Economic Activities: How the Poor Earn Income

Transcription

1 Economic Activities: How the Poor Earn Income This annex investigates the economic activities and income sources of the poor using the IHSES 2007 data, with a particular focus on self-employed agricultural workers. The main findings can be summarized as follows. Wage earners compose slightly less than 60 percent of employed poor workers. Around 50 percent of poor workers are employed either in agriculture (of whom 90 percent are self-employed) or in construction (of whom 99 percent are wage earners). According to the IHSES data, poor wage earners work longer hours than their nonpoor counterparts and are paid a very similar hourly wage. Hence, their being poor does not seem to depend on underemployment or low productivity. Labor productivity of self-employed poor workers, as measured by income per hour worked, averages two-thirds that of the nonpoor self-employed. Furthermore, poor workers on average seemingly have higher productivity as wage earners than when self-employed, while a much smaller gap exists for nonpoor workers. Self-employed agricultural productivity is estimated to be 40 percent lower than in all other economic sectors, both for poor and nonpoor workers. Nonpoor self-employed agricultural workers, however, display a large variability in productivity while poor ones appear more uniformly less productive. No clear evidence of plant production specialization is found between poor and nonpoor workers. 1. Economic Activities of the Poor According to the IHSES 2007, labor force participation in Iraq is very similar for the poor and nonpoor (around 43 percent). 2 Among the employed at the national level, the share of wage earners is lower among poor (58.3 percent) than among nonpoor (68.4 percent) workers (see Table 5.6-1). This is due to the situation in rural areas, where only 42.4 percent of poor employed workers earn wages (compared with 48.9 percent of nonpoor ones). Table Distribution of Employed Workers 15 and Older by Professional and Poverty Status Iraq Urban Rural Nonpoor Poor All Nonpoor Poor All Nonpoor Poor All Self-employed Wage earner Total Source: Our estimates using IHSES 2007 data 2 This annex considers those individuals who are at least 15 years of age. 388

2 Table Distribution of Employed Workers 15 and Older by Economic Activity and Poverty Status Economic Activity Nonpoor Poor All Of poor workers Wage earners Selfemployed Agriculture, hunting, forestry, and fishing Manufacturing Construction Wholesale and retail trade Transport, storage, and communications Financial intermediation, real estate Public administration and defense Education Health and social work Other community, social, and personal services Total Poor workers are relatively concentrated in a few economic sectors: almost half are employed either in agriculture or construction (Table 5.6-2). A further 22.7 percent work in trade or manufacturing. The comparable shares for nonpoor workers are 22.5 percent for agriculture or construction and 28.3 percent for trade or manufacturing. Furthermore, poor workers have very different professional status profiles in their two main sectors of employment: 90 percent of those in agriculture are self-employed while almost all those in construction are wage earners. In the next two highest sectors of concentration among poor workers, trade shows a less polarized distribution between the self-employed and wage earners (roughly a tilt), while manufacturing is similar to construction (although less extreme with a share of wage earners equal to 90.7 percent). Wages and Hours Worked In the entire economy, the median hourly wage (ID 1,900) is not significantly different between poor and nonpoor workers (Table 5.6-3). 3 Among the sectors where most poor workers are employed, the pattern is differentiated. In particular, while the median hourly wage in construction barely changes with poverty status (ID 2,200 for poor and ID 2,300 for nonpoor workers), the rate for poor workers in agriculture averages three-quarters of that for nonpoor workers (ID 1,500 versus ID 2,000, respectively). However, as Table notes, wage earners comprise only about 10 percent of all poor agricultural workers, or, around 3 percent of all employed poor workers. Very small and probably not statistically significant differences in wages across poverty status are also found in other activities such as trade; manufacturing; 3 Table presents the median because it is less sensitive to extreme values than the mean. However, the overall pattern would not change if the mean were considered. Weekly work hours have been calculated using questions 1314 (weekly hours of work), 1315 (annual days of holidays), and 1316 (annual days of vacation) in the IHSES More precisely, the following formula has been used: (q1314*52 ((q1314/7)*(q q1316)))/52. [5.6-1] That is, the number of hours corresponding to the days of holidays and vacations have been subtracted from the total number of annual hours of work, obtained multiplying the weekly figure by 52,.Weekly figures have then been obtained by dividing annual figures by the number of weeks in a year. 389

3 transport, storage and communications; public administration and defense; and other community, social, and personal services where a considerable share of poor workers are employed. Meanwhile larger wage gaps are observed where the shares of poor workers are smaller, that is, in education, health and social work, and financial intermediation and real estate. Finally, hourly wages in construction are the highest of any category for poor workers (and the second highest for nonpoor workers). To the extent that wages reflect labor productivity, the evidence in Table points to the possibility that productivity differentials between poor and nonpoor wage earners may not be large in many economic sectors. Table Median Hourly Wages by Economic Activity and Poverty Status (ID 1,000/hour) Economic activity Nonpoor Poor All Agriculture, hunting, forestry and fishing Manufacturing Construction Wholesale and retail trade Transport, storage, and communications Financial intermediation, real estate Public administration and defense Education Health and social work Other community, social, and personal services Total Median hourly wages also show regional variations (see Table 5.6-4). In particular, they vary for poor employed workers from ID 1,100 in Duhouk to ID 2,800 in Kirkuk, while ranging for nonpoor workers from ID 1,400 in Ninevah to ID 2,200 in Kirkuk. As these variation ranges indicate, the median hourly wage is higher for poor than for nonpoor workers in some governorates. The largest gaps in favor of poor workers are observed in Sulaimaniya and Kirkuk, while the largest in favor of nonpoor workers are found in Duhouk and Kerbela. This evidence seems to point to factors beside the hourly wage as determinants of an employed worker s poverty status. It also suggests that productivity differentials between poor and nonpoor wage earners vary across governorates. This is at least partly due to the geographical distribution of economic activities. 390

4 Table Median Hourly Wages by Governorate and Poverty Status (ID 1,000/hour) Governorate Nonpoor Poor All Duhouk Ninevah Sulaimaniya Kirkuk Erbil Diala Al-Anbar Baghdad Babil Kerbela Wasit Salahuddin Al-Najaf Al-Qadisiya Al-Muthanna Thi Qar Missan Basrah Total The median number of weekly hours worked shows substantive variation both across sectors and between poor and nonpoor workers (Table 5.6-5). In particular, at the level of the entire economy, poor workers work longer hours per week than nonpoor ones (36.0 versus 30.1 respectively). This pattern exists in most economic sectors, with only a few exceptions. One is construction, where the number of hours worked are very similar (37.5 for poor and 36.0 for nonpoor workers). Very small differences also are observed in trade, education, and other community, social, and personal services. Education is the sector where the shortest hours are observed (15.1 for poor and 14.8 for nonpoor workers). The longest weekly hours are lodged by poor workers in public administration and defense (44.8) and by nonpoor workers in trade (40.7). 391

5 Table Median Number of Weekly Hours Worked by Economic Activity and Poverty Status Economic activity Nonpoor Poor All Agriculture, hunting, forestry and fishing Manufacturing Construction Wholesale and retail trade Transport, storage, and communications Financial intermediation, real estate Public administration and defense Education Health and social work Other community, social, & personal services Total If the mean rather than the median number of weekly hours worked is considered (Table 5.6-6), the results differ somewhat, but the pattern remains very similar. Table Mean Number of Weekly Hours Worked by Economic Activity and Poverty Status Economic activity Nonpoor Poor All Agriculture, hunting, forestry and fishing Manufacturing Construction Wholesale and retail trade Transport, storage and communications Financial intermediation, real estate Public administration and defense Education Health and social work Other community, social, personal services Total

6 . The evidence in Tables suggests that (a) underemployment does not seem a cause of living standards for poor wage workers because they generally work longer hours than do nonpoor wage earners and (b) low productivity also does not seem to be a factor since poor and nonpoor workers generally are paid very similar hourly earnings. 2. Labor and Self-Employment Income Labor income a more comprehensive variable including not only wages but also self-employment earnings can also be measured using IHSES 2007 data, although only at the household level. Therefore, the labor income per hour presented in Table is calculated as the ratio between household labor income and the total hours worked by all household members. 4 It can be observed that this more comprehensive variable shows lower values for people living in poor than in nonpoor households. The difference is larger if the mean value is considered, but it is also present using the median value. Table Labor Income per Hour Worked by Poverty Status (ID 1,000/hour) Mean Median Deviation Nonpoor Poor All The gap in favor of nonpoor households also is observed at the governorate level with respect to both the mean and the median values (Table 5.6-8). 5 For some governorates (Baghdad, Ninevah, and Missan), the median values are equal between poor and nonpoor people. The largest differences (ID 700 per hour) are found in Sulaimaniya and Erbil. Median hourly labor income for poor workers ranges between ID 1,000 in Al-Najaf and ID 2,200 in Erbil. Meanwhile for nonpoor workers it ranges between ID 1,600 in Ninevah and Diala and ID 2,900 in Erbil. Hence the highest median governorate hourly labor income for a poor worker is 2.2 times greater than the lowest governorate rate, while the corresponding ratio among nonpoor workers is Table is based on the number of hours worked as measured in question 1204 of the IHSES 2007, where the time is recorded separately for each of the seven days preceding the interview. However, similar results can be obtained using question 701, which registers only the total hours worked in the same seven days. 5 The main exception is Erbil, where a higher mean but a lower median value is found for poor workers. Although here the result for the first indicator can be due to extreme values to which the second one is less sensitive. 393

7 Table Labor Income per Hour Worked by Governorate and Poverty Status (ID 1,000/hour) Poor Nonpoor Mean Median Mean Median Duhouk Ninevah Sulaimaniya Kirkuk Erbil Diala Al-Anbar Baghdad Babil Kerbela Wasit Salahuddin Al-Najaf Al-Qadisiya Al-Muthanna Thi Qar Missan Basrah Total The IHSES 2007 data allow one to measure self-employment income per hour worked at the household level as the ratio of total household income from self-employment to the total hours worked by the selfemployed members. This indicator can be interpreted as an average measure of labor productivity across self-employment activities and economic sectors. The figure for poor individuals is, on average, around two-thirds of that for nonpoor individuals using both mean and median values (Table 5.6-9). Furthermore, while the median hourly self-employment income is only slightly lower than the median hourly wage for nonpoor workers (ID 1,800 and ID 1,900 respectively), self-employment income for poor workers is substantially lower than wages (ID 1,200 and ID 1,900 respectively). Hence to the extent that these figures reflect productivity, one can argue that poor workers average higher productivity when employed as wage earners than when self-employed. A similar gap also exists for nonpoor workers, but it is small in size. Table Self-Employment Income per Hour Worked by Poverty Status (ID 1,000/hour) Mean Median Nonpoor Poor All

8 The average self-employment income per hour worked also is higher for nonpoor than for poor individuals at the governorate level (Table ). 6 There are two exceptions to this pattern, both using the mean (Baghdad and Ninevah 7 ) and the median values (Baghdad and Missan). Furthermore, the ratio between poor and nonpoor hourly incomes varies substantially across governorates: from 20 percent (in Sulaimaniya and Erbil) to 160 percent (in Ninevah) using mean values, and from 20 percent (again in Sulaimaniya) to 110 percent (Baghdad and Missan) using median values. Table Self-Employment Income per Hour Worked by Governorate and Poverty Status (ID 1,000/hour) Mean Poor Median Mean Nonpoor Median Duhouk Ninevah Sulaimaniya Kirkuk Erbil Diala Al-Anbar Baghdad Babil Kerbela Wasit Salahuddin Al-Najaf Al-Qadisiya Al-Muthanna Thi Qar Missan Basrah Total Self-Employed Agricultural Workers Given the sizable share of poor workers who are employed in agriculture (30.3 percent) and their frequency of self-employment (almost 90 percent), this section delves more deeply into the situation of self-employed agricultural workers. Using median values, self-employment income per hour worked is around 40 percent lower in agriculture than in all the other sectors, both for poor and nonpoor workers (Table ). However, the mean value of this productivity measure for nonpoor agricultural workers is larger than that in all the other sectors, indicating, together with the very high variability, that some of these nonpoor workers are very 6 Small sample sizes are likely responsible for the low values observed among poor self-employed workers in the Kurdistan regions. 7 The results for Ninevah using mean values could be driven by some extreme observations to which median values are less sensitive. 395

9 productive. In contrast, poor self-employed agricultural workers are more uniformly less productive. Using median values the productivity ratios of poor and nonpoor workers are similar in the agricultural and nonagricultural sectors: 75 and 79 percent respectively. However the presence of some highly productive nonpoor workers in agriculture reduces the productivity ratio in this sector between poor and nonpoor workers with respect to the rest of the economy calculated using mean values: the figures are 46 and 71 percent, respectively. Table Self-Employment Income per Hour Worked by Poverty Status and Economic Activity (ID 1,000/hour) The IHSES 2007 data contain information on other aspects of self-employed agricultural worker activities. We now compare the different types of plant production, the shares of auto-consumption, and the demographic characteristics of agricultural workers in poor and nonpoor households. The overwhelming majority of self-employed agricultural workers are engaged in crop production 91 percent at the national level, with little difference by poverty status (92 and 90 percent for poor and nonpoor workers, respectively) Mean Agriculture Median Figure Type of Plant Production by Poverty Status (Percent of market value of total plant production) Nonagricultural sectors Mean Median Nonpoor Poor All Nonpoor Poor 0 Source: Authors estimates using IHSES 2007 data 396

10 Very similar figures for poor and nonpoor agricultural workers also are found in considering the production of various types of plants as portions of total plant production (Figure 5.6-1) providing, in other words, no clear evidence of productive specialization between poor and nonpoor workers. Moreover, the share of total production not sold in the market (which can be interpreted as a proxy for auto-consumption) is very similar among poor and nonpoor households (Table ). These figures, however, are averages of more diversified ones, as can be seen if shares are considered at the level of per capita expenditure quintile. These shares increase from 28.8 percent at the poorest to 38.2 percent at the middle quintile before decreasing to a minimum of 21.2 percent in the richest quintile. This evidence suggests that the poorest households cannot afford to keep a large enough share of their production to satisfy their own consumption needs. Table Share of Total Production Not Sold in the Market by Poverty Status and Per Capita Expenditure Quintile % total production not sold in market Poor 29.7 Nonpoor 32.8 Per capita expenditure quintile 1 (poorest) (richest) 21.2 Finally, the IHSES 2007 also collects information about the characteristics of the first four household members involved in agricultural activity, and particularly in land cultivation. Table shows that the share of males among these agricultural workers is not significantly different across poor and nonpoor households (the average being 78 percent). Among poor households, the average age of the people participating in agricultural work is slightly lower than in nonpoor households. Finally, the share of working children is slightly higher in poor than in nonpoor households. However, many household members are not captured in the indicators supplied in Table , particularly for poor households in rural areas, where the average household size is 9.7 people. And the characteristics of the unconsidered members who participate in agricultural work may well be different from those of the first four members. For instance, the actual shares of children participating in agricultural work may be higher than the figures reported in Table , in particular in poor larger households. If this was the case, the actual demographic characteristics of household members participating in agricultural work could be different between poor and nonpoor households. 397

11 Table Household Members Participating in Land Cultivation by Poverty Status Male (%) Age (years) Working children (%) Nonpoor Poor All