Assessing Poverty in South Asia

Assessing Poverty in South Asia Introduction UNDP has taken an initiative to publish SOUTH ASIA POVERTY MONITOR periodically to assess the poverty situation at national grassroots level through the existing national expertise in South Asia. As part of this initiative a country report will be prepared for Bangladesh as well. The Bangladesh country report will be prepared through both quantitative and qualitative approach. Unnayan Shamannay is proposing to conduct the qualitative part of the study. Rationale for Qualitative Approach Statistical data do help very little in understanding what the variation means. Qualitative data, on the other hand, illustrate the value of detailed, descriptive data in deepening our understanding of individual variation.They give rise synergistically to insights and solutions that would not come about without them (Palton 1990:15-17) Qualitative approach can provide a depth of understanding of the issues associated with poverty that the more formal and statistically valid approaches may not. This class of studies includes the increasingly popular techniques of rapid and participatory rural appraisal and beneficiary assessment (WB 1992: 8-4). Objectives The objectives of this study are as follows: Assess poverty through qualitative methodology Complement the quantitative approach with qualitative one. Add qualitative dimension to the Bangladesh Country Report. Scope The scope of this qualitative study will be to: Identify indicators of poverty through a participatory approach Identify and monitor changes in the poverty situation Assess the impact of some of the poverty alleviation measures Analyse the findings Topics/Issues to be Addressed Poverty profile and poverty indicators are some of the important components of poverty assessment. Poverty assessment will be carried out in participatory manner. Broad topics on the extent of poverty, identification of sub-groups, nature of poverty, characteristics of the poor and risk management have been included in the proposed research agenda. Moreover, poverty monitoring will also be conducted periodically and it will act as a barometer to measure the changes in various socio-economic and welfare indicators relating to the lives of the poorest households. Methods to be Used All major qualitative research methods will be used in the study. Interview will be extensively used in the study including its key variants, namely participatory group discussion, focus group discussion, standardised open-ended interview and case study (Figure 1). In addition to interview, other methods of qualitative inquiry, namely observation and document analysis will also be made use of in the study. Selection of Sample Areas Qualitative exercises will be conducted in both urban and rural settings of the country. To cover the greater diversity in socio-economic environments, three different regional configurations of northern, central and southern parts of the country would be accommodated in the study. A total of six villages including two from each part would be covered under the study. In urban area, at least three slum areas would be covered to facilitate the comparison and triangulation of data and information. However, for monitoring of poverty in the selected six villages and three urban slums, certain number of the poorest households will be selected from each of the study sites. Out of six villages, three will be selected in such a manner where at least anti-poverty intervention by government is in operation. These three villages will serve as programme villages and they will be drawn from the three parts of the country including one from each. Besides, other three villages will also be selected nearby where there is no poverty focused government intervention. These three will serve as control villages in the three parts of the country. Sample Size Most of the topics would be addressed at the community level and no specific number of participants are needed to be ascertained beforehand. For poverty monitoring, a total of 120 poorest households will be selected  ¾ 90 from six villages and 30 from three urban slums. The poorest households will be selected through consultation with the respective community members. Tools to be Used In selecting tools desirable characteristics namely easy, simple, visual, non-verbal etc., must be taken into accounts. As PRA tools are recognized to have all these desirable characteristics, most of the tools will be drawn from its repertoire. Important PRA tools that will be extensively used in the poverty assessment include scoring and ranking, matrix ranking, wealth/well-being ranking, time line, social mapping, pie chart and so forth (Figure 1). Validity and Reliability Although the qualitative data are essentially based on the perception, opinion and judgement of the participants, the quality of data would be, nevertheless, refined through of the triangulation principle underlying the research design of this study. A combination of multiple sources, researchers/facilitators and on-the-spot cross-checking of data through discussion, debate and deliberation among the community participants would minimise the degree of error and bias of data to the minimum. Besides, the field observation by the researchers would in addition , act as a guard against any major inconsistency and biasness of data. Activities to be Undertaken For conducting the study a number of activities will be undertaken. The activities include: Identifying and reviewing available literature Now-a-days wide ranging literature on poverty is available. Different facets of poverty have been discussed in those literature. The indicators, measurement process, sampling frame etc. also differ. For a qualitative study for monitoring poverty, the volume of the problem further increases. The approach is not only different but gives a deeper insight. To make it complement the qualitative approach the literature on poverty needs a review. For this purpose all available literature on poverty will be reviewed. Analyse presently used indicators Before finalising the indicators for assessment of poverty there is a need for analysing the presently used indicators. This will provide a rational basis for the use of the indicators in the qualitative study. Pre-testing The indicators and tools to be used in the study will be pre-tested in the field. This will help understand the effectiveness of tools. Primary field visit Before starting field work a primary field visit is needed to get acquainted with the actual field condition. Training of field/research officers The field/research officers who will be engaged in this study are competent and experienced. Even then the field/research officers need training/orientation for doing such work. With this purpose they will be imparted in-house and field training. Processing of data/information The data/information processing in qualitative study is not similar to that of quantitative one. The information generated through qualitative approach is processed in a different manner. Different factors and aspects of reality are considered while classifying these information. Field activities The field activities to be undertaken in this study will require two types of work: a. in rural area and b. in urban area. Rural area: The activities in rural area will require identifying the group/sub-group, building up rapport with them and conducting the sessions. These activities have to be co-ordinated with the day-to-day activities e.g., ploughing or rowing time etc., of the participants. Rrban area: In urban area conducting participatory session is a difficult task. Urban life makes it difficult for the participants to spare time for such research. Besides building up a better rapport, tools need to be designed and adjusted accordingly. Document analysis Significant insights can be found through document analysis. Even discrepancies between reality and pronounced goals can be identified. CHAPTER X PROBLEMS IDENTIFICATION AND NEEDS ASSESSMENT BY THE POOR Problems and Needs Assessment By the Poor Problems facing the poor were identified by the poor themselves, and a list of felt needs were the outcome of the participatory discussion, debates and consensus among themselves. Two sets of problems and needs were assessed in a participatory manner each for the urban and rural areas. Urban Setting To the urban slum poor, homelessness and eviction from slums are the topmost problems. Other serious problems identified by the poor include lack of good health and water facilities, employment opportunities, security, education, latrine, gas, etc. (Exhibit 38). Regarding the needs assessment, the urban poor listed and prioritized their felt needs. Some of the most important are, latrine, shelter, drinking water, electricity, gas, security, rationing, employment and so on (Exhibit 39). Rural Setting Agricultural inputs, irrigation and culverts are considered to be the topmost problems by the rural poor. Apart from these, some other most serious problems mentioned by them are related to health, electricity, unemployment, flood, drinking water, industrialisation, veterinary facilities, silting up of rivers etc. (Exhibit 40). According to the needs assessment and prioritization by the rural poor, some of the most important needs as articulated by themselves are industries for employment, agricultural inputs at a fair price, rural roads, irrigation, electricity, school and madrasa, medical facilities etc. (Exhibit 41). Chapter IX Monitoring THE Impact of Public Expenditure on Poverty Objective The primary objective of monitoring of impact of public expenditure on poverty in this chapter is to understand the living condition of the poor. This is more of an illustrative exercise rather than a whole sector monitoring of poverty. The issue of representativeness has to be, therefore, viewed in this context. One of the stated objectives of the development strategy of both present and previous governments is to reduce poverty. A growing share of public expenditure is claimed to have been allocated to the development activities ostensibly aiming at poverty reduction in the recent past, and this is likely to be continued in the future. Against this background of increasing the public expenditure allocation to poverty alleviating projects, it is needed to know the effects and impacts of these expenditure on poverty alleviation. In this section a number of key questions have been addressed: Does the benefit of the public expenditure reach those lying at the bottom of the income scale ? Is there any sign of improvement in the condition of the poorest of the poor ? How do the selected poverty indicators behave ? Do they improve, deteriorate or oscillate ? In case of improvement, at what pace do they improve ? Based on the findings from these questions, an attempt will be made to assess the quality of public expenditure in terms of a set of selected indicators. To understand the trend of the impact of public expenditure on poverty, we started monitoring the behaviour of some selected indicators of poverty in both the urban and rural areas since 1993 as the base year. The qualitative and quantitative data generated throu gh the participatory tools have been used for this poverty monitoring. This is the first round of the periodic monitoring of poverty in a participatory manner. Poverty Assessment and Monitoring: Peoples Views The poverty assessment carried out under this study has two components. The community members actively participated in the assessment of their well-being by listing and categorizing of all the households by themselves in several groups based on their own criteria. This is, in fact, a subjective assessment. Secondly, after categorization, all households were arranged in descending order on the basis of well-being scores of each of the households resulting in the identification of the poorest of the poor in the respective communities lying at the bottom of the scale with quantitative precision which was again vetted by the community members/participants. The poverty of some of the poorest households in the community has been monitored on the selected indicators. As this monitoring is based on hard data, it, therefore, gives us an objective assessment of the living standard of the poorest. (Figure 9.1) The poverty sitution in the urban and rural areas has been assessed in a participator y manner. Instead of applying any pre-conceived ideas, standards, measures or categories by the researchers to measure poverty as is done conventionally, the criteria used in this study has been developed by the people at the community level. The basic question relating to poverty measurement or assessment is who is poor and how to identify him/her. Based mainly on qualitative data information Based mainly on quantitative data information Criteria Unlike a single standard or formula as applied in the conventional methodology, the community-members consider it appropriate to use a set of socio-economic criteria to assess the economic and social status of a household. For this purpose, the researchers and facilitators involved in the study initiated a series of group-level discussions and community-level validations. The community people developed their own criteria (Box 9.1) to assess the status of their own members and also to categorize them into a set of social classes. The more important criteria developed by the rural people in the selected villages are, among others, the amount of land owned and cultivated, the number of earning members, cash in hand, the housing condition, the amount of fixed assets, the family size, other sources of income, whether a household is female or male headed, etc. Prevalence of poverty Based on the above criteria, the community people identified the poor (moderate poor) and the poorest (extreme or hardcore poor) households in their own community. As poverty was assessed at the household level, the status of all the households in the community was assessed and categorized into four classes, namely well-off, medium, poor and poorest. In the urban slums, 72 percent of the households were found poor (moderate: 51, hardcore: 21) and 28 percent non-poor (middle: 19, well-off:9) (Tables 9.1 and 9.2). The incidence of poverty was, however, found to be widely different in different slums. In one sample slum there were no well-off households in 1996 although there were many in another sample. In the rural area, 75 percent of the households were classified as poor (moderate: 20 and hardcore: 55) whereas 25 percent were classified as non-poor (middle:14 and well-off: 11) (Table 9.3). Regionally, the incidence of poverty was more acute (moderate: 17, hardcore: 60) in the central part compared to that (moderate: 25, hardcore:47) in the northern part. The findings generated by the PRA exercise were further validated by the people in the respective community. So the scope of subjective bias, if any, was greatly reduced. Poverty Monitoring Using Panel Data Set (Quantitative) Being a value loaded term, poverty as such cannot be measured quantitatively/objectively. The debate on the issue abounds in the literature. But the symptoms and aspects of poverty can be measured and monitored by means of a series of socio-economic indicators that proxy the level of well-being of people. That is why, an attempt has been made in this section to measure and monitor poverty through a number of indicators/variables in two different years i.e., 1993 and 1996. Most of the indicators used for monitoring were suggested by the community members (Box 9.1.). The number of indicators used here are meant to have satisfied the desirable criteria, namely, unambiguity, consistency, specificity, sensitivity and ease of collection (Carvalho and White, 1994). Change in Demographic and Socio-economic Profiles of the Poorest Households During 1993-96 Demographic and Social Characterstics Family size and composition The population of the poorest households and their average family size grew by 5 percent over the monitoring period 1993-96 (Table 9.4). However, the populatioin growth rate is found to have been higher at 7.2 percent for the urban poor compared to 4.4 percent in the rural area over the same period. The family size of the poorest households in the rural area is, however, found to be higher at 4.2 in 1993 and increased further to 4.4 in 1996. The family size of the urban poor was lower at 3.5 in 1993, and it grew to 3.7 in 1996. In the rural area, the family size of the FFE-households is found to be much higher at 6.0 on an average in both the central and northern parts compared to those for the non-FFE households in both programme and control villages in 1996 (Table 9.5). Another important demographic characteristic of the poorest households is their family composition. In 1996, the FFE households are found to have a male majority  ¾ 61 percent compared to 49 percent and 41 percent for the non-FFE households in the programme and control villages respectively. The family composition is, however, found reverse for the poorest families in the urban slums. The poorest households had a female majority at 62 percent in 1996 (Table 9.6). The above findings pose some questions challenging the appropriateness of the main thrust of the development strategy being pursued by the government in the country. The much-publicized motto two children are enough seems to have been irrelevant so far as the poorest people are concerned in both the urban and rural areas. The increasing growth rates in populatioin and family size suggest that under the existing socio-economic conditions, their economic and social securities lie not in smaller family but in larger one. Earning members and incidence of child labour The poorest households and their different groups are found to have peculiar characterstics in the composition of their earning members. Overall, close to half of the earning members are men, and one-fourth are women and boys each in 1996 (Table 7.17). Against this general distribution of the earning members, the poorest families in the urban and rural areas are found to have different compositions of earning members by age and gender. In the urban slums, female earning members accounted for 43 percent (women: 36% and girls: 7%) among all the earners compared to 24 percent (women:23% girls:1%) in the rural area (Table 7.17 and 9.7). Female children are not found to have been as active in income earning activities previously as they are found to be in 1996. The preponderance of male income earners is found to be more prominent among the poorest households in the rural area. At the disaggregate level, the difference is more revealing in the rural areas. The participation of girls in income earning activities is found to be very minimal throughout the rural areas (Table 9.8). Among the FFE-households, womens participation in income earning activities is very small (3%), but it is widely observed (33%-36%) among the non-FFE households . Among the FFE households, the preponderance of male child labour is observed, and this remained unchanged throughout the monitoring period despite the programme intervention in the rural areas. The incidence of child labour among the earning members of the FFE households is found to be 40 and 41 percent in the central and northern parts respectively of the country, and this remained unchanged in both the areas during the period 1993-1996. The poor impact of the FFE programme on the incidence of child labour at large in the rural areas is also revealed sharply if we focus on the trend in the incidence of child labour. Overall, 25 percent of the boys of all ages were involved in income earning activities in 1993, and this remained almost at the same level (24%) in 1996. As the boys, the incidence of female child labour among the earning members is found to be at a much lower level (1.2%) in 1993 and this remained at that level 1996 as well. The above findings raise an important question to the fore: why is the FFE programme found to be ineffective in reducing the incidence of child labour ? The answer to this question should be searched not in the programme itself but in the economics. For the poorest households, the opportunity cost of sparing a boy from education is around Tk. 14 a day (wage rate) in 1996 (Table 9.9). The financial benefit gained from the FFE programme by a rural poor household is found not so significant at Tk. 4.85 (Tk. 0.81 per capita per day) a day for a boy (Table 9.10). The participatioin of a poor family in the FFE programme causes a substantial income loss to that family. As the benefit under the programme cannot offset the income loss that an extremely poor family has to incur, the appeal of the programme to a precariously income-poor family is found to be weak. This finding is found consistent with that of other studies (Ahmed and Billah,1995). Female-headed households One of the important demographic features of the poorest households is that close to one-third of them were female-headed during the reference period (Table 9.11). More than half of the sample households (55%) are found to be female-headed in the urban slums compared to 23% in the rural households during the same period. Another important demographic feature of the three groups of the poorest households is that only 5 percent of the FFE households have been female headed compared to 25 percent and 40 percent for the non-FFE households respectively in the programme and control villages in 1993 (Table 9.12A). This composition remained unchanged even in 1996. The above findings suggest that the FFE households are found to be relatively stable not only in respect of assets (details later) but also demographically. The preponderance of female-headed households among the non-FFE household groups imply that these households are not only income-poor but also subject to a higher degree of vulnerability and defencelessness both economically and socially. In the urban slums, a significant portion of the poorest households happened to be female-headed during the monitoring period (Table 9.12B) The gender focus of poverty is found more pronounced among the poorest segment of the slum-dwellers compared to those in the rural area. Table 9.11 shows that more than half (55%) of the sample households have been female-headed compared to that (23%) among the rural counterparts during the same period. Begging households Altogether, 6% of the poorest households are found engaged in begging. In the urban slums, none of the poorest households is found in this category (Table 9.13) and all begging households under our sample belong to the rural area. Besides, all these households are found among the non-FFE groups. (Table 9.14). These households are more vulnerable and extremely poverty-ridden mainly due to some unfavourable demographic factors. The households engaged in begging are relatively small (3.8) in family size compared to the sample average (4.2) in 1996. Moreover, the dependancy ratio for the begging households is lower (2.7) compared to that for the sample households (3.0) in 1996.The predominance of women among the earning members points to the poor income level of these households. As the dependency ratio is very low, it implies that most of the family members are forced to go for earning activities due to their poverty. Income Source of income The poorest households have limited sources of income. The urban poor are usually engaged in unskilled manual labour. Similar is the case with the rural poor (Table: 9.15) as well. Sale of labour has been the main source of the rural poor accounting for 82% of their total income in 1993. This has marginally increased to 84 in 1996. Agriculture is the second most important source of income making up only 12% of the total income of the rural poor in 1993 and 10% in 1996. Only 1% of the income of the rural poor has been derived from livestock, a new source of income, in 1996. Nominal income In the rural area, the income of the poorest households has been found to be miserably low during the monitoring period. The per capita daily income of these households was Tk. 6.9 in 1993. This increased to Tk. 7.4 in 1996 showing an 7% growth (Table 9.16). Their per household daily income grew by 12% from Tk. 29 in 1993 to Tk 33 in 1996. The higher growth rate of nominal income is mainly due to a positive growth of the nominal wage rate (12%) alongwith a growth of the number of earning members (5%) of the poorest households. The low per capita income is partly due to the large family size and its growth over the monitoring period. The low income of the poorest households is the result of a number of socio-economic factors, e.g., low wage rate (Table 9.9), poor asset base, poor human capability due to illiteracy (Tables 7.31 and 7.32), low access to economic opportunities, etc. The impact of the FFE programme does not seem to have been appreciable on the level of income of the programme households. Although the programme has had some positive impact on the growth of income (15% in per capita and 18% in per households terms during 1993-1996), its contribution to the growth is difficult to ascertain. However, other findings indicate that the contribution of the programme to the income of the programme households is insignificant (Tk. 0.81 per capita/daily, Tk. 4.85 per household/daily, 15% of the average household income) (Tables 9.10 and 9.16). The per capita nominal income of the poorest households in the urban slums was Tk. 12 a day in 1993 and increased to Tk. 19 a day in 1996 representing a 31 percent growth (Table 9.17). The per household daily income of the urban poor increased by a higher rate of 40 percent from Tk. 41 a day to Tk. 58 during the same period. The income of the urban poor increased by a much higher rate than that of the rural poor in both per capita and per household terms because of the higher growth rates of wage (29%) (Table 9.9) and of earners per household (17%) (Table 9.7), lower family size (3.7), etc. Moreover, gainful economic opportunities are greater in the urban area relative to the rural area. Real income The income of the poorest households in real terms (in kilogram of coarse rice) is found to have declined across the board during the monitoring period. In the rural area, the per capita real income of the poorest households declined by 22% on an average from 0.9 in 1993 to 0.7 kilograms of coarse rice in 1996 (Table 9.18). Barring the FFE households, the per household real income has registered a sharp decline during the same period irrespective of differences in regional diversity. Due to the income support under the FFE programme, the FFE households could avoid the sharp fall of income. The per capita real income for the FFE households has declined by 13% against a 20 to 25 percent decline for the non-FFE households over the same peiod. Overall, despite an 7% increase in per capita income in nominal terms on an average during 1993-96 (Table 9.16), the corresponding real income took an appreciably higher downward trend (22%) (Table 9.18) caused by a 24 to 43 percent price hike of c oarse rice in the rural areas during the same peirod (Table 9.19). The per capita real income of the urban poor remained unchanged, whereas, the per household real income marked an upward trend (5%) during the monitoring period (Table 9.20). The per capita real income of the urban poor is almost double at 1.4 kg a day of that of the rural poor in 1993 which remained almost unchanged during the same period. The per household real income of the poorest households stood in urban slums at 4.7 kg and 5.0 kg a day in 1993 and 1996 respectively recording a 5% growth. The poorest households in the urban slums are relatively better off than their rural counterparts in respect of per household real income which declined by 16% for the latter during the same period (Table 9.18). Wage rate The unskilled wage rate is considered to be an important indicator for monitoring poverty. The wage rate of all categories of unskilled wage labourers is found to have increased in both the rural and urban areas (Table 9.9). In the rural area, the daily nominal wage rate increased by 11.7% from Tk. 17.2 in 1993 to Tk. 19.2 in 1996 (Tables 9.7, 9.16, 9.24 and 9.25). The wage rate is found to be much higher for the urban slum-dwellers, and it grew by 29% from the level of Tk. 35.8 in 1993 to Tk. 46.1 in 1996 (Tables 9.9, 9.21, 9.22 and 9.23). Although the wage rate for unskilled labourers increased during the monitoring period, the purchasing power of the poor labourers did not rise due to a higher rate of price increase in the case of coarse rice. The average wage rate for unskilled wage labourers, in fact, declined across the board in real terms during the monitoring period. However, the poor in the northern part had to sustain a much higher rate of fall (22%) in real wage rate compared to 14% for those in the central part during this period (Table 9.24). Consumption Consumption of food The consumption of rice and wheat  ¾ the staple food items of the poorest households  ¾ is found to have recorded opposite trends among these households in the urban and rural areas. In the urban slums, the per capita daily consumption of food (rice and wheat) was 442 grams in 1993 and it rose to 514 grams in 1996 representing a 16 per cent growth (Table 9.25). The increase in the consumption level of food in terms of both per adult equivalent unit and per household units has also been substantial, 18 and 25 percent respectively during the monitoring period. These findings, however, conceal the substantially low level of food intake observed in one of the slums where poverty is found to be more acute (Table 9.26). In the rural area, the trend in food consumption is, however, found to have consistantly sunk during the monitoring period in per capita and per adult equivalent and per household terms (Table 9.27). The per capita daily consumption of rice and wheat declined from the level of 585 grams in 1993 to 566 in 1996 showing a 3 percent decrease. The food consumption per adult equivalent unit is found to have been at a much higher level  ¾ 797 grams in 1993 and 786 grams a day in 1996 – recording a relatively small fall during the period. Per household consumption, likewise declined during the same period. The declining trend in food intake is true of both the programme and non-programme households during the same period. The consistent fall in the level of consumption of food is largely due to the fall in real income and expansion of the average family size of the poorest households during the monitoring period. Box 9.2: Food Security: A Quantitative Assessment In order to assess the poverty status of sample households, the heads of the households were asked to make self-assessments in respect of poverty. Their self-assessed status may be categorized as follows: Chronic deficit households reporting food shortage throughout the year; Occasional deficit households reporting food shortage occasionally in a year;