WPj 3b6*T POLICY RESEARCH WORKING PAPER 3 068 Poverty and Economic Growth in Egypt, 1995-2000 Heba El-laithy Michael Lokshin Arup Banerji The World Bank Development Research Group k Poverty Team June 2003 I POLICY RESEARCH WORKING PAPER 3068 Abstract After a decade of slow economic growth Egypt's rate of age composition of poor households. The results show growth recovered in the late 1990s, averaging more than that average household expenditures rose in the second five percent a year. But the effect of this growth on half of the 1990s and the poverty rate fell from 20 poverty patterns has not been systematically examined percent to less than 17 percent. But, in addition to the using consistent, comparable household datasets. In this ongoing divide in the urban-rural standard of living, a paper El-Laithy, Lokshin, and Banerji use the rich set of new geographicaUlregional divide emerged in the late unit-level data from the most recent Egyptian household 1990s. Poverty was found predominantly among less- surveys (1995-96 and 1999-2000) to assess changes in educated individuals, particularly those working in poverty and inequality between 1995 and 2000. Their agriculture and construction, and among seasonal and analysis is based on household-specific poverty lines that occasional workers. These groups could suffer the most account for the differences in regional prices, as well as from the slowing economic growth evident after 1999- differences in the consumption preferences and size and 2000. This paper-a product of the Poverty Team, Development Research Group-is part of a larger effort in the group to understand the impact of economic growth on poverty. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Patricia Sader, room MC3-632, telephone 202-473-3902, fax 202- 522-1153, email address psader@worldbank.org. Policy Research Working Papers are also posted on the Web at http:// econ.worldbank.org. The authors may be contacted at hflaithy@gega.net, mlokshin@worldbank.org, or abanerji@worldbank.org. June 2003. (28 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by Partnerships, Capacity Building, and Outreach POVERTY AND ECONOMIC GROWTH IN EGYPT, 1995-2000 Heba El-laithy Faculty of Economics, University of Cairo, Cairo, Egypt hflaithvyigega.net Michael Lokshin* Development Research Group, The World Bank, 1818 H Street NW, Washington, D.C., 20433, USA mlokshin(l)worldbank.org Arup Banerji Middle East and North Africa Region, The World Bank, 1818 H Street NW, Washington, D.C., 20433, USA abanerj i2worldbank.org * Corresponding author. The findings, interpretations, and conclusions of this paper are those of the authors and should not be attributed to the World Bank, its Executive Directors, or the countries they represent. We thank Martin Ravallion for extremely helpful suggestions and comments on this paper. 1. Introduction The late 1990s were, in economic terms, a watershed period for Egypt. After a decade of slow economic growth, averaging less than 4 percent per year between 1987 and 1995, overall growth rebounded to reach an average of 5.6 percent per year between 1996 and 2000. What was the effect of this growth on the economic dimensions of poverty (that is, poverty in consumption and income terms) in this nation of 65 million people? This paper answers that question, using a comprehensive set of household-level data from 1995-1996 and 1999/2000. The economic background for the growth and poverty story in Egypt during the late 1990s involves both a liberalizing and an inward-looking economy. Following a significant liberalization of Egypt's economy beginning in 1991, Egypt experienced a growth spurt. During this period, Egypt began a slow process of controlling inflation, dismantling the historically overwhelming state domination of the economy, and opening markets to greater competition. As a result, the private sector emerged as a more important economic player in terms of output and jobs. At the same time, the economy remained inward-looking, as high tariff rates and an appreciating currency made the domestic market more attractive for Egyptian businesses. Consequently, the growth spurt took place because of the service and industry sectors, and a strong demand for domestic products - especially tourism, construction, and import-substituting manufactured goods. Our paper fills a critical gap in the literature on poverty in Egypt for three reasons. By making use of a rich and yet rarely available dataset on household expenditures in Egypt for two recent time periods - 1995/96 and 1999/2000 -- we offer the first analysis of the evolution of poverty in Egypt in late 1990s. Second, we use the opportunity provided by the large datasets to construct 'household-specific' poverty lines that effectively address methodological problems that have cropped up in previous papers in relation to economies of scale and regional differences. In doing so, we provide an example of a methodology that can be useful for similar work in other countries. Third, 3 our findings, especially on the distinct change in the spatial distribution of poverty over the course of 1990s, have critical implications for policymakers and advisors. Limited formal and consistent empirical evidence exists on the levels and changes in poverty and inequality in Egypt during the 1990s. To a large extent, this stems from the fact that, up to now, no nationally representative unit-level data was available for public use in Egypt. The first study to introduce a consistent. approach to poverty measurement in Egypt in 1995/96 was El-Laithy and Osman (1997), but follow-up studies did not use the same methodology to look at the evolution of poverty after 1996. Datt, Jolliffe and Sharma (2001) constructed a poverty profile for 1997 using data from a specially commissioned integrated household survey, but the sample size was relatively small (2,500 households) and the survey, being a one-off exercise, did not permit analysis of the evolution of poverty across time. The only recent paper that has used the 1999/2000 household data to examine poverty is El-Ehwany and El-Laithy (2001), but the paper was a preliminary effort, and used non-weighted data. While these studies give useful information on poverty in Egypt, the poverty lines they use take into account neither regional and urban-rural differences in prices (with the exception of Datt, Jolliffe and Sharma 2001), nor the differences in costs of living due to economies of scale and demographic composition of the households. As a result, these papers could well overestimate poverty in large households and in households with children. At the same time, poverty rates in urban areas of Egypt could be underestimated, because urban households are on average smaller than rural households, and prices of most nonfood commodities and services and some food items are higher in the four metropolitan cities of Egypt. We use a methodology that takes these problems into account. Although we follow the standard cost-of-basic-needs method to construct poverty lines, we introduce a new approach to estimate region-specific household poverty lines'. The unit-level data in the most recent Households Income, Expenditure and Consumption Survey (HIECS) allows us to construct poverty lines for each household depending on its location, size, age, and gender composition. Economies of scale are also taken into account. 'This new approach was developed in a course of multiple discussions with Martin Ravallion. 4 The results presented in this paper are critical for understanding Egypt's policy options over the coming decade. While poverty declined for Egypt as a whole between 1995/96 and 1999/2000, there were significant differences in poverty alleviation and household expenditure increases across different regions and different sectors of employment. With a labor force growing at about 2.7 percent annually, over half a million job-seekers are entering the Egyptian labor market every year. The economic boom during the late 1990s clearly helped many in the labor force to improve their economic welfare. However, with a rapid decline in the construction sector since 1999, and a volatile tourism sector buffeted by the effects of regional security crises, the first decade of the 21 st century will present policymakers with a challenge in terns of income expansion and poverty reduction. Policymakers also need better information on the dynamics and causes of poverty because the government has had fewer resources available for social assistance program in recent years. A better understanding of changes in poverty and inequality in Egypt could facilitate both a more effective design of the social policies and greater efficiency in social assistance programs. This paper is organized as follows. The next section discusses the data and methodology issues. Section 3 presents estimates of poverty and inequality in Egypt in 1999/2000. Section 4 describes the changes in poverty and inequality in Egypt between 1995 and 2000, and Section 5 paints a profile of the poor in Egypt. Section 6 presents the results of a multivariate poverty profile analysis and poverty simulations. Section 7 concludes. 2. Data and definitions Our poverty analysis is based on the 1995/96 and 1999/2000 rounds of the HIECS conducted by the Central Agency for Public Mobilization and Statistics (CAPMAS), the official statistical agency of Egypt. The HIECS rounds were administered over 12 months, with 10 visits to each household over a period of one month. The 1995/96 survey's data were collected from October 1995 to September 1996. The 1995/96 HIECS includes 14,805 households, of 5 which 6,622 reside in urban and 8183 in rural areas. The sample frame of the 1995/96 survey is based on an. updated frame of the 1986 Population Census of 503 area sampling units that include 276 units in urban and 227 in rural areas. Data for the most recent survey were gathered from October 1999 to September 2000. The 1999/2000 sample comprises 47,949 households, of which 28,754 reside in urban and 19,195 in rural areas. This is the largest survey of its kind conducted in Egypt.2 The 1999/2000 HIECS was based on the 1996 Population Censuses sample frame of 600 area sampling units distributed between urban and rural areas (360 and 240 units respectively). The surveys of 1995/96 and 1999/2000 are highly comparable in terms of sampling procedure and data collection methodology. The samples of both surveys are stratified multistage random samples. The sample designs of both surveys are nationally representative and the size of these surveys is large enough to allow for inferences at the regional and governorate levels, with the exception of Border govemorates where the sample sizes are small.3 Definitions of Regions in Egypt The regional aggregates used in this paper are for Metropolitan, Lower Egypt, Upper Egypt and Border areas. Ninety-five percent of the population of Egypt lives in the first three regions. The Metropolitan governorates essentially comprise the four major cities of Cairo, Alexandria, Port-Said and Suez, all in northern Egypt. Lower Egypt (essentially the region of the Nile delta) is also in the northern part of Egypt, and Upper Egypt, perhaps counter-intuitively, is the area mostly south of Cairo, with governorates largely following the meandering upper reaches of the Nile. The Border areas are the less populated desert areas bordering the Red Sea, the Sinai, and the vast Marsa Matruh and El Wadi El Gadid areas west of the Nile.4 In 1999/2000, the Metropolitan cities had about 18 percent of Egypt's total population, Lower Egypt had 43 percent, mostly in urban governorates; and Upper Egypt, 2 Starting from 1957 the household budget surveys were conducted in Egypt in 1964/65, 1974/75, 1981/82 and in 1990/91 (CAPMAS VARIOUS ISSUES). 3 For more information on the sample issues see, e.g., CAPMAS, Household, Income, Expenditure and Consumption Survey Report, Volume one, 2001. 4 Also see Table 2 for the specific governorates in each region. 6 had 38 percent, with more than two-thirds (27 percent of the Egyptian population) residing in rural areas. Welfare Indicator We measure welfare in terms of an aggregate money metric indicator, defined as the amount of money required, given a set of prices, to attain a particular level of utility. This indicator allows us to rank households' welfare levels by comparing household observable consumption. The consumption-based approach is particularly suited for measuring poverty in developing countries, since it relies for poverty comparisons on the notion of deprivation from certain commodities and resources (both food and nonfood) deemed essential for attaining a minimum level of well-being within a given society (Hentschel and Lanjouw 2000). Consumption is the preferred indicator of well-being also because it incorporates the life-cycle hypothesis. Households might use savings and credit to smooth fluctuations in income, and in that case, consumption would better reflect their actual welfare. Finally, respondents in developing countries might be reluctant to reveal information about their income, so consumption is usually measured more precisely than income. The information on the components of total consumption we use in this paper draws upon responses from several sections of the survey. In the diary the respondents report daily food expenditures for a period of one month. Information about their. expenditures on nonfood items is collected in the main questionnaire for the previous. three months, or the previous year, depending on the type of commodity. All data are converted to an annual basis. Total household consumption is then measured as the sum of food consumption (home produced and purchased on the market), total nonfood, expenses, and actual or imputed rental values of housing. In poverty comparisons, the value of household consumption for 1999/2000 was deflated to 1996 Egyptian pounds (LE). Poverty Lines Our analysis foilows the cost-of-basic-needs (CBN) approach in order to construct household- and region-specific poverty lines. According to CBN methodology, poverty 7 lines represent the level of per capita expenditures at which the members of a household can be expected to meet their basic needs. The household-specific poverty line is calculated as a sum of the food and nonfood poverty lines. An initial step in defining the food poverty line is to construct a minimum food basket, that can be anchored to normative nutritional requirements. We estimate minimum caloric requirements for 52 different types of individuals, by urban and rural, gender and age categories, using World Health Organization standards (WHO 1985). The total household caloric requirement is the sum of minimum caloric requirements of all household members. Next, based on reported prices for each region in Egypt, we evaluate the cost of obtaining a hundred calories for the households with expenditures close to the previously estimated (El-Ehwany and El-Laithy 2001) poverty lines, i.e., from the second quintile of expenditure distribution. The cost of a household-specific food poverty line is calculated as the product of a household's caloric requirements and the cost to that household of a hundred calories. The share of nonfood expenditure is estimated by fitting Engel's curves of the food share onto total expenditure controlling for the household's. demographic composition. The total poverty line is then calculated by dividing the cost of the food poverty line by the estimated share of nonfood expenditure. We use two alternative poverty lines. The upper poverty line allows a nonfood expenditure share typical of those individuals whose expenditure on food is equivalent to the food poverty line. The lower poverty line restricts a nonfood expenditure to the share typical of those individual whose total expenditure is equivalent to the food poverty line (Ravallion et al. 1994). Unless specified, the results presented in the paper are based on this lower poverty line. To obtain the poverty lines for 1995/96 we deflate the food and nonfood components of the 1999/2000 poverty lines using food and nonfood CPIs. The food CPIs are calculated for each region based on information from the HIECS on food expenditure and quantity purchased. The nonfood components of the poverty lines are deflated with the official regionally disaggregated nonfood CPI. Seven region-specific cost-of-living indices for 1995-96 are derived as a weighted average of the food and nonfood CPI. We use these composite price indices to update the 1995/96 poverty lines. 8 Thus, constructed poverty lines ensure that differences in relative prices, food and nonfood consumption preferences, activity levels, and size and age composition of poor households are accounted for. 3. Poverty and inequality in Egypt 1999/2000 In 1999/2000, the poverty rate in Egypt stood at 16.7 percent, based on the lower poverty line (Table 1). Approximately 10.7 million Egyptians were poor according to this estimates. The poverty gap index was 2.97 percent, implying a per capita poverty deficit of LE 248 and an' average deficit of LE 1,482 for the poor. The poverty severity index of 0.8 was low by the standards of middle-income countries. Applying the upper poverty line, the poverty rate in Egypt rises to 42 percent, bringing the number of poor to almost 27 million. The distribution of the poor is quite uneven across the regions. In rural areas poverty rates reach 22.1 percent. Most of the poor are concentrated in Upper Egypt. Thirty-four percent, or 5.8 million individuals, are poor in the Upper Rural region. In the Upper Urban region the poverty rate is lower, but still reaches 19.3 percent, or 1.4 million people. The Upper Rural regiori contributes the most to national poverty. Almost 54.4 percent of all poor in Egypt live in the Upper Rural region, even though its share of the total population is only 26.7 percent. Moreover, the Upper Rural region's share in overall poverty increases with distribution-sensitive measures, reflecting the low standards of living of the poor in this region. The ranking of the regions remains unchanged for the poverty gap and severity of poverty indices, suggesting not only that poor households represent a large proportion of Upper Egypt's population, but also that their expenditures are far below the regional poverty line. The rate of poverty, particularly extreme poverty, is relatively low (9.2 percent) in urban areas where 41.5 percent of the population resides. The lowest incidence of poverty is in the Metropolitan region where only 5.1 percent of households are poor. These households constitute 3.9 percent of all poor in Egypt. The incidence, depth and severity of poverty vary considerably within each region. Table 2 shows poverty measures for various govemorates in urban and rural areas. The poverty indices in the urban areas of all governorates in Upper Egypt exceed 9 the corresponding national averages, except for Menia governorate. The poverty rates in rural Upper Egypt governorates surpass the national average, except for Giza and Aswan. Poverty incidence is highest in the govemorate of Assuit followed by Sohag and Beni Suef. The same pattern holds for the poverty gap and severity indices. In Lower Egypt, Menufiya is the only governorate in which poverty measures exceed the national average. In the Metropolitan region, Alexandria has the highest poverty rate. The incidence of poverty in Cairo amounts to 5.01 percent, ranking fifth among the urban govemorates in Egypt. The national level of expenditure inequality, measured by the Gini coefficient, reached 0.378 in 1999/2000 (Table 3). Urban areas of Egypt had higher levels of inequality in comparison with rural areas. The Upper Urban and Metropolitan regions had the highest inequality levels (Gini 0.406 and 0.386, respectively). The rural areas of Lower and Upper Egypt experienced relatively low inequality: 0.248 and 0.273, respectively. In some areas economic growth was positively correlated with changes in inequality - mean per capita expenditure in Metropolitan and Border Urban areas increased in tandem with inequality. Per capita expenditures in Border Rural areas declined as inequality declined. In Lower and Upper Egypt, however, changes in inequality and mean per capita expenditures go in opposite directions. 4. Poverty Trends: 1995/96 to 1999/2000 Following a rapid economic growth and an increase in per capita expenditure, poverty rates in Egypt decreased between 1995/96 and 1999/2000. The average annual per capita expenditure in 1999/2000 (evaluated at 1995 Metropolitan prices) was LE 1,599, compared with LE 1,407 in 1995/96 - an increase of 3.24 percent per year.5 The incidence of poverty, using the lower poverty line, decreased from 19.4 percent in 1995/96 to 16.7 percent in 1999/2000, or by about 800,000 people. The headcount, depth and severity of poverty calculated for both lower and upper poverty lines also showed statistically significant declines. The decreases in the poverty gap and severity indices indicate that expenditure distribution improved for the poor. The overall inequality in per s I LE = 0.30 USD in' 1995 prices 10 capita expenditure, measured by Gini coefficient (Table 3), increased from 0.345 in 1995/96 to 0.378 in 1999/2000. The growth of household expenditures, however, was not homogeneous across the regions (Table 4). While average per capita expenditures declined by 0.3 percent in Upper Rural Egypt and by 1.3 percent in Upper Urban Egypt between 1995 and 2000, it increased by 8 percent per year in the Metropolitan region. The annual growth rate of average household expenditures reached 2.9 percent for urban and 1 percent for rural areas in Lower Egypt. To answer the question of how the gains from aggregate economic growth were distributed across households in relation to their initial expenditure we calculate the growth-incidence curves (Ravallion and Chen, 2002). The growth-incidence curves indicate growth rates by household expenditure quintiles. Integrating this curve up to the poverty headcount index gives a measure of the rate of "pro-poor growth" consistent with the Watts index for the level of poverty (Watts 1968). Figure 1 shows the growth-incidence curves for five regions and Egypt as a whole. At the national level, per capita expenditure of the lower percentiles of expenditure distribution grew at a higher than average annual rate, suggesting that poor benefited more than the non-poor from economic growth. The aggregate picture, however, masks the considerable regional differences. Per capita expenditure of the lower percentiles of the expenditure distribution in Metropolitan and Upper Urban regions grew slower than their respective averages. In Lower Egypt and Upper Rural Egypt, growth of per capita expenditure of the poorest percentiles exceeded the regional average. Growth in Lower Egypt and Upper rural regions mainly benefited the poor. As a result of this unequal growth among different income groups, the inequality in expenditure distribution differed across regions. The Metropolitan region and Upper Egypt experienced an increase in inequality (for instance, the Gini coefficient increased from .374 to .396 in the Metropolitan region). At the same time, the Gini coefficients declined, by about three percentage points, for both urban and rural areas of Lower Egypt (Table 3). Using the Theil index (Theil 1979) to decompose the inequality at the national level into inequality between and within regions shows that 82 percent of the inequality in 1999/2000 can be explained by within-region variation, while 18 percent can be explained by between-region variation. The corresponding figures for 1995/96 are 87 percent and 13 percent. Comparisons over time suggest that inequality between regions is growing - the share of overall inequality explained by differences in regional means has increased by more than four percentage points over the period. The interactions between changes in the mean expenditure and inequality could explain the observed poverty dynamics across regions (Table 1). Poverty rates decreased substantially in the Metropolitan region (from 13.1 percent to 5.1 percent) and in Lower Rural Egypt (from 21.5 percent to 11.8 percent). In Upper Egypt, the poverty headcount rose from 29.3 percent to 34.2 percent in rural areas and from 10.8 percent to 19.3 percent in urban areas. These differences in poverty across regions are statistically significant, and the ranking of regions remains unchanged for other measures of poverty - depth and severity. Poverty Decomposition: Growth and inequality A standard view of the relationship between economic growth and poverty is that continued economic growth would reduce poverty. But growth may also be associated with rising inequality, which then tends to offset part of the gains the poor make from growth. The decline in poverty depends crucially on the extent to which economic growth has reached the poorer members of the population. How much would poverty have declined in Egypt if there were no changes in welfare distribution? To answer this question we apply growth-redistribution decomposition (Datt and Ravallion 1992). The change in poverty between two years can be decomposed into three components. The growth component is the difference between the two poverty indices when keeping the welfare distribution constants. The redistribution component is the change in poverty if the mean of the two distributions is kept constant. The residual component shows the change in poverty as a result of the interaction of growth and inequality. Table 5 presents the results of such decomposition for the national and regional levels. The 2.68 percentage point reduction in poverty in Egypt between 1995 and 2000 could be decomposed into a 1.72 percentage point decline 12 in poverty resulting from the increase in the mean per capita expenditure and a 1.1 percentage point decline due to the change in the distribution. In other words, if the mean per capita expenditure in Egypt remained unchanged between 1995 and 2000, the poverty incidence would decline by 1.1 percentage points just because of the change in the welfare distribution. On the other hand, if inequality stayed constant during this period, the growth in per capita expenditure would reduce poverty by 1.72 percentage points. Changes in both per capita expenditure and inequality resulted in decreased poverty rates. From 1995/96 to 1999/2000, growth and redistribution components worked in opposite directions in the Metropolitan region, where reduction in poverty (-9.18 percentage points) resulting from increased per capita mean expenditure was hampered by worsening inequality (1.07 percentage points). In Upper and Lower Egypt both growth and redistribution effects affected poverty in the same direction. In Lower Egypt, growth in per capita expenditure accompanied decline in inequality, resulting in the reduction in poverty levels (growth components are -0.61 and -4.45 for urban and rural areas, respectively, while redistribution components were -1.51 and -5.93). The opposite occurred in Upper Egypt, where per capita expenditure dropped and inequality worsened. These turns for the worse were reflected in the rise of all poverty measurements. Growth in the Metropolitan region was high enough to improve poverty levels but it was not a pro-poor growth. Poverty decomposition: Regional How did these regional differences in the changes in poverty affected the incidence of poverty at the national level? To answer this question we use regional poverty decomposition (Ravallion anrd Huppi 1991). The change in poverty over time can be decomposed into three components. The intra-regional effect is a contribution of within-region change in poverty to the overall change in national poverty. The interregional population shift measures how much national poverty would have changed if population shifted across regions but poverty within regions remained unchanged. The third component accounts for the interaction of the intra- and inter- regional effects. Applying the regional decomposition to Egypt over the period 1995/96- 1999/2000 shows that almost all of the reduction in poverty at the national level can be 13 attributed to the intra-regional effect. This effect reduced poverty by 2.77 percent (Table 6). The inter-regional population movement and covariance effects were small. With regard to regional roles in reducing poverty, the Metropolitan region may have contributed the most, lowering the overall incidence of poverty by 1.69 percentage points. The drop in poverty rates in the Metropolitan region and Lower Egypt dominated outweighed increases in poverty in Upper Egypt, resulting in a net decline in the level of national poverty. Thus, the decrease in national poverty between 1995 and 2000 could be attributed to decreases in poverty in the Metropolitan region and Lower Rural Egypt and, to smaller extent, to the decline in poverty in Lower Urban Egypt. 5. Who are the poor? Poverty profiles reveal differences in the relative poverty of certain subgroups of the population, and analysis of the poverty profiles is essential for understanding the causes of poverty. A profile of the, poor based on 1999/2000 HIECS data reveals that characteristics of the poor in Egypt are similar to those found in other countries of the Middle East. The Egyptian poor tend to live in large families, have low levels of education, work in an informal sector and be concentrated in low-paying unskilled activities. The rate of poverty of female-headed households is higher than the average for urban Egypt. Table 7 shows the proportion of poor households by the education level of the head and type of locality. Education appears to be a key dimension of welfare in Egypt. About a third of poor households are headed by an illiterate person. The incidence of poverty declines as household heads become educated. The households whose heads have completed basic education are more than twice as unlikely to be poor compared with the households headed by illiterate individuals. The likelihood of being poor falls almost to zero for households with heads holding university or postgraduate degrees. The effect of a head's education on the welfare status of a household seems to be much stronger in urban areas. Employment status of the household head also appears to influence the poverty status of households. The poor have lower rates of labor force participation than non-poor 14 (34.6 percent versus 37.8 percent respectively). Urban households with unemployed heads were twice as likely to be poor as urban households headed by a person with salaried employment. The lowest poverty incidence in urban areas occurs among households headed by a non-working head. In rural areas, employment status of the head appears to be uncorrelated with the poverty status of a household. Another important factor affecting the profile of the poor is sector and industry of employment. Households in which the head works in the private sector are more likely to be poor, both in urban and rural areas of Egypt, (the private sector is most prevalent in rural areas). At the national level, 20 percent of households with heads working in the private sector were poor in 2000. This is almost twice as high as the poverty rate among households with the head working in the public sector (11 percent). The influence of the household head's sector of employment on welfare status is more pronounced in the urban areas of Egypt, where most of the public and government sector jobs are concentrated. Looking at the industry of employment, the largest proportion of poor is concentrated in agriculture and construction. Table 7 shows that about 27 percent of poor households have heads employed in agriculture, followed by 20 percent in construction. The lowest incidence of poverty is registered in manufacturing and trade and finance. 6. Multivariate poverty profile and simulations So far in the paper we have documented the incidence and changes in poverty rates from 1995 to 2000. These indicators are intrinsically ex-post measures of well-being. At the same time poverty-reducing policies are forward looking. Policy makers try to design interventions that protect populations from future poverty. Such interventions are often based on an ex-ante assessment of probability to fall into poverty. To assess the probability of households in Egypt to be poor we rely on a two-step approach. Let total household consumption C, be a function of household characteristics Xi and assume that Cj is log-normally distributed. In the log form: In (Cd =X,,8+ ei, JI5 1 5 where e, is a normally distributed error term. Then the probability of household i to be poor is: Pi=prob(ln(Cd