and three-quarters of rural households have access to safe water, mainly by drilling wells, the primary source of drinking water. But the other characteristics of housing reflects a high level of poverty. Half of the houses use sheets as roofs, but almost all of them have their walls made of mud or similar materials and less than 40 percent have a cement 19 SCADD stands for Stratégie de Croissance Accélérée et de Développement Durable ( Strategy of Accelerated Growth and Sustainable Development). 62 floor. And in rural areas, the population barely use any kind of hygienic toilets, with nearly 8 out of 10 households having no toilet (modern or even latrines). In addition, Burkina Faso households have limited access to infrastructure and basic services. This limited access not only reflects the poor living conditions of households, but has a direct impact on their ability to carry on economic activities. Electricity is an important input for carrying on many activities. Electricity can make possible certain basic manufacturing activities, enabling households to diversify their sources of income by moving towards the most productive activities, for example processed agricultural products. But electricity is scarce and less than 3 percent of rural households use it as the main source of lighting, severely limited by access. Firewood is the main source for cooking and less than 1 percent of households use gas or electricity for this purpose. Table 4.1: Characteristics of rural households Urban Rural All T-Test Household Characteristics Household size 5.9 7.9 7.4 *** Dependency ratio 0.7 1.4 1.2 *** % With no education of the best educated member 17.6 63.4 50.9 *** Head of household characteristics % Women 14.8 13.1 13.6 Average age 44.5 47.0 46.3 *** % Married 70.1 87.3 82.6 *** % With no education 44.7 88.0 76.2 *** % Household involved in agriculture 22.2 89.2 71.7 *** Housing characteristics % Owning dwelling 61.6 92.9 84.4 *** % Walls made of cement/brick 50.5 6.4 18.4 *** % Roof made of sheets/cement 94.9 53.6 64.9 *** % Floor made of cement/tiles 91.5 38.7 53.1 *** % Main source of lighting is electricity 59.6 2.6 18.2 *** % Living in households using clean energy for cooking 21.8 0.6 5.2 *** % Main source of drinking water is potentially safe 94.4 75.2 79.4 *** % Individuals living in households with piped water 49.8 0.7 11.4 *** % Having a toilet system 80.6 22.5 38.4 *** Source. Author’s calculations using the EMC 2014 4.2 Stylized facts on labor market and income source in rural areas More than others, the poor rely on labor for their livelihood and at first sight, the Burkina Faso labor market suggests a dynamic picture. The working population 15 years of age and older is nearly 81 percent, and even higher in rural areas, 87 percent. It looks like everyone has to work to compensate for low income. But more than half of this workforce are unpaid family workers. This category of workers provide valuable help in farm and non-farm household 63 enterprises. But it is also obvious that most of the unpaid family workers would choose a different job if better opportunities were present. In addition, the fact that agricultural activities are not market-oriented brings ILO to exclude unpaid family workers involved in subsistence farming from the labor force. Classifying unpaid family workers as unemployed produces a big drop in the working population to less than 38 percent nationally, and only 34 percent in rural areas. Figure 4.1. Working population (age 15 and older) by area of residence 100.0 80.0 60.0 40.0 20.0 0.0 Urban Rural All Urban Rural All Unpaid family worker are occupied Unpaid family worker are not Source. Author’s calculations using the EMC 2014 When considering the main job, rural employment is characterized by very high concentration in agriculture,20 cropping mainly cereals and livestock. Main jobs in the non- agricultural sector represent one-fourth of the workforce, mostly in family enterprises, and less than 6 percent of the whole rural labor force work for a wage. The family enterprises are mostly in small trading and the processing of food products. But the main employment does not reflect the whole picture of the rural labor market in Burkina Faso. Because the agricultural season does not last the whole year and maybe also because of low productivity in agriculture, the population is involved in other activities. One third of individuals aged 15 or older declared having a second job during the last 12-month period. While agriculture is still very important when considering the second job, it is not the most prevalent activity. Many families combine agriculture and livestock activities, having de facto multiple jobs; but the population also diversifies its activities and sources of income with non-farm activities which represent nearly 48 percent of secondary employment versus 46 percent in agriculture; again wage jobs are less prevalent. So the whole rural labor market picture (considering main employment and secondary jobs) show agriculture accounting for nearly than two-thirds of the jobs. The non-agricultural sector is important even in rural areas with one-third of the jobs. Still, there the rural sector in Burkina Faso is weak; during the last 15 years the structure of its labor market remains unchanged and not only is rural poverty still high, but the gap in relation to cities is increasing. There is a negative correlation between agriculture and poverty in rural areas. Individuals from poor households work more in agriculture while better-off households are more involved in non-agricultural activities. The population of the first three quintiles have eight (main) jobs out of 10 from farming (crops or livestock), this proportion drops to 6 out of 10 in the fifth quintile. The comparison between rural and urban areas confirms the negative correlation between 20 The subsequent calculations are made after removing unpaid family workers from the workforce. 64 poverty and agriculture. The share of non-agricultural jobs is higher in urban areas and poverty is lower. While most of the jobs are in agriculture at the national level, there are some slight differences between the regions, explained by the level of urbanization of the region and maybe some other geographic factors. First the Centre region which contains Ouagadougou, the national capital, has 60 percent of its jobs in non-farm activities. This region is specific and its situation and is largely due to the relative dynamism of African capital cities compared to secondary cities and remote areas. For example, the capital has a high percentage of civil servants and it offers opportunities for small trade and non-trade jobs with little start-up capital. The Centre-est region not far from the capital city is the other region with a medium level of non-agricultural activities (35 percent). Figure 4.2: Active population 15 years and older by main occupation and welfare quintile 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 All 1 2 3 4 5 All National Rural Wage agriculture Wage other Farmer Self-employed Source. Author’s calculations using the EMC 2014 Figure 4.3: Active population 15 years and older by secondary employment and welfare quintile 100.0 80.0 60.0 40.0 20.0 0.0 All 1.0 2.0 3.0 4.0 5.0 All National Rural Wage agriculture Wage other Farmer Self-employed Source. Author’s calculations using the EMC 2014 65 Figure 4.4: Active population 15 years and older by main and secondary jobs and welfare quintile 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 All 1.0 2.0 3.0 4.0 5.0 All National Rural Wage agriculture Wage other Farmer Self-employed Source. Author’s calculations using the EMC 2014 The high concentration of rural jobs in agriculture implies that most of the households draw at least part of their income from this sector of activity. More than 90 percent21 of rural households have an income from agriculture (crops or animal husbandry); wage earners in this sector are rare. Rural households draw an income from agriculture regardless of their welfare level; even among households in the fifth quintile, 80 percent have an income from this sector of activity. But agriculture is not the sole source of income. As we have seen earlier, households diversify their activities using multiple strategies. First, they use the fact that household size is important to have multiple individuals in the labor market. Second, the agricultural off-season is the ideal time to be involved in some other activities and improve income and welfare. So households’ members are involved in multiple jobs, either concomitantly with agriculture or during the agricultural off-season; non-agricultural income is also very present in rural areas. In fact, 6 households out of 10 have an income from non-agricultural employment. Most of those households who are involved in non-agricultural activities own their small enterprise (56 percent); wage income is enjoyed by only 13 percent of rural households. Salaried income is present when they are wage earners either from the private sector, or public sector. On the private side, most of the enterprises are located in urban areas where there is infrastructure (electricity, roads) and a potential market. On the public side, service delivery is a real problem in Burkina Faso and civil servants are not often present in remote rural areas. More than a quarter of households enjoy a non-labor market income, mainly from remittances (26 percent). The other type of income (public transfers, real estate, interest from capital, etc.) are scarce (less than 2 percent of households). Unsurprisingly, agriculture is the most important income source. It represents nearly 61 percent of the total rural household income. Less than half of a percentage point of this income comes from wages, so the total agricultural income is from farming. In Burkina Faso farms are small and the production is for self-consumption, so the biggest part of this income is in nature, not so much cash except for those who grow cash crops. Non-agricultural activities account for 36 percent of income, two-thirds of this income derived from self-enterprises. The low level of wage income (less than 7 percent) reflects the rarity of wage earners in country side. Other income represents just 3 percent of total income, most of it from private transfers. In 21 A household can be involved in more than one activity. 66 particular, it is interesting to note the scarcity of public transfers in a country where households are vulnerable to many hazards (climate, shocks, etc.). Just for comparison, the distribution of income at national level shows that 41 percent of total national income comes from agriculture and 53 percent from non-agricultural activities. The Burkina Faso income structure adds another dimension of vulnerability to households. Agriculture is subject to many type of shocks including rainfall variability (drought, flood, etc.), locust attack, prices volatility, etc. For example, a decrease of more than 30 percent in the volume of rainfall and a similar increase a year later is not unusual in Burkina Faso. During the last decade, such decreases or increases have been recorded in Bobo-Dioulasso in 2006, 2010, 2011, 2012 and 2014; in Dori from 2005 to 2007 and in 2014; and in other part of the country.22 In addition to rainfall volatility, price variation is another issue faced by households. For example, cotton prices peaked at $210 per metric ton in 2011, but were around $60 in 2016. Having most of its income derived from agriculture makes income volatile and households vulnerable. Table 4.2: Total household income by source and welfare quintile National Rural All 1 2 3 4 5 All % Having income Agriculture 70.7 95.3 95.3 94.1 92.0 80.9 90.2 Wages 1.9 4.1 2.9 2.2 2.2 1.1 2.3 Farm 70.5 95.3 95.1 94.0 92.0 80.7 90.1 Non-agriculture 68.5 57.7 60.6 61.0 61.7 65.5 61.9 Wages 23.4 10.1 11.7 11.7 13.3 16.3 13.1 Self-enterprises 57.4 53.9 56.2 55.7 57.3 56.4 56.1 Other 29.4 24.6 28.6 28.2 24.4 29.9 27.4 Private transfers 26.4 24.5 27.7 26.9 22.8 27.9 26.1 Others 4.8 0.6 1.4 1.5 1.9 3.2 1.9 % Of total income Agriculture 40.8 75.8 69.9 65.5 60.3 50.6 61.0 Wages 0.4 1.1 0.7 0.6 0.4 0.1 0.5 Farm 40.4 74.7 69.1 64.9 59.9 50.5 60.5 Non-agriculture 53.2 22.4 28.0 32.0 36.7 45.0 35.9 Wages 17.4 2.3 3.3 5.3 5.2 10.1 6.3 Self-enterprises 35.8 20.1 24.7 26.7 31.5 34.9 29.5 Other 6.1 1.8 2.1 2.5 3.0 4.3 3.1 Private transfers 3.2 1.7 2.0 2.3 1.6 3.3 2.4 Others 2.8 0.1 0.1 0.2 1.4 1.1 0.7 Source. Author’s calculations using the EMC 2014 With this configuration there is some degree of income diversification in rural Burkina Faso. We analyze income distribution according to the degree of diversification at the household level. Diversification can be a strategy for accumulating wealth. This can be true for better-off 22 The statistics come from the Statistical yearbook of Burkina Faso, 2014 edition. 67 households, for example those with one or more members with well-paid jobs, who invest in some other activities. But diversification can also be a necessity, for example for poor households who need to complement their low income. In a situation of high poverty, diversification is a good strategy since it helps households to cope with shocks. A household is specialized in a source of income if it derives at least 75 percent of its total revenue from that source; households that obtain less than 75 percent of their total income from four sources are considered not specialized or diversified. In rural Burkina Faso, half of households are specialized in agriculture, roughly 1 out of 6 are specialized in non-agricultural activities, nearly 7 percent on migration and nearly one- third are diversified. Given the efforts deployed by households in agriculture, and effort measured as the proportion of households involved in those activities, this level of specialization is logical. But it also puts households, and particularly poor households, in a difficult situation. Households in rural Burkina Faso are subject to many shocks as noted earlier. Having most of their income come from agriculture makes these households vulnerable because in the event of a shock, they can lose a big part of their income. With the imperfection of the insurance market, as seen previously they have to rely on their own solidarity mechanisms to cope with it. In rural Burkina Faso, income diversification is not necessary correlated with welfare, but the type of specialization is. Better-off households are more specialized in non-agricultural activities, either self-enterprises or wage jobs. Only 8 percent of households in the first quintile are specialized in non-farm enterprises and less than one percent are specialized as wage earners; those statistics are respectively 21 percent and 7 percent of those in the fifth quintile. At the same time, better-off households are less specialized in agriculture and are also less diversified. It seems that even though diversification clearly improves income because households are involved in activities other than their primary ones, it does not really take households to the next level of welfare. Poor households start with agricultural activities, then they diversify, but the additional income is low so those households are still poor and, despite diversification, are either specialized in agriculture or not specialized. The pattern of diversification also varies by region. The Centre region where Ouagadougou is located has two households out of five specialized in non-agricultural activities; one-fourth are diversified and less than 24 percent are specialized in agriculture. On the other hand, high specialization in agriculture is in regions such as Est, Boucle du Mouhoun, Cascades and Hauts-bassins, some being very poor and others not. So at regional level, probably the type of agriculture and its level of productivity can account for some differences. 68 Figure 4.5: Percentage of households by type of specialization and welfare quintile (*) 100.0 80.0 60.0 40.0 20.0 0.0 All 1 2 3 4 5 All National Agriculture Non-agri wages Non-agri selfRuralMigrant Non-specialized Source. Author’s calculations using the EMC 2014 (*) A household is specialized in an activity if 75% of its income comes from that activity Figure 4.6: Percentage of households by type of specialization by region (*) 100.0 80.0 60.0 40.0 20.0 0.0 Agriculture Non-agri wages Non-agri self Migrant Non-specialized Source. Author’s calculations using the EMC 2014 (*) A household is specialized in an activity if 75% of its income comes from that activity Box 4.1. Income aggregate The income aggregate is constructed using the RIGA (Rural Income Generating Activities) approach. Household income has three components: agricultural income, non-agricultural income and other income. Agricultural income consists of wages, crop income and livestock income. Non-agricultural income comprises wages and income from non-farm enterprises, and other income includes public transfers, remittances and other income (rental, interest, dividends, etc.). Income is calculated using different modules of the questionnaire. The employment module collects wage income. There are two modules on production activities, a module on agriculture, and a module on non-farm enterprises. Another module is dedicated to remittances, and a last module collects other non-labor income. The survey was implemented in four visits, and different modules were administered during different visits. Agricultural income. This income comprises agricultural wages and farm income. The EMC survey did not collect livestock data. Livestock is important in Burkina Faso, representing 10 percent of the GDP. So not taking into account livestock underestimates both total income and agricultural income. Wages are taken from the employment 69 module of the second visit. Agricultural wages are defined by the type of industry reported by the respondent. Only those industries related to agriculture are included. Wages comprise salary and different bonuses, in cash. Some workers have some benefits in kind but those have been dropped - first because only a few people received them, and second there were too many outliers. Wages are computed for the primary and the secondary employment at the individual level, and then aggregated to the household level. The agriculture module was administered during the fourth visit, between November 2014 and January 2015 when the harvest was almost completed. Farm income is computed by considering the whole production, crop by crop, valuing it using production prices and subtracting the costs of labor and input. In contrast to wages, farm income is directly computed at the household level. Non-agricultural income. This income is computed the in same way. Wages from non-agricultural work come from the employment module and are computed for the primary and secondary employment. The type of industry is used to classify non-agricultural wages. The module on non-farm enterprises which was administered during the second visit is used to calculate income from enterprises. Information is collected on the revenues received and operating costs for the last month of operation. Using this information, the monthly value-added is computed. This value-added is annualized by multiplying the former by the number of months the enterprise operated during the last 12 months. Then taxes, which are collected for the last 12 months, are subtracted to provide net income. Other income. A module on remittances was administered during the second visit and is used for the calculation of this type of income. A module for all other income sources is also available. It is worth noting that remittances are under-estimated. According to the 2014 EMC, the total amount of private transfers is 74.3 billion FCFA, 48.5 billion from abroad. The total amount of remittances according to the 2014 balance of payment is 179 billion FCFA. It is true that the two concepts do not coincide exactly. For example if someone who lives abroad has a bank account in his country and transfers money to this account, it is a remittance in the sense of the balance of payment, but not a transfer in the sense of household survey since there is no beneficiary. Still the difference is huge and it is not an exaggeration to think that the EMC survey has captured just one third of the remittances. 4.3 Agricultural sector Agriculture is the backbone of the Burkina Faso economy and because the majority of households derive their income from it, improving agricultural productivity is a key driver of poverty reduction. The sector is dominated by traditional subsistence farming. In 2014, traditional crops accounted for three-fifth of total production (by value) with sorghum, millet and maize accounting for 40 percent. The other important traditional crops are rice, peanut and cowpeas. Cash crops, cotton and sesame also total 40 percent of production. The potential of the country in other high value-added crops like fruit and vegetables is weakly exploited. There is not an obvious correlation between diversification in agricultural and poverty at the regional level. Two of the three poorest regions, Nord and Centre-Nord, are highly specialized in traditional crops and particularly in dry cereals (millet and sorghum); the sole cash crop is sesame which represents some 6 percent of total production. But the Sahel region, one of the least poor is also specialized in traditional crops. The difference between the three regions may come from livestock which is important in Sahel but is not taken into account in this analysis.23 On the other hand, being specialized in cotton does not protect against poverty. In three regions, Hauts-Bassins, Boucle du Mouhoun and Cascades, cotton accounts for more than half of production. While the first and the last regions enjoy moderate poverty rates, Boucle du Mouhoun is the region with the second highest poverty headcount. Again, the Hauts-Bassin region is one of the most important in livestock, but livestock is also important in the two other 23 There are three important regions for livestock in Burkina Faso: Sahel, Est and Haut-Bassins. The Nord region, which border Sahel region is among the less dynamic regions for livestock. 70 regions. The Est and Sud-ouest regions are the most diversified agriculturally, producing cotton, sesame, cereals and peanuts, but the two regions are not less poor either. So the pattern of poverty is not linked to the degree of agricultural diversification as can be seen when looking at the pattern of production by welfare quintile. Cotton which is the main cash crop is relatively important in the first quintile (the poorest) and the fifth quintile (the richest). But it is interesting to note that millet and sorghum are relatively more important in the poorest households and rice is more important in the wealthy rural households. Table 4.3: Distribution of agricultural production (by value) by area of residence and type of crop Cereals Tubers Grains Cotton Fruit/Veggies Total Urban 58.2 0.3 23.3 15.9 2.2 100 Rural 47.5 0.6 15.1 36.2 0.7 100 Total 47.8 0.6 15.4 35.4 0.8 100 Source. Author’s calculations using the EMC 2014 Figure 4.7: Distribution of rural agricultural production by welfare quintile and type of crop 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 1 2 3 4 5 Total Millet/sorghum Corn Rice Cowpea Peanuts Sesame Cotton Tubers Fruit/Veggies Source. Author’s calculations using the EMC 2014 71 Figure 4.8: Distribution of rural agricultural production by region and type of crop 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Millet/sorghum Corn Rice Cowpea Peanuts Sesame Cotton Tubers Fruit/Veggies Source. Author’s calculations using the EMC 2014 Burkina Faso agriculture operates at a small level. Total annual production is less than 600,000 FCFA a year. To put this production into perspective, remember that the poverty line is 153,530 FCFA per capita per year and that the average household size in rural area is 8 persons. So an average household needs 1.2 million FCFA to be out of poverty, twice the annual production of an agricultural household in a rural area. This agriculture is household-needs oriented rather than market-oriented. Less than 6 percent of total production was sold during the data collection period. Even though part of that production was still in stock (and in the case of cotton it will be sold), it is hard to imagine that a substantial percentage of cereals and other food crops will be brought to market when household needs are barely satisfied. The scale of production is small, with the average household cultivating less than 4 hectares, and half of households less than 2.5 hectares. While this area of land is not as small as in other countries like Mali and Niger, it would certainly be considered small if the production were for market. Productivity is also low, around 160,000 FCFA per hectare. As seen earlier, yields of the main crops (millet and sorghum, maize, cotton) have not improved much in the two last decades. While population is growing rapidly with obviously additional food needs, production increases come from more areas being cultivated and not from productivity gains. 72 Figure 4.9: Kernel density of logarithm of agricultural production per hectare (in FCFA/Ha) 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 4 6 8 10 12 14 16 18 Log of value harvest per hectare (FCFA/Ha) Productivity all crops Productivity without cotton Source. Author’s calculations using the EMC 2014 The agriculture sector performs largely below capacity. Agriculture is not mechanized and farmers rely on small equipment like hoes, pickaxes, sickles, machetes, etc. The average value of the equipment owned by a farm is 135,000 FCFA, less than 250 dollars. Tractors or other big equipment are nonexistent and less than 2 farms out of 5 use a plough or other type of equipment that relies on animal traction. Labor is abundant, due in part to the large size of households and the contribution of children. The total amount of labor on a farm is 411 days, 20 percent of this amount of work provided by children under the age of 15. But although labor is abundant, the whole labor force is shared among many activities. Assuming that a regular worker works 200 days a year and there are on average four adults in a household in a rural area, the amount of work provided by adults on a farm represents less than 50 percent of household potential. In addition to the under-utilization of labor, the use of inputs is also limited. Half of the farms use manure, but only 4 out of 10 use chemical fertilizers and one third rely on pesticides. And according to the 2015 agricultural survey, less than 5 percent of farms practice irrigation. The pattern of agriculture in Burkina Faso cannot be fully understood without taking cotton into. There are huge disparities between cotton producers and other agricultural households. Households involved in cotton are less likely to have female heads (4 percent versus 13 percent for non-cotton), household heads are more educated, and households are smaller in size. Total production is 8 times greater than in non-cotton households; area cultivated is two times greater and productivity is 3.5 greater. Cotton producers also mobilize more inputs and have more equipment. Household size is slightly larger among households that produce cotton but family labor is even more heavily used in these households, providing on average 42 percent more family labor than non-cotton producers. Cotton-producing households use also non-family labor (51 days of work a year) while this resource is virtually unused by non-cotton-producing 73 households. Nearly all cotton producers use chemical fertilizers and 90 percent use pesticides. The value of their equipment is three times greater than that of non-cotton-producing households. Table 4.4: Characteristics of rural agriculture by welfare quintile 1 2 3 4 5 All Head being female (%) 11.9 10.2 13.2 12.0 16.1 13.1 Average age of the head 51.2 49.3 48.4 46.4 43.2 47.0 % Head with some education 7.1 7.3 10.3 10.3 19.5 12.0 Dependency ratio 1.74 1.55 1.48 1.28 1.03 1.36 Average number of plots 3.9 3.8 3.6 3.3 2.9 3.4 Production total (Fcfa) 580330 578313 529011 551067 679695 589689 Production sold (Fcfa) 24483 26671 36107 34455 39864 33326 Area cultivated (Ha) 4.1 4.4 3.9 3.6 3.1 3.7 Men family labor (days) 171 186 167 161 130 159 Women family labor (days) 195 220 188 158 121 168 Kids family labor (days) 110 123 102 76 43 84 Non-family labor (days) 8 13 12 11 16 12 Non-family wages (Fcfa) 4059 7599 6923 9572 12518 8819 Non-family wages if labor (Fcfa) 16988 25827 23569 32788 35956 29260 Use organic fertilizer (%) 57.6 59.8 57.7 57.5 49.4 55.6 Value of organic fertilizer (Fcfa) 621 1045 511 959 626 744 Value of organic fertilizer if use (Fcfa) 1078 1749 885 1585 1267 1321 Use chemical fertilizer (%) 37.0 43.8 43.3 42.4 40.5 41.5 Value of chemical fertilizer (Fcfa) 33713 49038 46463 46494 45849 44886 Value of chem fertilizer if used (Fcfa) 91195 112025 107219 109583 113144 108203 Use pesticides (%) 25.7 32.5 31.1 32.1 33.3 31.4 Value of pesticides (Fcfa) 9661 13452 14246 13193 14471 13296 Value of pesticides if use (Fcfa) 37655 41341 45837 41095 43399 42318 Acquire seeds in the market (%) 92.4 92.0 90.3 87.8 76.5 86.4 Value of seeds bought (Fcfa) 7873 226634 11415 14043 11252 47185 Value of seeds bought if used (Fcfa) 8520 246456 12647 15986 14704 54642 Use tractors or similar equipment (%) 0.5 0.3 0.6 0.8 0.6 0.6 Use plough or medium equipment (%) 42.3 43.2 41.3 38.4 30.6 38.0 Use small equipment (%) 95.9 95.8 95.0 92.8 81.9 91.0 Equipment value (Fcfa) 123472 134910 145188 140745 129831 134949 Equipment value if have it (Fcfa) 128756 140783 152828 151666 158481 148314 Farm more than an hour to market (%) 33.5 31.2 32.1 35.4 32.8 33.0 Farm more than an hour to transport (%) 51.0 53.2 49.8 51.2 49.5 50.8 Farm more than an hour to the road (%) 41.2 42.1 38.0 37.5 35.9 38.4 Having an account Bank or MFI (%) 4.8 7.7 8.7 9.3 18.8 11.1 Source. Author’s calculations using the EMC 2014 An estimation of an agricultural production function provides some insight into the low level of productivity in Burkina Faso. The dependent variable is the logarithm of the value of production per hectare. The explanatory variables are the household characteristics and its head, 74 the farm characteristics including area cultivated, labor, level of input used, equipment, etc. and dummy on the type of crops grown and geographic variables. The model is estimated for all households and then separately for household that grow cotton and those that do not. The results are in Table A8 in the annexes. First, the results show that household characteristics are correlated with productivity. The sign of the coefficient associated with the variable identifying the gender of the head is positive, and female-head households are on average 14 percent more productive than male-head households; but the result does not hold for households cultivating cotton. The effect of human capital variables is mixed. The experience, measured by the age of the head has a positive effect on productivity, but education variables have no effect. Actually education is generally low in rural Burkina Faso and the absence of effect is not a surprise. The effect of land size (measured by the natural log of the area cultivated) is negative and strongly significant in all regressions indicating that productivity declines with land size. This result is consistent with the inverse relationship between productivity and land size found in many other studies.24 Since households involved in cotton have on average larger farms than non-cotton producers, this result may give female-head households a relative advantage in terms of productivity. Second, labor and non-labor inputs are also strongly positively correlated with productivity. The elasticities of male household labor, female household labor, child family labor and non-family labor are all positive with a positive sign. The highest is the elasticity of male family labor which is 0.07. In other words, a 10 percent increase in male family labor (in number of days per hectare) is associated with a 0.7 percent increase in productivity. Female labor and child labor elasticities are respectively 0.04 and 0.01 and the elasticity of hired labor is 0.06. This results suggest that men working on their own farms are more productive than persons coming from outside the household; but women and children are less productive than hired labor. As for non-labor input, the coefficients associated with those variables are also significant. The value of chemical fertilizers and pesticides have a positive effect on productivity. If the value of chemical fertilizer increases by 10 percent, productivity increases by 2.3 percent; we have a similar effect with the use of pesticide. Labor and non-labor inputs are used in conjunction with capital to generate production. Despite the fact that capital resources are low, the coefficient associated with the value of agricultural capital per hectare (in log form) is positive and significant. The coefficient estimate suggests that a 10 percent increase in the value of agricultural capital leads to a 0.7 percent increase in productivity. While this elasticity is relatively low, the positive correlation shows that there is room to improve productivity with a minimum of equipment involved in the production process. Other factors considered also have solid correlation with productivity. The type of crops grown by the household affects productivity. In fact, growing dry cereals is associated with a decrease in productivity while rice, sesame, cotton and tubers are associated with an increase in productivity. The strongest effect is with cotton: a household growing this crop doubles its productivity, after controlling for all other characteristics. Rice also has a significant effect, with 24 See for example Carletto, Savastano and Zezza. “Fact or Artefact: The Impact of Measurement errors on the Farm Size - Productivity Relationship,� Journal of Development Economics 103 (2013): 254-261. 75 an increase of 17 percent compared to households not growing rice, after controlling for all the other factors. The impact of shock also is negative; households affected by a shock see an 8 percent decrease in its productivity compared to households that have not been affected by any shock. Having any bank account, either in the formal banking system or in a micro-finance institution, is associated with an increase in productivity of nearly 12 percent. Finally, regions, which take into account unobservable characteristics, also have a significant coefficient. These results provide insight into the low productivity of Burkina Faso agriculture and explains the difference in welfare among households. Wealthier rural do not have larger plots. But they use more fertilizer and pesticides (when measured per hectare), the value of equipment is higher and they use more hired labor. All these factors are the main determinants of productivity and explains why they have better income. In addition their demographic composition (lower in size and lower dependency ratio) explain differences in income and in welfare. In fact, rural households face several poverty traps which hamper their ability to improve productivity. First, agriculture is not mechanized whereas equipment has a real impact on productivity. In addition, the imperfection of the credit market makes it difficult to borrow and acquire equipment. Only 11 percent of households have a bank account and the poorest households in the lowest quintile are more penalized, with less than 5 percent of them having a bank account versus 19 percent of those in the highest quintile. Without having access to the formal finance system, farmers rely solely on family and friends who have weak financial capacity to finance equipment and will rarely lend their money for the medium term. In 2014, in less than 2 percent of existing rural areas have loans been granted for equipment. It is interesting to notice that if credit were accessible, farmers would be willing to borrow, as they do when it comes to other matters. For example, nearly one-third of loans have been to acquire input, one- third coming from friends and family, one-third from a cooperative and one-third from a supplier or the formal banking system. The second poverty trap is the low use of fertilizer and pesticide, which also has a negative impact on agricultural productivity. Access to labor input is better, but even that is not optimally used by households. The third point is the specialization in households. Most of the areas cultivated are mainly used for dry cereals, crops with a negative impact on productivity. Cotton, rice and tubers have a better impact on productivity and are probably a pathway to improving it; of course it can be worth exploring other potential high productivity crops like fruit and vegetables. In addition to all these factors, households have limited access to market. Half of households have to walk more than an hour to find transportation and 38 percent are more than an hour from the nearest road. In such conditions, even if farmers were able to produce a surplus, they would have difficulty getting it to market and selling it at a fair price. 4.4 Non-farm enterprise income Non-farm enterprise is the second most important income source in Burkina Faso rural areas. According to the 2014 EMC survey, there are 2 million non-farm enterprises in rural areas, on average 1.1 enterprise per household. More than 60 percent of households rely on non- farm enterprises for their livelihood. These enterprises belong equally to households at all level of welfare, the poorest and the wealthiest; 61 percent of households in the first quintile have a household enterprise and this percentage is 63 percent in the fifth quintile, meaning that poorer and better-off households rely on this source of income. Most of the household enterprises belong to women (three out of five), who usually run them as their secondary job in addition to 76 their activities in agriculture. Only one-third of rural enterprises are busy all year round. Half of them are seasonal activities, 8 months a year on average. The remaining may not operate at all due to difficulties such as lack of customers or lack of inputs. The average age of the owner is 36, showing that the person has some experience of the activity. These enterprises are concentrated in activities with few barriers to entry (low capital and relatively unskilled). The most represented sectors are retail (40 percent), food processing (20 percent) and extractive industries (1 in 6). Such activities may start with a low level of equipment and without any specific technical training. Manufacturing that requires more technical skills accounts for only 11 percent of enterprises. The demographics of non-farm enterprises present a mixed picture. The average number of years of existence is 7, showing that the enterprises last enough years which might allow them to grow; in fact, one enterprise out of six is more than 15 years old. But, at the same time, half of them are less than 5 years old and one-fourth were created during the last three years, showing clearly less dynamism. The older enterprises are found in manufacture, food processing and services. The relationship between family enterprises and the administration are nonexistent and most of them are in the informal sector. A formal or modern enterprise is known by the administration, at least is registered with the tax authorities. In addition, it must have a basic formal accounting system. Household enterprises in Burkina do not meet any of these two criteria. Only 0.7 percent are registered with the tax authorities and 0.1 percent have a formal accounting system. These enterprises may be unregistered because the tax authorities are less present in the countryside, but that is probably not the reason as, even at national level, only 1.6 percent of enterprises are known by the tax authorities. The truth is that most non-farm enterprises are so small that it is not worth trying to collect any tax. Non-farm enterprises also operate at a small level, and working conditions are precarious. The main place of business is outdoors, either a specific spot by the side of the road or a market place or as a street vendor. One third of the enterprises operate at home and only 7 percent own a specific business premises. The place where the business operates depends on the type of industry. In the case of extractive industries, the virtually all of them work outdoors. Manufacturing and food processing have a local plant, 60 percent of them operating from home. Only one third of retailers work from home, most of them are in the street. In addition to not having a business premises, activities are carried on without basic commodities such as electricity or water. This precariousness in the exercise of the activity makes these companies vulnerable (and perhaps also the households to which they belong), as adverse weather conditions may force them to stop working. In fact, the mobile phone is the only modern working tool that is now entering the world of individual enterprises. The low level of business is evident in the means of production used. In addition to the absence of a business premises and basic commodities, the start-up capital of the average enterprise is 80,000 FCFA (less than $150) and consists essentially of tools and basic equipment. Less than 3 percent of enterprises have machines, less than 6 percent have motorbikes and automobiles, and less than 1 percent have furniture. At this low level of business, it is difficult to achieve good productivity and a decent income. Start-up capital correlates to level of welfare. 77 Very small enterprises with an average start-up capital of 22,000 FCFA belong to the poorest households while the start-up capital of the wealthiest households is more than ten times greater. So although owning a family enterprise does not make the difference between households, the size of the enterprise is discriminant, with bigger enterprises belonging to better-off households. Moreover, growth potential is limited. Growth requires more financing. But the relationship between owner enterprises and the banking system is limited and only 5 percent own a bank account in the formal system or in a micro-finance institution. Although more than 94 percent obtained loans during the last 12 months, it was through informal channels, particularly parents and friends, and the loans were small and not dedicated to developing the enterprise. Non-farm enterprises’ labor comes mainly from family, and these enterprises are created more for survival than to create wealth. An average individual enterprise employs just its owner and at times some unpaid family workers. Less than 3 percent of the family enterprises use hired labor (1 percent in the first quintile and 4 percent in the fifth quintile). To put it differently, an individual enterprise in Burkina Faso is created only to employ the person who created it. In addition, the human capital is low; only one out of ten owners have attended school. This low level of education cannot be supplemented by other family members as they are in the same situation. These statistics reflect the low skill level of the workforce, and this low skill level combined with the low level of capital can only result in the production of low quality products that cannot always compete with imports. The small scale of production largely justifies the modest outcome from non-farm enterprise business. The annual turnover is 662,000 FCFA. A unit of production creates on average a value-added of 550,000 FCFA per year. Since hired labor is rare and very few units pay taxes, the total value-added turns out to be very close to net income which is 534,000 FCFA. This net income varies from 192,000 FCFA for enterprises in the first quintile to nearly 10 times greater for enterprises in the fifth quintile, making a massive difference. As earlier, the performance of non-farm enterprises is analyzed using a regression technique. The dependent variable is productivity, measured by the value-added per hour of work. The explanatory variables include the characteristics of the owner (gender, human capital) and of the enterprise (namely the total hours worked, the use and value of equipment, and some other characteristics). The model is estimated using the Heckman technique, at the first stage the probability of owning a non-farm enterprise and productivity at the second stage; the results are in Table A5 in the annex. Productivity depicts an inverse relationship with the annual volume of hours worked as in the case of farm enterprises. Enterprises run by women are also less productive than those run by men, the difference being estimated at 56 percent. Since 6 out of 10 non-farm businesses belong to women, this result provides one explanation of low productivity and income derived from these enterprises. Productivity is also correlated with experience, the experience of the owner and the experience of the enterprise itself. The more the person is in the business, the more his enterprise gains in terms of productivity; and the more the enterprise is operating, the more its productivity improves. Productivity is positively correlated with the value of capital and hired labor. The elasticity of productivity on capital is 0.017, meaning that an increase of the value of capital of 10 percent will increase productivity by 0.17 percent. So owning machines and other equipment increases productivity. As for labor, an enterprise that hires people is 34 percent more productive that one that only uses family workers. Having some 78 degree of formal existence, for example having registered with the tax authorities, having a professional place of work, having electricity, or owning a mobile phone, improve the enterprise’s performance substantially. These results confirm that productivity and income are positively correlated with the size of non-farm enterprises. When the enterprises have a minimum size which allow then to own some equipment, if they operate in a professional local and if they have a minimum degree of formal existence, then their performance improves substantially. The issue with a low income- generating process is that the creation of non-farm enterprises follows population growth. Enterprises are created at a very small scale of production, more to fill the non-working time encountered during the agricultural off-season. But without qualifications, skills and access to credit, those enterprises remain too small to provide decent incomes to their owners and they struggle to impact poverty. 4.5 Private transfers 4.5.1 The size and origin of private transfers Private transfers are an important source of household income in many developing countries, but they present a real challenge of measurement from household surveys. According to World Bank25, at the macroeconomic level, international remittances which are part of private transfers were estimated $582 billion in earnings in 2015. In 27 countries, remittances were equal to more than 10 percent of gross domestic product (GDP) in 2014; in ten countries they were equal to more than 20 percent of GDP. At the household level a survey on different research on private transfers in 9 countries show that a minimum of a quarter of households receive private transfers26. Having a better understanding of private transfers is important for designing social policy because they provide economic and social benefits similar to those of public programs, for example pension for the senior or insurance in the case of unemployment or a shock. In Burkina Faso in 2014, the World Bank estimation of international remittances is $396 million (roughly 200 billion FCFA) while the calculation of the 2014 household survey is 74 billion FCFA for all transfers, with 40 percent (30 billion FCFA) originating from outside the country. Clearly this household survey underestimates private transfers, since the international private transfers as measured by the survey represent just 15 percent of remittances as estimated by the World Bank. Measuring transfers from household survey poses the same difficulties as measuring other source of income, reluctance to declare income, memory effect, etc. However, the underestimation underlined above is overstated because the concept of remittances (as defined by the Bank) and transfers (as seen in household surveys) do not coincide. In particular, remittances include compensation of employees and savings and direct investment of migrants which are not include in transfers as capture by surveys. But we keep in mind that there is an important underestimation of private transfers and be cautious when drawing conclusions. In Burkina Faso, a quarter of household enjoy private transfers and they represent 3.2 percent of the total household income. As it has been stated earlier, private transfers seems 25 http://www.worldbank.org/en/topic/migrationremittancesdiasporaissues 26 Daniel Cox & Emmanuel Jimenez, Achieving social objectives through private transfers: A review, The World Bank Research Observer, July 1990 79 largely underestimated. However they represent the fourth source of income after agriculture, wages and non-farm enterprises and largely before public transfers. Private transfers have different motives and most of them are fulfilled in Burkina Faso. First there are customs and social norms and the more valid population has the moral obligation to support the others, either by altruism or for self-interest. In a country where about 80 percent of the population lives in rural areas and depends on a low productive rainfed agriculture, many are poor and they benefit from the generosity of the less poor. But private transfers can also be the consequence of economic shocks or other motives. For example through mutual agreements, households use transfers as risk-sharing mechanisms. These mutual insurance allow household to transfers transitory income among them with the aim to smooth their consumption. Also the principle of reciprocity motivates transfers between households. In this view, each transfer acts as counterpart of previous transfers involving family members and friends in social arrangements. Households receive transfers both inside and outside the country, and Côte d’Ivoire is the main origin of such transfers. The most important fraction of Burkinabe who migrated live in Côte d’Ivoire. These millions of individuals who work essentially in coffee and cocoa plantations and industries send money to their relatives in Burkina Faso. These transfers represent a quarter of the total transfers (inside and outside the country) and 70 percent of transfers from outside the country. Doing so, populations in Burkina Faso benefit from positive externalities of the economic health of Côte d’Ivoire. Data also show that transfers from Côte d’Ivoire are directed to rural areas (34.2 percent against 12.7 percent in urban areas). This can be explained by the fact that plantations in Côte d’Ivoire attract agricultural labor force largely available in rural Burkina Faso. In return, these workers send transfers to their relatives (parents, families) who live in rural areas in Burkina Faso. Transfers from urban Burkina Faso (Ouagadougou the capital city and other urban areas) represent 40 percent of the total. If transfers are motivated by altruism, one can understand why inside country transfers come more from urban than rural areas. Urban households are richer than rural households and the first are more likely to send transfers than the second according to the principle of altruism. However, transfers from rural areas are also important. Figure 4.10: Share of transfers by their origin 40.0 35.0 30.0 Share of transfers 25.0 20.0 15.0 10.0 5.0 0.0 Ouaga Other urban Rural Côte d'ivoire France/Italy WAEMU/Ghana Other Origin of transfers 80 Source: Author’s calculation using 2014 EMC household survey 4.5.2 Motive of private transfers Transfers play a key role in equalizing welfare among members of family in a broader sense. Intergenerational transfers are important in Burkina Faso. Indeed of 35 percent of the total amount of transfers flow from the children to their parents, playing probably the role of pensions in a country where most of the active population is involved in informal employment (either in agriculture or in urban activities) without a possibility to have a formal one. At the same time, 7 percent of the transfers represent a support from the parents to their kids. The low percentage of transfers from parents to kids is not a surprise since most of the children live in the same households with their parents until they are able to move to build their own household. But is also important to point out that the bigger percentage (nearly half) of the transfers flow from other family members than children and parents, in particular siblings. It is very common in African countries in general and in Burkina Faso in particular for the elder to support the younger between siblings, and it translates in transfers among those households. This type of transfers are not only motivated by altruism, as stated earlier they are a substitute of insurance in the case of shock (unemployment, natural hazards, etc.). The solidarity mechanism induced by transfers go beyond the family since 10 percent of the transfers are from non-family members. Households receive transfers for many reasons, but supports to family appear to be the main motivation of transfers in Burkina Faso. Regardless of socio-demographic characteristics of the head, transfers are provided to households as assistance for household needs. This means that transfers provided to households are essentially one time transfers with the aim to smooth household’s consumptions. Only 10 percent of transfers are directed to funding investment in education, health or economic activities. As it appears from this analysis, the motivation of household transfers in Burkina Faso is essentially to smooth household consumption instead of productive investments. But as seen in introduction the underestimation can play a role here. An important part of transfers can be on own investment of a migrant in his own country, and since no household benefit from it, this type of transfers are not capture by household surveys. Figure 4.11: Share of transfers (in the total) by motive 100.0 90.0 80.0 70.0 Share of transfers 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Support Education Health Events Economic activities Others Reasons of transfers 81 Source: Author’s calculation using 2014 EMC household survey 4.5.3 Private transfers, poverty, inequality and vulnerability Private transfers contribute to alleviate poverty and to make other household less vulnerable. The probability of receiving transfers is a decreasing function of pre-transfers annual per capita income (Figure 3, left part). Households at the left end of the distribution of the pre-transfers per capita income are nearly half receiving transfers, while only 20 percent of the richer ones (as measured by income per capita) enjoy transfers. Clearly the flow of transfers goes more to the poorer, contributing to alleviate poverty. Those transfers seems to reduce also inequality with the Gini index of per capita income decreasing by 2 percentage points before and after transfers. But in absolute value, the curve of total transfers function of pre-transfers per capita annual income has a U, first decreasing for the poorer households and then increasing for the better-off households who receive the most important transfers (Figure 3, right part). This result is not easy to explain without further investigation. But one assumption is that as seen earlier, the poorest households have a high frequency to receive transfers and this high frequency is translated into some important amount while the richest, even if the frequency is low, when they benefit the amounts received are more important because they can invest for example in having migrants abroad with higher economic power. Figure 4.12: Probability of receiving transfers and annual transfers by log of pre-transfers per capital annual income 0.5 11.1 Probabiliy of receiving transfers 0.45 11 Log of annual transfers 0.4 10.9 10.8 0.35 10.7 0.3 10.6 0.25 10.5 0.2 10.4 0.15 10.3 4 9 14 4 9 14 19 Log of pre-transfers per-capita annual income Log of pre-transfers annual per capita income Source: Author’s calculation using 2014 EMC household survey The analysis of transfers by household characteristics confirms that the more vulnerable and poor household have higher probability to benefit from transfers. The gender of a household’s head is correlated to it probability to have a transfers. Data show that households headed by women are more likely to receive transfers compared to those headed by men. About 46 percent of households headed by a woman receive a transfer while only 23 percent of households headed by a man received a transfer. The average amount of transfer received by a household headed by a woman (65,281 CFA) is 2.6 times the average amount received by a household headed by a man (25,239 CFA). Transfers represent 12.7 percent of the total income of households headed by women while they represent 2.5 percent of the total income of households headed by men. The fact that households headed by women received more transfers than those headed by men can be explained in part by their vulnerability. It has been estimated 82 that the probability for a household being poor is 30 percent higher when the head of household is a woman to compare to a household having a man as a head. But also many households headed by a woman depends on the financial support of a husband who is not consider as a household member (in the case of polygamy) according to the survey definition. The support to the more vulnerable population can also be seen when considering handicap. A household having a disabled person as a head has a probability of 22 percent higher to receive transfer than a household with a non-handicapped head. This result reinforce the role of transfers as safety nets for the population facing adverse conditions in their life. Household with a head at the working age are less likely to receive transfers than those having a young or a senior as a head. The probability of receiving transfers in function of age also have a U shape, with probability being relatively high at the young age and at the higher age, and low in the middle of the life. The shape of this curve and the ones of the other curves of transfers and income function of age shows that private transfers are directed to those who need them most, at least in terms of frequency. As it can be seen in Figure 5 (the left and the right), the amount of transfers are relatively high when personal income is low and become low when personal income is higher. In this case, transfers also respond to some intergenerational and solidarity pattern. The young receive transfers from their parents but also their elder siblings. The senior head have active children who provide them financial assistance through transfers. For example among the senior (50 years and older), 34 percent of the recipients received transfers from their children. In addition, in Burkina Faso, older individuals play central role in the social structure and have an important social network. In return of their social engagement, they received many transfers from members of their community. Figure 4.13: Probability of receiving transfers by head of household age 1 0.9 Probability of receiving transfers 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 20 40 60 80 100 120 Age Source: Author’s calculation using 2014 EMC household survey 83 Figure 4.15: Amount of transfers and total income by head of household age 11.6 13.4 13.2 11.4 Log of annual transfers (FCFA) Lof of annual income (FCFA) 13 11.2 12.8 11 12.6 12.4 10.8 12.2 10.6 12 10.4 11.8 11.6 10.2 0 50 100 150 11.4 0 20 40 60 80 100 120 Age Age Source: Author’s calculation using 2014 EMC household survey 84 5. CONCLUSIONS Despite some major progress during the past 15 years, poverty is still high in Burkina Faso, and in rural areas in particular where nearly half of the population still live below the national poverty line. Poverty projections show that, with the current trend, the country will not be able to reach one of the twin goals, which is eradicating poverty by the year 2030. Poverty in the country is not only a deficit in consumption, but it has multiple aspects. Different forms of food insecurity affect the population. Foodstuffs are not equally accessible to all strata of the population. Some people are chronically food insecure while some others are regularly affected by shocks, making them transiently food-insecure at some periods of the year. In addition, the living conditions of the population are difficult. Houses are built of poor materials. Access to electricity, energy, sanitation and other basic commodities is limited. Many factors explain the poor situation of the population of Burkina Faso. The first one is demography. Burkina Faso has very rapid population growth, around 3 percent a year. This rapid population growth is the consequence of very high fertility, 6 children per woman. Like many countries in Sub-Saharan Africa, Burkina Faso intends to become an emerging market economy in the next 25 years. The current population structure, with a wide- base age pyramid poses a real challenge to this objective. Actually, Burkina Faso cannot reach a high stage of development without undergoing a demographic transition. Demographic transition means a shift from a largely rural agrarian society with high fertility and mortality rates to a predominantly urban industrial society with low fertility and mortality rates. At an early stage of this transition, fertility rates fall, leading to fewer young mouths to feed. During this period, the labor force temporarily grows faster than the population dependent on it, freeing up resources for investment in economic development and family welfare. Other things being equal, per capita income grows faster too.27 This step can take more than 20 to 30 years or even longer. In Burkina Faso, mortality is falling but it is not yet the case for fertility. The growth enjoyed by Burkina Faso during the last 15 years could have been more pro-poor with lower population growth. High fertility rates are a real challenge for growth and poverty reduction and getting a better understanding of the determinants of fertility and the channels by which it can be reduced is a path for better results on poverty reduction. The second challenge is education. Human capital in Burkina Faso is extremely low. Education improves human capital and has a positive impact on income and on poverty reduction. Education and in particular women’s education has a positive impact on many other phenomena, including the use of contraceptives and fertility, under-nutrition, etc. The country has made a lot of progress but that progress is limited. The completion rate in primary school is better than it was 15 years ago, but expected years attendance at school is still low, 7.5 years, meaning that the average child completes only up to the first year of secondary school. There is no doubt that pro-poor growth must begin with robust agricultural growth. However, while this 27 Lee & Mason, Finance and Development, a Quarterly Journal of the IMF, September 2006, Volume 43, Number 3. 85 has proven elusive in the past, the aspirations remain and the new world offers some prospects for success. The third challenge for poverty reduction in Burkina Faso is improving productivity in agriculture. This sector is the main income source for the vast majority of the population. But performance in the sector is poor. Agriculture is characterized by low mechanization, low access to inputs and improved seeds, to labor and to credit. Most farmers continue to use traditional techniques which have proven to be ineffective in boosting productivity. The potential of the country in irrigation is not totally exploited. But, at the same time, there are many initiatives in this sector. It is important to conduct a rigorous evaluation of the investment in the sector in order to be able to inform policy makers on the potential of the future projects on reaching the goal of improving productivity and income. Agriculture also needs to invest in research to improve seeds and techniques, invest in different type of infrastructures (storage, processing, roads, etc.) to facilitate access to market. The issue of diversifying agriculture towards new crops with higher potential value-added and choosing a variety of crops with higher productivity are paths to growth and poverty reduction. The fourth challenge is to increase the resilience of the population by implementing safety nets. The poorest accumulate too many handicaps (low human capital, low productive capital, no access to credit, etc.), they are too far from the poverty line and are less likely to benefit from poverty reduction strategy without specific targeting. The poorest are also vulnerable to the many shocks that affect the country and are most affected by food insecurity. In the absence of social security mechanisms, households rely on their own resources to cope with adverse situations. Many households rely on friends who are as poor as themselves. Others have to sell their assets, making them more vulnerable. A system of safety nets would prevent people to fall into deep poverty, to keep their dignity and sometimes to keep their children at school. In addition to overcome the negative effects in case of shocks, social safety nets improve human capital (education, preventive and curative health) and allow beneficiaries to build productive capital. 86 ANNEXES Table A1. Growth Inequality Decomposition National Urban Rural 2003-2009 Average Average Average 2003 2009 2003 2009 2003 2009 effect effect effect Poverty headcount 51.8 46.7 24.6 27.9 57.9 52.6 Change -5.1 -5.1 -5.1 3.3 3.3 3.3 -5.3 -5.3 -5.3 Growth -2.4 -2.6 -2.5 7.5 7.6 7.6 -4.1 -4.1 -4.1 component Redistribution -2.6 -2.8 -2.7 -4.3 -4.2 -4.2 -1.2 -1.2 -1.2 Residual -0.2 0.2 0.0 0.1 -0.1 0.0 0.0 0.0 0.0 2009-2014 Average Average Average 2009 2014 2009 2014 2009 2014 effect effect effect Poverty headcount 46.7 40.1 27.9 13.7 52.6 47.5 Change -6.6 -6.6 -6.6 -14.2 -14.2 -14.2 -5.1 -5.1 -5.1 Growth component -3.2 -3.7 -3.4 -9.1 -8.5 -8.8 0.2 0.4 0.3 Redistribution -2.9 -3.4 -3.2 -5.7 -5.1 -5.4 -5.5 -5.3 -5.4 Residual -0.5 0.5 0.0 0.6 -0.6 0.0 0.2 -0.2 0.0 2003-2014 Average Average Average 2003 2014 2003 2014 2003 2014 effect effect effect Poverty headcount 51.8 40.1 24.6 13.7 57.9 47.5 Change -11.7 -11.7 -11.7 -10.9 -10.9 -10.9 -10.4 -10.4 -10.4 Growth component -5.3 -6.6 -5.9 -1.5 -0.9 -1.2 -4.0 -5.1 -4.5 Redistribution -5.2 -6.4 -5.8 -10.0 -9.4 -9.7 -5.2 -6.4 -5.8 Residual -1.3 1.3 0.0 0.6 -0.6 0.0 -1.2 1.2 0.0 Source: Authors calculations using INSD surveys QUIBB-2003, EICVM-2009, EMC-2014 Table A2. Sectoral Decomposition of a Change in Poverty Headcount 2003-2009 2003-2014 2009-2014 Poverty in period 1 52.734 52.734 47.972 Poverty in period 2 47.972 40.107 40.107 Percentage Percentage Percentage population Absolute Percentage population Absolute Percentage population Absolute Percentage base year change change base year change change base year change change Change in population 87 Agriculture urban 3.7 -0.046 1.0 3.7 -0.521 4.1 4.3 -0.563 7.2 Industry/Construction urban 1.9 0.052 -1.1 1.9 -0.242 1.9 2.4 -0.363 4.6 Commerce/Services urban 7.0 0.486 -10.2 7.0 -0.397 3.2 7.8 -0.981 12.5 Unemployed urban 2.9 0.009 -0.2 2.9 -0.385 3.1 4.3 -0.578 7.3 Agriculture rural 77.9 -4.127 86.7 77.9 -8.162 64.6 68.3 -3.533 44.9 Industry/Construction rural 1.1 0.038 -0.8 1.1 -0.235 1.9 1.7 -0.409 5.2 Commerce/Services rural 2.8 0.147 -3.1 2.8 -0.109 0.9 2.9 -0.269 3.4 Unemployed rural 2.7 -0.135 2.8 2.7 -0.085 0.7 8.4 0.155 -2.0 Change in poverty Total Intra-sectoral effect -3.575 75.080 -10.135 80.270 -6.542 83.180 Population-shift effect -1.504 31.570 -2.832 22.430 -0.969 12.320 Interaction effect 0.317 -6.650 0.341 -2.700 -0.354 4.500 Change in poverty -4.762 100.0 -12.627 100.0 -7.865 100.0 Source: Authors calculations using INSD surveys QUIBB-2003, EICVM-2009, EMC-2014 Table A3. Ranking of regions using the poverty headcount (from the least poor to the most poor region) PC0 PC1 PC2 PC3 PC4 PC5 Average Hts Bassins 4 4 4 4 4 4 4.0 Bcle Mouhoun 12 12 12 12 12 12 12.0 Sahel 2 2 2 2 2 2 2.0 Est 10 11 9 11 9 9 9.8 Sud Ouest 7 8 10 6 8 10 8.2 Centre Nord 9 5 6 5 7 7 6.5 Centre Ouest 11 10 8 9 11 11 10.0 Plateau central 8 9 7 10 10 8 8.7 Nord 13 13 13 13 13 13 13.0 Centre Est 5 7 11 8 5 5 6.8 Centre 1 1 1 1 1 1 1.0 Cascade 3 3 3 3 2 3 2.8 Centre sud 6 6 5 7 6 6 6.0 Source: Author’s calculations using INSD survey EMC -2014 Table A4. Regression of the logarithm of per capita consumption National Urban Rural Parameter T-Student Parameter T-Student Parameter T-Student Sociodemographics Kids less than 5 -0.104 -15.9 -0.149 -10.5 -0.094 -12.9 Kids less than 5, squared 0.008 8.6 0.017 5.4 0.007 6.7 Boys 5-14 -0.134 -19.2 -0.154 -9.9 -0.117 -15.2 Boys 5-14, squared 0.013 10.3 0.016 4.1 0.011 8.3 Girls 5-14 -0.139 -18.3 -0.179 -11.1 -0.125 -14.7 Girls 5-14, squared 0.017 11.2 0.028 6.6 0.014 8.8 Men 15-64 -0.040 -4.9 -0.089 -6.6 -0.020 -2.0 Men 15-64, squared 0.002 2.0 0.007 3.9 0.001 0.5 88 Women 15-64 -0.039 -5.2 -0.080 -6.1 -0.022 -2.5 Women 15-64, squared 0.004 5.0 0.008 4.6 0.003 3.2 Men 65 or more -0.080 -2.0 0.024 0.3 -0.083 -2.0 Men 65 or more, squared 0.050 1.7 -0.064 -0.9 0.060 1.9 Women 65 or more -0.120 -5.0 -0.118 -2.4 -0.093 -3.6 Women 65 or more, squared 0.009 0.6 -0.002 -0.1 0.003 0.2 Head married (yes) -0.110 -2.2 -0.282 -3.2 0.161 2.6 Head is a woman (yes) -0.352 -15.7 -0.439 -12.8 -0.147 -4.9 Head not a Burkinabe (yes) -0.280 -4.9 -0.316 -4.6 -0.057 -0.5 Head is disabled (yes) -0.045 -2.2 -0.005 -0.1 -0.059 -2.5 Age Age of head 0.005 2.4 0.012 3.3 0.001 0.3 Age of head squared 0.000 -3.4 0.000 -3.4 0.000 -1.1 Age of spouse -0.010 -4.1 -0.010 -2.1 -0.010 -3.4 Age of spouse squared 0.000 4.7 0.000 2.9 0.000 3.4 Education of head None ref Primary 0.103 7.2 0.126 6.0 0.060 3.1 Low secondary 0.228 11.1 0.218 8.3 0.160 4.6 Upper secondary general 0.415 14.6 0.357 10.4 0.468 7.6 Upper secondary professional 0.484 7.4 0.449 6.0 0.247 1.4 Post-secondary & University 0.700 20.5 0.608 15.1 0.704 7.5 Education of spouse None ref Primary 0.084 4.7 0.102 4.0 0.100 4.1 Low secondary 0.110 4.8 0.146 5.0 0.144 3.4 Upper secondary 0.095 2.2 0.177 3.5 -0.040 -0.3 Post-secondary & University 0.231 3.7 0.282 4.0 0.286 1.8 Labor market of head Non-participant ref Participant -0.003 -0.1 -0.023 -0.6 0.099 1.6 Institutional sector of head Public administration ref Public enterprise 0.017 0.3 0.015 0.2 -0.232 -1.3 Private enterprise -0.035 -1.0 -0.044 -1.1 -0.143 -1.8 Individual enterprise -0.055 -2.0 -0.067 -2.0 -0.097 -1.6 Type of industry of head Agriculture ref Industries 0.285 12.0 0.244 7.5 0.282 7.5 Construction 0.235 7.1 0.187 4.8 0.380 4.6 Commerce 0.333 16.7 0.325 12.2 0.269 7.4 Restaurant/Hotel 0.427 8.5 0.374 6.1 0.342 3.4 Transportation 0.299 8.1 0.263 6.0 0.404 4.3 Education/Health 0.278 7.8 0.191 4.5 0.668 8.3 Other services 0.288 12.5 0.244 8.3 0.290 6.0 89 Institutional sector of head Public administration Public and private enterprise 0.163 2.5 0.085 1.1 0.445 3.0 Individual enterprise 0.044 0.9 0.000 0.0 0.169 1.7 Unemployed 0.121 2.5 0.088 1.5 0.255 2.5 Type of industry of head Agriculture ref Industries, construction -0.022 -0.8 0.027 0.7 -0.069 -1.6 Commerce 0.064 3.3 0.084 3.0 0.127 3.8 Restaurant/Hotel 0.134 3.2 0.168 3.2 0.186 2.3 Education/Health 0.143 2.5 0.152 2.3 0.104 0.8 Other services -0.006 -0.2 -0.002 -0.1 0.123 1.6 Time to the grocery market Less than 15 minutes ref 15-29 mns -0.033 -2.5 -0.035 -1.7 -0.034 -2.1 30-44 mns -0.034 -2.3 -0.105 -4.0 0.002 0.1 45-59 mns -0.046 -2.3 -0.100 -2.4 -0.023 -1.1 60mns et + 0.004 0.3 -0.076 -1.7 0.029 1.6 Time to the nearest pharmacy Less than 15 minutes 15-29 mns -0.043 -3.2 -0.036 -1.8 -0.035 -1.9 30-44 mns -0.045 -2.8 -0.049 -1.8 -0.023 -1.2 45-59 mns -0.081 -4.0 -0.142 -3.4 -0.050 -2.2 60mns et + -0.023 -1.3 -0.056 -1.2 -0.005 -0.3 Time to the civil registration center Less than 15 minutes 15-29 mns -0.045 -3.2 -0.046 -2.0 -0.027 -1.6 30-44 mns -0.030 -1.8 0.005 0.2 -0.037 -2.1 45-59 mns 0.004 0.2 -0.022 -0.3 0.003 0.1 60mns et + 0.002 0.1 -0.053 -0.9 -0.003 -0.2 Time to the police station Less than 15 minutes 15-29 mns -0.037 -2.1 -0.032 -1.5 -0.036 -1.0 30-44 mns -0.051 -2.7 -0.015 -0.6 -0.087 -2.6 45-59 mns -0.057 -2.8 0.012 0.4 -0.121 -3.5 60mns et + -0.108 -5.9 -0.105 -3.2 -0.134 -4.4 Physical and social Capital Area land cultivated 0.004 2.6 -0.002 -0.4 0.006 3.6 Area land cultivated, squared 0.000 -3.2 0.000 -0.3 0.000 -3.8 Member of any association (yes) 0.076 7.7 0.086 4.8 0.063 5.6 Regions Nord ref Hts Bassins 0.148 6.9 0.157 4.2 0.111 4.2 Bcle Mouhoun 0.005 0.3 0.005 0.1 -0.013 -0.5 90 Sahel 0.368 15.9 0.305 7.4 0.382 14.2 Est 0.123 5.5 0.092 2.3 0.112 4.3 Sud Ouest 0.175 7.5 0.315 7.6 0.128 4.7 Centre Nord 0.280 12.6 0.276 6.9 0.254 10.0 Centre Ouest 0.096 4.3 0.102 2.6 0.082 3.2 Plateau central 0.102 4.5 0.021 0.5 0.134 5.1 Centre Est 0.244 11.1 0.170 4.5 0.291 11.2 Centre 0.277 12.3 0.291 8.2 0.183 6.0 Cascade 0.274 11.7 0.214 5.3 0.292 10.5 Centre sud 0.163 7.1 0.074 1.8 0.209 7.9 Residence Rural (yes) -0.034 -2.7 Constant 1.101 12.0 1.309 10.3 0.443 2.8 Statistics # observations 10411 4003 6408 2 R 0.572 0.630 0.402 Source: Authors calculations using INSD survey EMC-2014 91 Table A5. Probit model of FIES food insecurity Model 1: Shock in detail Model 2: Shock grouped National Urbain Rural National Urbain Rural Household demographics Household size 0.000 0.026 -0.007 -0.002 0.021 -0.008 Household size (squared) -0.000 -0.001 -0.000 -0.000 -0.001 -0.000 Dependency ratio 0.063 0.173 0.037 0.072 0.188 0.048 Head female (yes) 0.035 0.138 -0.033 0.050 0.165 -0.012 Single -0.157 -0.203 -0.203 -0.177 -0.225 -0.194 Polygamous -0.072 -0.496* -0.054 -0.089 -0.493* -0.049 Divorced -0.321 -0.619 -0.270 -0.352 -0.681* -0.267 Widower -0.051 -0.319 -0.055 -0.095 -0.368 -0.071 Human capital Household head age 0.005 0.030 0.001 0.004 0.030 0.000 Squared head age -0.000 -0.000 0.000 -0.000 -0.000 0.000 Primary -0.063 -0.177 -0.042 -0.069 -0.197 -0.040 Low secondary -0.098 -0.264* 0.097 -0.076 -0.244* 0.125 Upper secondary -0.345 -0.269 -0.874*** -0.360* -0.303 -0.855** University -1.023*** -0.815** -1.744*** -1.027*** -0.805** -1.761*** Household assets Electricity (yes) -0.296*** -0.305*** -0.127 -0.299*** -0.288*** -0.154 Toilets with flush (yes) -0.055 -0.212 0.319 -0.005 -0.172 0.363 Automobile (yes) -0.427 -1.037*** 0.020 -0.405 -1.025*** 0.028 Motocycle (yes) -0.486*** -0.610*** -0.462*** -0.487*** -0.615*** -0.467*** Bicycle (yes) -0.118** 0.137 -0.242*** -0.116** 0.138 -0.240*** Refrigerator(yes) -0.275* -0.218 -0.091 -0.277* -0.212 -0.084 Acces to infrastructure Transport < 30 min (yes) -0.085* -0.027 -0.097* -0.095* -0.027 -0.110** Shocks Shocks 0.292*** 0.435*** 0.242*** Natural 0.286*** 0.129 0.311*** Price -0.083 0.081 -0.121** Employment 0.209* 0.333* 0.036 Death and Illness 0.058 0.364*** -0.008 Security (crime, theft) 0.040 0.302* -0.018 Social Issue 0.246** 0.180 0.231* Other shock 0.283** 0.596** 0.207 Social capital Association (yes) -0.039 -0.150 -0.023 -0.039 -0.147 -0.023 Geographic variables Bcle Mouhoun -0.338** -0.546*** -0.321** -0.290** -0.586*** -0.254 Sahel 0.350*** -0.073 0.356** 0.407*** -0.150 0.441*** Est 0.517*** 0.621*** 0.517*** 0.559*** 0.605*** 0.576*** 92 Sud Ouest 0.167 -0.051 0.166 0.215 -0.065 0.234 Centre Nord 0.161 0.023 0.176 0.218 0.005 0.253 Centre Ouest -0.023 0.220 -0.035 -0.003 0.182 -0.000 Plateau central -0.333** 0.218 -0.351** -0.299** 0.152 -0.304* Nord -0.161 -0.489*** -0.130 -0.141 -0.602*** -0.084 Centre Est 0.226* 0.020 0.263* 0.234* -0.053 0.284* Centre 0.361*** 0.380*** 0.257 0.358*** 0.361** 0.277* Cascade -0.071 0.183 -0.144 -0.035 0.183 -0.081 Centre sud 0.237* 0.532*** 0.205 0.279* 0.477*** 0.268 Rural 0.064 0.000 0.000 0.083 0.000 0.000 Constant -0.260 -1.083** 0.071 -0.310 -1.095** 0.010 # Observations 9901 3718 6183 9901 3718 6183 Source. Author’s calculation using the 2014 EMC survey 93 Table A6. Regression on calories consumption Model 1: Shock in detail Model 2: Shock grouped National Urbain Rural National Urbain Rural Household demographics Household size -0.100*** -0.261*** -0.070*** -0.099*** -0.258*** -0.069*** Household size (squared) 0.002*** 0.006*** 0.001*** 0.002*** 0.006*** 0.001*** Dependency ratio -0.526*** -0.888*** -0.366*** -0.519*** -0.892*** -0.359*** Head female (yes) 0.054 -0.106 0.124* 0.046 -0.099 0.112* Single -0.931*** -1.089*** -0.071 -0.934*** -1.078*** -0.089 Polygamous -0.780*** -0.625* -0.018 -0.781*** -0.603* -0.034 Divorced -0.560** -0.680* 0.166 -0.553** -0.672* 0.167 Widower -1.044*** -1.063*** -0.201 -1.036*** -1.047*** -0.203 Human capital Household head age -0.027*** -0.073* -0.019*** -0.028*** -0.073* -0.019*** Squared head age 0.000*** 0.001* 0.000*** 0.000*** 0.001* 0.000*** Primary -0.020 -0.066 -0.012 -0.021 -0.057 -0.012 Low secondary 0.262*** 0.167* 0.188** 0.269*** 0.177* 0.189** Upper secondary 0.342*** 0.160 0.499*** 0.322*** 0.136 0.485*** University 0.959*** 0.654*** 0.312 0.940*** 0.637*** 0.296 Household assets Electricity (yes) 0.254*** 0.293*** 0.177*** 0.259*** 0.279*** 0.189*** Toilets with flush (yes) 0.201* 0.320** -0.079 0.199* 0.317** -0.067 Automobile (yes) 0.217** 0.304** 0.360** 0.213** 0.291** 0.369** Motocycle (yes) 0.171*** 0.256*** 0.157*** 0.168*** 0.248*** 0.157*** Bicycle (yes) -0.210*** -0.163* -0.127*** -0.216*** -0.155 -0.136*** Refrigerator(yes) 0.319*** 0.200* 1.064*** 0.323*** 0.209** 1.038*** Acces to infrastructure Transport < 30 min (yes) -0.040* 0.026 -0.050* -0.052** 0.014 -0.060** Shocks Shocks -0.133*** -0.127 -0.100*** Natural -0.010 0.179 -0.025 Price -0.207*** -0.215* -0.196*** Employment 0.012 0.014 0.116* Death and Illness 0.024 0.105 0.031 Security (crime, theft) 0.034 -0.002 0.041 Social Issue -0.153** -0.193* -0.122 Other shock -0.048 -0.089 -0.042 Social capital Association (yes) 0.041** 0.062 0.024 0.045** 0.072 0.025 Geographic variables Bcle Mouhoun -0.136** -0.147 -0.150** -0.090 -0.100 -0.103 Sahel 0.833*** 0.366*** 0.855*** 0.856*** 0.419*** 0.878*** Est 0.123* -0.164 0.130 0.130* -0.119 0.138 Sud Ouest 0.123 0.418** 0.109 0.157 0.465*** 0.148 94 Centre Nord 0.257*** 0.310* 0.237*** 0.312*** 0.384*** 0.292*** Centre Ouest 0.016 -0.200 0.017 0.031 -0.149 0.033 Plateau central 0.020 -0.090 0.021 0.056 -0.034 0.060 Nord -0.038 -0.205 -0.041 -0.022 -0.159 -0.026 Centre Est 0.296*** -0.031 0.343*** 0.310*** 0.026 0.359*** Centre 0.322*** 0.252** 0.224* 0.328*** 0.231** 0.251** Cascade 0.241*** 0.089 0.296*** 0.278*** 0.114 0.337*** Centre sud 0.093 -0.053 0.109 0.122 0.009 0.142 Rural 0.016 0.000 0.000 0.014 0.000 0.000 Constant 3.906*** 6.170*** 2.514*** 3.927*** 6.204*** 2.528*** 2 0.336 0.395 0.275 0.331 0.392 0.264 R # Observations 9901 3718 6183 9901 3718 6183 Source. Author’s calculation using the 2014 EMC survey 95 Table A7. Regression on the dynamic food insecurity National Urban Rural Chronically Transcient Chronically Transcient Chronically Transcient food food food food food food insecure insecure insecure insecure insecure insecure Household demographics Household size 0.325*** 0.211*** 0.647*** 0.319*** 0.282*** 0.186*** Household size (squared) -0.005** -0.003*** -0.013*** -0.006*** -0.004** -0.003*** Dependency ratio 0.438 0.613*** 0.644 0.233 0.484 0.781*** Head female (yes) -0.226 -0.142 0.188 -0.006 -0.403 -0.254 Single -0.070 -0.236 0.317 -0.126 -0.534 -0.604* Polygamous -0.207 -0.394 -0.357 -0.525 -0.564 -0.693** Divorced -1.254 -0.626 -0.387 -0.065 -2.007** -1.269** Widower 0.286 -0.135 0.547 -0.081 -0.169 -0.460 Human capital Household head age 0.040** 0.045*** 0.004 0.058* 0.045** 0.043*** Squared head age -0.000 -0.000** 0.000 -0.000 -0.000 -0.000** Primary -0.088 -0.151 -0.111 0.045 -0.152 -0.259* Low secondary -0.299 -0.322* -0.225 -0.330 -0.332 -0.150 Upper secondary -0.326 -0.666** -0.697 -0.485 -0.042 -1.002** University -1.130 -1.011** -22.129*** -0.895* -0.032 -1.920** Household assets Electricity (yes) -1.290*** -0.692*** -1.362*** -0.800*** -0.950* -0.467* Toilets with flush (yes) -0.700 -0.244 -2.449*** -0.655* 0.100 0.235 Automobile (yes) -1.531** -0.643* -1.914** 0.141 -1.969** -1.902*** Motocycle (yes) -0.979*** -0.484*** -1.392*** -0.686*** -0.929*** -0.437*** Bicycle (yes) 0.107 0.109 0.153 0.223 -0.026 0.014 Refrigerator(yes) -0.036 0.158 0.410 0.165 -22.201*** -0.819 Acces to infrastructure Transport < 30 min (yes) 0.223* -0.006 -0.046 -0.009 0.259* -0.006 Shocks Shocks 0.452*** 0.235*** 1.071*** 0.354** 0.274* 0.156 Social capital Association (yes) -0.141 -0.002 0.129 0.153 -0.182 -0.034 Geographic variables Bcle Mouhoun 0.950*** 1.000*** -0.272 0.237 1.114*** 1.065*** Sahel -4.218*** -1.562*** -2.583*** -0.157 -4.274*** -1.687*** Est -0.431 0.322 0.724* 0.859*** -0.491 0.194 Sud Ouest -0.356 -0.690** -0.957 -0.804** -0.331 -0.745** Centre Nord -1.842*** -1.019*** -2.702*** -1.066*** -1.776*** -1.070*** Centre Ouest -0.119 0.385* -0.134 0.372 -0.082 0.344 Plateau central -0.664* -0.507** -0.308 0.456 -0.686* -0.631** Nord 0.559* 0.427* 0.679 1.025*** 0.537 0.278 Centre Est -0.869*** -0.536** -0.264 -0.418 -0.998** -0.631** 96 Centre -0.713** -0.681*** -0.696* -0.627** -0.570 -0.682** Cascade -1.680*** -0.667*** -2.130*** -0.290 -1.679*** -0.826*** Centre sud -0.376 -0.338 0.005 0.555 -0.425 -0.504 Rural 0.144 0.184 0.000 0.000 0.000 0.000 Constant -3.915*** -2.218*** -5.502*** -3.481*** -2.928*** -1.279*** # Observations 9901 3718 6183 log pseudo likelihood -14770424 -2574393.7 -12016265 Source. Author’s calculation using the 2014 EMC survey Table A8. Regression on the agricultural productivity All No cotton Cotton Productivity (output value/hectare) Coef. P>t Coef. P>t Coef. P>t Household demographics and education Female 0.14 0.00 0.12 0.00 0.11 0.54 Dependence ratio 0.06 0.02 0.06 0.03 0.11 0.18 Square dependence ratio -0.01 0.08 -0.01 0.10 -0.01 0.47 household head age 0.01 0.02 0.01 0.01 0.01 0.63 Square household head age 0.00 0.02 0.00 0.02 0.00 0.55 Handicap (yes) -0.05 0.32 -0.07 0.16 -0.02 0.92 Primary school years 0.00 0.49 -0.01 0.24 0.00 0.74 Square primary school years 0.00 0.11 0.00 0.04 0.00 0.58 Secondary school years 0.00 0.79 -0.01 0.40 0.11 0.00 Square secondary school years 0.00 0.51 0.00 0.16 -0.01 0.01 Capital Land surface -0.03 0.00 -0.03 0.00 -0.03 0.00 Equipment 0.07 0.00 0.07 0.00 0.07 0.00 Labor use Family workers (men) 0.07 0.00 0.08 0.00 0.09 0.05 Family workers (women) 0.04 0.00 0.05 0.00 -0.06 0.15 Family workers (kids) 0.01 0.05 0.01 0.19 0.06 0.01 Non family workers 0.06 0.00 0.06 0.00 0.06 0.01 Inputs use Organic fertilizer use 0.00 0.84 -0.01 0.55 0.02 0.78 Chemical fertilizer use -0.03 0.64 -0.26 0.00 0.34 0.25 Pesticides use -0.12 0.15 -0.39 0.00 0.09 0.66 Seeds 0.20 0.00 0.18 0.00 -0.12 0.79 Organic fertilizer qty 0.00 0.56 -0.01 0.39 0.00 0.98 Chemical fertilizer qty 0.02 0.00 0.05 0.00 0.00 0.97 Pesticides qty 0.02 0.03 0.06 0.00 -0.02 0.45 Seeds qty -0.01 0.01 -0.01 0.02 0.00 0.75 Income (non farm and other) Non agricultural wage -0.06 0.06 -0.03 0.33 -0.25 0.04 Non farm income -0.01 0.78 0.00 0.98 -0.01 0.86 Transfers -0.04 0.11 -0.03 0.29 -0.08 0.35 Other income 0.00 0.99 -0.07 0.36 0.42 0.04 Products 97 Cereals -0.11 0.00 -0.11 0.01 -0.09 0.32 Corn 0.04 0.12 0.07 0.01 -0.13 0.18 Rice 0.16 0.00 0.13 0.00 0.27 0.00 Cowpeas 0.04 0.14 0.04 0.15 -0.01 0.93 Peanut 0.00 0.99 0.00 0.84 0.02 0.77 Sesame 0.18 0.00 0.24 0.00 -0.02 0.76 Cotton 0.69 0.00 Tubers 0.19 0.05 0.26 0.01 -0.18 0.60 Fruits and vegetables 0.02 0.66 0.03 0.35 -0.08 0.54 Economic environment Distance to market -0.03 0.31 0.01 0.81 -0.10 0.22 Distance to transportation service 0.04 0.18 0.01 0.74 0.10 0.24 Distance to the closest road 0.00 0.99 0.00 0.95 -0.01 0.89 Household Own account 0.11 0.00 0.06 0.10 0.30 0.00 Shocks Shocks -0.09 0.00 -0.06 0.02 -0.22 0.00 Region Boucle du Mouhoun -0.42 0.00 -0.22 0.00 -0.61 0.00 Sahel 0.08 0.23 0.29 0.00 Est -0.74 0.00 -0.55 0.00 -0.98 0.00 Sud Ouest -0.49 0.00 -0.35 0.00 -0.54 0.00 Centre Nord -0.23 0.00 -0.03 0.69 Centre Ouest -0.48 0.00 -0.24 0.00 -0.75 0.00 Plateau central -0.38 0.00 -0.16 0.02 -0.83 0.00 Nord -0.40 0.00 -0.18 0.01 Centre Est -0.38 0.00 -0.16 0.01 -0.87 0.00 Centre -0.68 0.00 -0.45 0.00 Cascade -0.34 0.00 -0.29 0.00 -0.25 0.06 Centre sud -0.71 0.00 -0.43 0.00 -1.41 0.00 Constant 10.01 0.00 9.75 0.00 11.44 0.00 R2 0.3403 0.287 0.2716 # Observations 5671 4786 885 Source. Author’s calculation using the 2014 EMC survey Table A9. Regression on the the transfers Probability (transfers=1) Coef Household demographics Household size -0.025*** Household size (squared) 0.001** Gender (female) 0.619*** Household head age -0.013* Squared Household head age 0.000*** Monogamist -0.245** Polygamist -0.200* Divorced -0.314 Widower -0.341** Household head education 98 Primary 0.195*** Low secondary 0.104 Upper secondary -0.089 University 0.290** Handicap Handicapped (yes) 0.218*** Shocks Natural -0.068 Price -0.027 Employment 0.329*** Death and Illness 0.135*** Security (crime, theft) -0.001 Social Issue 0.189 Other -0.093 Income Income before transfers (log) -0.039** Region Bcle Mouhoun 0.247** Sahel -0.069 Est -0.017 Sud Ouest -0.347*** Centre Nord 0.333*** Centre Ouest 0.218** Plateau central 0.056 Nord 0.132 Centre Est 0.01 Centre 0.05 Cascade 0.465*** Centre sud -0.142 Residence area Rural 0.033 Constant -0.09 # Observations 9423 Source. Author’s calculation using the 2014 EMC survey 99