EXECUTIVE SUMMARY Linking growth and poverty is a crucial element for evaluating the effectiveness of government policies under the PRSP process. Burkina Faso has benefited from more than 3 percent growth in per-capita incomes since the devaluation in 1994. The steady increase in incomes, albeit from a very low level, should over time have lifted some Burkinabk above the poverty line and led to a reduction in poverty rates. To gauge how the rising incomes have affected the population, the government has conducted household surveys from time to time (1994, 1998, and 2003) to collect information of household consumption and incomes. This study uses the household data from 1998 and 2003 data to (i) consider the measurement of poverty over time; (ii) study the links between growth and poverty in 1998-2003 and under possible future growth paths; (iii) examine the relationship between poverty and social services; and (iv) illustrate equity considerations in the execution of fiscal policy choices. In doing so, it builds on the existing analysis by the government, donors, and independent researchers. Measuring Poverty Measuring changes in poverty is not an exact science and subject to data quality issues. Over time, the National Institute of Statistics and Demography (INSD) has improved the data collection process, but has also made changes to survey designs that have reduced the comparability of different surveys. Care needs to be taken to construct a comparable measure for household welfare to best assess changes in poverty. A comparable welfare indicator can be constructed for 1998 and 2003 since the effects of changes in survey design can be clearly identified and largely eliminated. The resulting consumption measure for households contains equally measured items from the 1998 and 2003 surveys by excluding a limited number of items measured differently in both surveys. A consistent set of monthly price data is applied to household consumption items to eliminate the effect of prices changes over time and space. However, a similar methodology cannot be applied to 1994 survey data because of changes in the recall period in the survey design, the effect of which cannot be eliminated. Using a comparable poverty measure, we find that the poverty headcount declined by about 8 percentage points between 1998 and 2003. Adopting the previously announced 46.4 percent headcount for 2003 and the related poverty line as baseline, the comparable poverty headcount in 1998 was 54.6 percent. The poverty decline was stronger in rural than in urban areas, and inequality remained largely unchanged on the national level between 1998 and 2003. The conclusion that poverty declined between 1998 and 2003 is robust to changes in the poverty line. Using a household income measure, rather than consumption also allows drawing the conclusion that poverty declined during 1994-98. Growth, Poverty, and Inequality Growth during 1998-2003 was driven by a large expansion of the primary sector, following the 1997-98 drought. Increases in volume and value of cereal production exceeded population growth, and by 2001 the volume and value cotton production bypassed levels attained viiiin 1997 after recovering from the crisis caused by the “white fly” infestation in 1998. Growth in the primary sector was the result of increased surfaces, with little change in productivity. The secondary sector expanded at about the same rate as overall GDP, with the informal sector playing a large role in the expansion of manufacturing (food products, textiles, minerals, and metal works), indicating that growth estimates may be fairly imprecise. In the tertiary sector growth was sluggish owing to government restraint and sluggish growth in trade. Again, informal activities were the major dynamic force. National account data and household data for 1998-2003 line up reasonably well. The corrected household consumption data implies annual average per capita expenditure growth of about 2.3 percent per year. From national account data, we arrive at changes in real per capita private consumption of about 3 percent per year on average. Given the variety of errors possible in the way these data have been established, the match is reasonably close. As regards correlates of poverty, results are similar for the 2003 survey as those found in previous studies. Larger household size, lower education levels, occupation in agriculture, and remoteness tend to be correlated with lower per-capita consumption levels. By contrast, families of landowners and those receiving transfers from relatives tend to have higher expenditure per capita. The decline in national poverty rates between 1998 and 2003 is largely a result of the growth in agricultural output, both in subsistence farming and cotton farming. The respective socio-economic groups producing agricultural non-tradables (60 percent of the population in 2003) and agricultural tradables (1 8 percent of the population in 2003) registered a sharp decline in poverty rates of 7.5 and 6 percentage points, respectively. However, inequality increased in rural areas, reducing some of the potential growth impact on poverty rates. Between 1998 and 2003, the share of subsistence farmers in the population has declined, whereas the share of cotton farmers has been on the rise after the confidence crisis related to the “white fly” infestation had been overcome. As a result of this shift of the rural population, there were less subsistence farmers among the poor in 2003 than in 1998 but more cotton farmers. The urban sector also contributed to declining poverty by absorbing an additional share of the population without experiencing rising poverty rates. Overall, urban poverty rates declined little. However, this fact masks that the urban population was on the rise and thus urban areas were able to absorb rural population without a dramatic increase in poverty, in part through a decline in urban inequality. In particular, the share of socio-economic groups in the private non- tradable sector and the informal sector and the unemployed has been rising considerably. The first two groups benefited sufficiently from the growth and declining inequality in the secondary and tertiary
View Full Document
Unlocking...