Measuring poverty is a challenging and complicated task because poverty itself is complex and multidimensional. Although conventional income and consumption based measures are essential for poverty monitoring, they seriously fail to capture critical aspects of well-being. Well-being, however, is more difficult to define as it encompasses a multitude of dimensions, both economic and non-economic. While the estimations of asset holding, income, and consumption is a direct quantitative measure of poverty, the level of human development captured through education and availability of health facilities and basic amenities is supposed to be an indicator of overall well-being. Moreover, many dimensions of deprivation experienced by the poor can be best identified by soliciting their own perception and insights. Hence, before going into various dimensions of well being, we will have a look on the perception by the poor themselves about poverty, and ill being.

Against this background, the purpose of this paper is to analyse present status of rural poverty in Bihar and changes therein during the last two decades. The paper is largely based on a comparison of panel data generated through two primary surveys during 1981-83 for the first time and 1999-2001 for the second time from the same 12 villages, spread over six districts, of the state. On the basis of the data collected at two points of time, it is argued in this paper that although there has been marked improvement in the living status of people a large proportion of lower castes and classes population continues to live in abject poverty and are leading a very poor quality of life.

The paper is organised in three sections. Section I is based on qualitative data collected in the 12 villages through participatory rural appraisals (PRA) and presents perception of villagers regarding poverty and changes in the living conditions over time. Section II presents a crude estimation of poverty on the basis of income and expenditure criteria. The third section (Section III) presents a detailed account of pattern of consumption expenditure and food sufficiency to the households. A changing nature of consumption expenditure asset holding position and food availability has also been presented in this section. Section IV deals with other aspects of human development such as education, heath, access to institutions and civic facilities etc. Section V presents conclusion.

Participatory studies of poverty reveal that the poor use different criteria to assess their own situation and its change over time. An analysis of the criteria used by rural people in Bihar shows that they use a wide range of criteria in which, apart from ownership of assets and land, education and health status and physical ability feature prominently when grouping and ranking the state of their own well-being. Further, the poor attach considerable importance to personal freedom and dignity, in spite of the fact that they have become increasingly vulnerable with the decline in patron-client relationship and secure employment.

We received a standard response while conducting the wealth ranking exercise – "most of us live in distressful conditions and are therefore poor". However, further probing and clarification reveal that in all the villages, the villagers feel that identification with the poor will result in some monetary benefit from the state. In most cases, people perceive and relate to poverty along caste lines also. Almost everywhere most of scheduled caste-groups, especially mushar and chamar, are perceived as poor and their position in the continuum delineated by the people themselves is consistently low. Most of them a) live in segregated colonies in mitti-phus (mud-hay) houses; b) are landless; c) both men and women work as agricultural labor; d) have many children; and are likely to have the highest number of out of school children; e) visit the government hospitals when ill; and f) borrow to meet food requirements of the family etc. Generally speaking, people identified the following indicators of their well being or ill being:

• widowhood and deserted women
• disability and old age
• quantity and quality of land available with households
• Availability of irrigational facilities
• presence or absence of person working in government
• ownership of mechanised agricultural and household articles
• saving capacity and ability to lend
• self-sufficiency with respect to food
• ability to access quality educational and health services
• number of children particularly girl children
• Caste and class of the households

Although the list of indicators identified by the poor themselves are more or less similar in different villages, there are some variations in their perceptions in respect of each across the villages. A district-wise comparison reveals that, in Purnea not only are people living in abject poverty, but there is also less difference between the richest and the poorest except in terms of land holdings. Similarly, land seems to be more equally distributed in Gopalganj district. On the other hand, in the districts of Madhubani and Purnea a relatively higher proportion of households is on the brink of starvation during lean seasons, while there seems to be a sense of discontentment amongst the people in the villages of Gaya and Nalanda districts. In these districts, villagers categorically stated that the food situation has drastically improved over the years. In respect of number of school children who are out of school, though, the same indicators appeared in all the villages, a comparison across districts reveals that relatively fewer children are out of school in Rohtas and Gopalganj and that even the poorest and lowest castes in these areas try to educate their children. In Nalanda and Gaya too, most families attempt to educate their children. However, factors outside their control such as caste-tensions, closing down of school, lack of teacher etc. prevent the poorest from availing any benefit. In Purnea and Madhubani districts not only are the highest number of children out of school, people in these districts as well as Gaya district are labelled as the 'poorest', when their children work in the fields. Across districts, the households with a) a more permanent structure (pucca houses), b) food to eat all year around, and c) ability to educate children in private schools and treat their illness in private clinics are considered 'wealthy'. Though, in all the villages households where women do not work are considered as rich, in many it was also quoted as the reason for the decline in the economic condition of the household in particular and community group in general.

The difference between districts also arises in terms of ownership of assets. Only a handful of very rich households own any mechanised agricultural implements (tractor, pumpsets, thresher etc.) in the districts of Purnea and Madhubani. In contrast, a larger proportion of households in Rohtas, Nalanda, Gaya and Gopalganj districts own both mechanised agricultural and household goods. Further, from the PRA reports it is obvious that material poverty is closely linked to the level of household consumption irrespective of the ownership of assets. For instance, in spite of owning fewer assets, the poor in Purnea district perceive an improvement in their lives and attribute this improvement to migration. Across districts, in all the villages, there is a perceptible air of two contradictory feelings: joy and anger, both among the poor and very poor middle and lower caste people. They attributed their joy to the increased freedom to choose both place of work and employer for whom they would work and to decide who to vote for. In fact most of them welcome the increasing monetisation of village economy and casualisation of labour and see it as a sign of reduced caste-based subservience. In contrast, especially in the villages of Nalanda and Gaya districts, amongst the lowest caste (also the poorest) there is a growing sense of frustration and anger, which in many cases (villages Rupaspur-Salempur and Kanadi of Gaya district) is tightly reigned.

Villagers’ responses to changes in the economic conditions of the households and village life as a whole have also been collected through the household questionnaire. Specifically their responses to ‘improvement’, ‘deterioration’, and ‘no change’ in their economic conditions as well as village life in general have been classified for different castes and classes and presented in table1.

Table 1 reveals that among different social classes, a significant proportion of middle peasants (67.68%) responded in favour of improvement in their economic conditions. Moreover, approximately 80 per cent of the people belonging to this class are of the opinion that the social and economic conditions in the village as a whole have improved in the last ten years. The responses of the middle peasant class are substantiated by the fact that this class mainly comprises of middle castes—Kurmi, Yadav, and Koeri —that are the most upwardly mobile castes in rural Bihar. The positive response from these castes are borne out by the fact that more than half of the persons belonging to them agree that there has been an improvement in their economic conditions. On the other hand invariably most of the people from the agricultural labour class are of the view that either their economic situation has deteriorated or remained unchanged, except in case of cultivating casual agricultural labour class. Regarding the conditions of village life in general most of the people from all the classes said that there has been an improvement over the last ten years. On the other hand the response of Kurmi and upper caste Muslims followed by upper caste people to conditions in village life has been less than average positive response. This may be due to the growing caste-class conflict and tensions in the rural areas, which in turn is attributable to the perception of a substantial proportion of Scheduled Caste that their economic conditions as well as village life have either worsened or remained unchanged.

In terms of land size groups, most of the people from higher land size households are of the opinion that over the years their socio-economic conditions as well as that of the village as a whole has improved. Again most of the people from landless class feel that their conditions have either deteriorated or remained unchanged.

The perception of people regarding their own economic conditions and that of village life in general has a far reaching implications. While on the one hand, it gives an idea of how villagers perceive their own relative position in the society, on the other, it throws light on the their hopes and aspirations. In the light of the above mentioned perception of poverty and change in the social status, we will try to evaluate different dimensions of poverty as indicated by villagers themselves in the following section.

Besides perception of poverty by the poor themselves we have also calculated income poverty on the basis of income and consumption expenditure of the households. For comparing the variations in income and consumption levels we have calculated average monthly per capita expenditure (MPCE) and income (MPCI) of households belonging to different castes and classes. For the calculation of MPCE all types of family expenditure has been taken into consideration. The items of expenditure which have been taken into account are expenditure on cereals, non-cereal food items such as edible oil, spices, milk and milk products, meat, eggs, fruits, etc. non-food items such as smoking, drinks, fuel, cosmetics, etc. recreation, books, magazines, clothing, social ceremonies etc. Accordingly, for calculating average monthly income all possible sources of income of households have been considered. These include income from agriculture, animal husbandry, self-employment in business and other non-farm activities, regular employment, casual employment, migration, and some other sources such as rent, pension, usury, leasing etc. The average consumption expenditure and income of households shows that apart from low levels of annual income and consumption expenditure there is also a wide variation in the income and consumption expenditure across different castes and classes (Table 2).

The average MPCE is higher than MPCI. This could be due to several reasons; first, there is always a possibility of underestimation of income of rural households as working members of the family are involved in wide array of economic activities including multiple work undertaken by the same person. Second, additional consumption of poorer households arising from access to common property resources. Third, incurring of debt to meet current consumption need etc. However, the comparison of MPCE and MPCI gives some interesting findings across caste, class and landsize categorisation. For example, households from some categories show higher level of income over consumption—among castes Kurmi shows far higher level of MPCI over MPCE and among Yadav MPCI is marginally higher than MPCE. Across class, big peasants show marginally higher MPCI over MPCE. Across landsize groups, households in larger landsize group of 10 acre or more shows higher MPCI over MPCE. In case of highest landsize group of 20 acre and above, MPCI is almost double of MPCE. Interestingly, the average level of MPCE shows contrasting picture; MPCE of scheduled caste is less than half of Kurmi; that of agricultural labour classes are around half of big peasants and landlord classes and that of the landless is less than one-third of landsize group of 10 acre or more.

The above analysis gives an idea about the levels of living in economic terms across different categories of households. But there are large differences among household falling within each category of caste, class and landsize group. Thus, we present percentage distribution of households across different consumption groups across caste, class and landsize categories. The MPCE group classification had been taken from 55^th round NSS consumption expenditure survey to enable us to compare poverty level between these two surveys. Given the Planning Commission poverty line of rural Bihar as Rs. 333, percentage of households falling below poverty line (having MPCE up to Rs. 330) is 46.71 per cent, which is far higher than NSS figures of rural poverty 41.50 percent. However, average MPCE figure of Rs. 450 in the present survey is higher than the NSS figure of Rs. 385. This could be because data in our sample is collected at one point of time period whereas NSS figures are collected in four sub-rounds. The much higher poverty level in our survey is attributed to larger concentration of households (nearly one-fifth of all surveyed households) in the 0-250 MPCE group (Table 3).

One reason for this could be that our survey is based on class stratification, which gives more weightage to agricultural labour households that forms the bulk of poor households. The weighting diagram used by NSS would, therefore, be quite different. Still so high concentration of household in 0-250 MPCE group cannot be explained only through differences in weighting patterns of sample surveys. It seems that in case of rural Bihar there is an underestimation of the lowest rung of the poor households in rural Bihar in NSS 55^th round survey. In contrast, in MPCE group of Rs. 765+, our survey gets higher proportion of households in comparison to NSS round. So as a whole, the stark contrast of level of consumption gets diluted in NSS survey.

The consumption class in terms of MPCE presents contrasting picture across caste, class and land size categories (Table 4). Nearly one-third of households of scheduled caste and one-fifth of OBC II and Koiri caste households are very poor (0-250). Two-third of all scheduled caste and Koiri caste households are below the poverty line. Across class more than one-third of attached agricultural labour class are very poor and more than four-fifth are below the poverty line.

Across land size category nearly one-third and two-third of landless households are very poor and below the poverty line respectively. In contrast, around one-tenth of households belonging to Brahmin and Kurmi caste, in big peasants and landlord class and landholding size more than 5 acre have MPCE more than Rs. 765. In landsize group, 20 acre or more, two-third of households falls in the MPCE group of Rs. 945+.

A similar distribution based on monthly per capita income (MPCI) shows that three-fifth households belonging to scheduled caste and backward caste I are very poor (0-250 MPCI group) and more than two-third of scheduled caste, backward I and Muslim households are below the poverty line.

Across class and landsize group, the percentage of households from very poor to poor and further to rich show a continuous progressive pattern in terms of social hierarchy and landsize ownership. Half and two-third of households belonging to agricultural labour class and landless size groups are very poor or poor. In contrast, one-sixth or more households belonging to big peasants and landlord class and landsize holding of 2.5 acre and more are rich (MPCI>765). In 20+ acre landsize group two-third of households are rich.

Although over the yeas the poverty ratio has declined, the analysis of poverty in terms of overall consumption expenditure groups shows that, there is still a large concentration of poor towards the lowest rung of the consumption expenditure class. This lowest group of consumption class largely constitutes some of the most vulnerable groups of society such as agricultural labour, landless, poor middle peasants, etc.

To elucidate the contrast evident in total consumption expenditure we present pie charts reflection composition of different items for top 10 percent and bottom 10 percent of households. The mean level of consumption of bottom 10 percent is Rs 189 and that of top 10 percent is Rs. 1507. The breakdown of consumption basket shows that of bottom 10 percent of household more than half of consumption expenditure is incurred by on cereal & cereal substitutes. For the top 10 percent it constitutes only 15 percent. In contrast, more than half of top 10 percent households expenditure is accounted from by miscellaneous good & services constituting small consumer durable, transport, court expenditure, taxes & cess, religious & marriage expenditure and other items & services. In the case of bottom 10 percent it constitutes only 2.89 percent (Fig. I).

Bottom 10 per cent Top 10 per cent

Access to adequate and nutritious food by households has been one of the most important considerations in determining their poverty levels. However, since the consumption basket of the households may differ across the regions and social groups, a general poverty line approach may not be truely indicative of the relative deprivation of the households in terms of access to food. Moreover, the recent NSS estimates of consumption of food items and nutrients indicate that the per capita cereal intake has declined although per capita real expenditure on all food items has increased over the years. This means that a fall in the cereal consumption has been offset by rise in non-cereal consumption. Again, the degree of shift in the consumption basket may differ over the regions and social groups of population depending on the capacity of households to command food. In the following paragraphs we will look at the trends in the per capita consumption of food grains (cereals as well as non-cereals) and non-foodgrains items across different social groups.

Per capita consumption of cereals, foodgrains, and non-foodgrains in the households belong to different castes and class groups are presented in Table 5.

Although the per capita consumption expenditure on cereals (rice and wheat) is slightly higher in case of lower castes and classes, the per capita consumption expenditure on pulses and non-foodgrains show a clear bias in favour of upper castes and classes. As a result the total consumption of and expenditure on all food items taken together is much higher in case of household belonging to higher castes and classes. Besides this, there is comparatively low proportion of consumption expenditure on non-food grain items by lower caste and class households. The qualitative difference in the food consumption expenditure pattern of upper caste and SC households is more clearly presented in Fig II.

As can be seen from the Fig. 6.1 upper caste households spends 41 per cent of food expenditure on main food grain , rice and wheat, while in case of SC households this percentage is as high as 61. Accordingly, the percentage of expenditure on pulses, milk, meat/eggs, fruits etc. is substantialyy higher among upper caste households in comparison to that among SC households.

Since the present data relates to normal agricultural year, the wide variations in total food consumption or consumption expenditure is not related to general shortage of food and highlights inter-group distributional issues more prominently. Obviously during the times of shortages (not necessarily during famines, droughts, or floods but also during slack seasons of a normal year) some groups can suffer absolute starvation.

The consumption pattern of different population groups in terms of castes and class (Table 5) brings out the following points very clearly.

Due to the acute poverty the households face acute food security problems subject to seasonal variations. Even in the absence of natural calamities such as famines, droughts, floods etc. the overall food availability to households is subject to variations depending on the crop cycle, irrigational facilities, cropping patterns etc. Besides, the seasonal variation of food availability is also region specific, which may be associated with the climate or geography of the regions concerned. In our surveyed districts, the months of August to middle of November emerged as the slackest period when more than 40 per cent of the households faced food crisis. The second slack period observed prominently was January to March (Chart 1).

These two periods can be identified as pre-harvest time of kharif and rabi respectively. Besides these, a few households also marked May to July as lean period i.e. at the beginning of agricultural season. Chart 6.1, shows that in different villages, the period of food shortage range form Bhado (August-September) to Kartik (October-November). Starvation period for all villages ranges over two months either from Bhado (August-September) to Ashwar (September-October) or from Ashwin (September-October) to Kartik (October-November). In fact, in all villages, the problem of food security becomes so severe at times that people are forced to depend on a variety of low quality food items like sattu, ghor, wild sag and animal products like local fish, kekra, ghongha, mouse etc. Although, many of these items might be even more nutritious than common food items, but these are generally looked down as food that is of inferior quality and are also not available in sufficient quantity in most of the times.

Besides the regional variations the period of food deficiency also varied for different sections of the population. On the basis of the perception of respondents more than 25 per cent of the households faced food deficiency for one month or more. Besides there are at least two to three per cent of households that are chronically food deficient for more than six months.

The largest concentration of such households is in non-agricultural households. As a whole, however, most of the households hit by food deficiency are faced with food shortage for 1 to 3 months in a year.

Hence, although seasonal variations in food availability have a wide ranging effect on population they affect the poor more severely even to the extent of causing irreversible damage to their assets and human resource. In such a situation seasonality also has a lagged effect on households even during the non-slack periods.

3. Changes in food security over time

Although the food availability situation is heavily biased towards upper caste and class, the overall food availability has increased over the years. This is reflected in the perception of none other than villagers themselves responding to the questions. "Do you get enough food round the year?"—the percentage of households from different land owning size and class during the two referred survey periods (1981-82 and 1999-2000) who said that they were not getting enough food is presented in Table 7.

It can be seen from Table 7 that the percentage of households "not getting enough food" round the year has come down to 26 in 1999-200 from 68 in 1981-82. Today there is no household with more than 5 acres of land which faces any kind of food deficiency. Accordingly very few households from middle peasant, big peasant, and landlord class report that they are "not getting enough food". However, the situation in regard to food availability in the case of landless and agricultural labour households is still grim through has been some improvement. Approximately 60 per cent of these households find that they do not get enough food round the year. The position of non-agricultural class is same as that of poor middle peasants.

The overall improvement in the food scenario is largely the outcome of the overall general improvement in the availability of main food grains such as rice and wheat. The consumption of inferior grains such as millets, jowar, bazra, etc. has been almost completely replaced by the consumption of rice and wheat, which is now almost universalised. However, there has been little improvement in consumption of non-foodgrain items such as pulses, vegitables, milk, eggs/meat, fish, fruits etc. In order to observes the changes in the dietary pattern in 1999-2000 over 1981-82, we have calculated number of days of consumption of different items in a month in the year 1999-2000 and 1981-82. Since, apart from seasonal variations, availability of different food items is subject to regional variation, we have calculated these figures at the village level. Average number of days of consumption of food items in a month for the two periods are presented in Table 8. Since the consumption of rice and wheat is completely complement each other, we have not separated these two items, and the average figure shows consumption of these two items taken together.

Table 8 reveals a marked improvement in terms of consumption of rice and wheat. The average number of days of consumption of rice and wheat has improved from 33 days in 1981-82 to 57 days in 1999-200. This implies that now the households are able to consume rice and wheat almost daily. Also, the regional variations in terms of number of days of consumption has drastically fallen in 1999-2000 compared to that in 1981-82. Consumption of pulses and vegetables has also improved slightly during the intervening period. The consumption of milk, which was very low in 1981-02, as the average number of days of consumption was only 3 in a month has now improved to 13 days in a month. However, increase in the consumption of these quality food items has been largely limited to upper castes and class of the households. Increased animal husbandry activities and improved diary farming over the years has contributed to an increase in the consumption of milk. Consumption of many other quality food items such as eggs, meat, fish, fruits etc., however, show either very little improvement or even deterioration during the intervening period.

Again the class-wise pattern shows considerable venation inequality in the improvement of food quality. Whole households belonging to lower castes and class have improved cereals availability they do not show much improvements in terms of quality food items. Since in the case of lower caste and class households, one of the major source of some of the quality food items such as fish, fruits, etc. has been forest, ponds, and other commons, fast depletion of these sources over the years has resulted in to deterioration in the consumption of these items over the years. Today the consumption of the quality items is largely limited to some upper classes of the households. Though, as is clear from the table, even among the lower classes the consumption basket has widened to include variety of food items..

Skipping of meals has been one of the most important coping mechanisms of households particularly in a poor economy. Although there is no normative number of meals per day, the data collected on a 'number of meals taken per day' is indicative of two important aspects: one, this give an idea of intra-household variations in the frequency of meals taken by males, females, and children, and second it also indicates the extent of variations in the availability of meals across different social groups such as caste and class. Table 7 shows the distribution of persons skipping meal at the time of food scarcity.

Notably, as the table shows, adult males hardly skip any meal. In contrast, in nearly three-fourth of all surveyed household adult females skip meals at the time of scarcity. Across castes, only in backward caste I, some adult males skip meal at the time of scarcity, while those among Yadav, Koiri and Kurmi castes do not do so at all. There are some variations in incidence of adult females skipping meals, but when other females are added to them, then castewise disparity get reduced by substantial margin.

Asset holding position of household is an important indicator of poverty. For the assessment, we can classify the Households' total assets in two parts—'productive' and 'non-productive'. Productive assets may include all kinds of modern and traditional agricultural machinery, transport vehicles, carts, livestock, land etc. However, For the purpose of clarity we have treated certain productive assets such as House, livestock, and land separately, because of their specific roles in rural areas.

All kinds of non-productive assets have been clubbed together under consumer durable assets. The per capita value of these assets has been presented in Table 9. At per capita level, as housing constitutes more than four-fifth of total value of assets excluding land. Productive assets and livestock together constitute hardly one-tenth of total assets -- both being almost equal.

Across castes, assets per capita among Kurmi are far higher than that of other castes as the components of the assets of the farmer, except livestock is far above that in the case of this rest of the castes. In regard to the possession of livestock, Yadav score over the other castes although in terms of per capita total assets they fall far below the forward castes (Brahmin and Bhumihar) and landed backward castes, Kurmi and Koiri. In respect of class and land categories the assets per capita increase progressively as one moves up in the social hierarchy and land sizes respectively. But per capita assets of big peasants are marginally higher than those of landlords by virtue of their higher level of productive asset. The value of per capita non-productive assets which is largely consists or consumer items, is the highest among the landlord class, which is inductive of high social status it enjoy in society. In respect of the land holding categories, percapita assets of households with landholding size of 20 acres and above, much higher than the rest of all land size groups.

House property constitutes more than four-fifth of all asset per capita (except land) in rural Bihar as a whole. It would be interesting to examine the proportional breakdown of moveable assets (excluding land and house assets) in productive and non-productive categories, which the latter constitute nearly three-fifth of all assets, and the former in a few categories constitute more than half of total moveable assets.

Possession of different types of domestic consumer durable is not only an indicator of quality of life but also it may be taken as directly indicative of the level of poverty of a household. In fact, in the absence of proper income data, possession of domestic consumer durable gives an idea about the living standard of the households. In Table 10 the percentage of households possessing different types of these consumer durable has been presented. The table shows that apart from the ownership of cot (Charpoy and palang), the ownership of lantern/petromax, mosquito net, torch, and watch is widespread. Besides these, there is a substantial proportion of households own some other consumer durable such as bycycle, chair/table, radio, etc. However, possession of these durables shows considerable variation over different regions and class of households. Some of the rich villages viz. Paharpur Dayal in Gopalganj and both villages in Rohtas show better asset holding position in comparison to that in other villages. Accordingly, almost all big farmer and upper class households possess items like cot, chair and table, torch, and watch, while only a small proportion of small farmers and lower class households possess these items.