Income is considered as one of the most important determinants of livelihood in a society. Â But it must also be admitted that ignoring distributional issues, increasing per capita income alone does not ensure welfare for all. It is mainly because level and sources of household income are responsible for existing differential livelihoods. However, estimation of such income, especially in rural areas, is always very complex and a hard task. The reasons are not far to seek. There is hardly any record-keeping system for inputs purchased and output produced in the vast spectrum of informal transactions in rural areas. Most of the income-related questions are answered from memory. Again, advertently or inadvertently, the general tendency to under-report income makes it all the more difficult job to arrive at a correct estimate. Besides, sometimes the 'guess-based' accounts of income given can not be cross-checked due to lack of alternative information. On the other hand, the most difficult problem arises when households' own production is never considered as an income for the household. Â The same happens with the in-kind income that households receive. And only for these reasons, estimates of rural income remain noisy and especially, the balance is tilted towards undervaluation.
In the estimation of household income as used in this article, we have adopted a different technique to minimise errors and omissions. Â For example, (a) receipts in kind or in cash, have been treated as income. In this case, money value is imputed to in-kind earnings, and the value represents the prevailing prices in survey areas; (b) households' self-produced products and by-products, such as crop, livestock and poultry, fisheries and forest products are bracketed as income on the assumption that expenditure saved is also income earned and, (c) the income from crops has been estimated by the value of the main product and by-product. The value of inputs used has been deducted to arrive at net income of the household. Prices of inputs supplied by the household are imputed at the prevailing prices in the market.
However, still a few limitations remain to be resolved. For example, no allowances could be given to the depreciation of fixed assets and for the owned house. And as we all know, there is an acute dearth of information on these counts in rural Bangladesh. Â Again, receipts from sale of assets and loans are also not included as income since there is always a possibility of very high or very low reporting on them. We admit that the government services, such as non-market transfers received from health and education, should have been included in income. Unfortunately, we fail to collect information on these items. And finally, we have expressed household income in terms of US dollars and for that the exchange rate of the year of data collection is used.
The income earning capacity of sample households would obviously depend on a number of factors. For example, land is the most important asset which plays an important role in influencing household income. It may be owned land or rented-in land. Likewise, income depends on the number of working members within the household as well as on non-land assets used in production. Again, land availability is not sufficient for income; soil type, topography, depth of flooding, use of modern seeds and improved farm management practices also affect income from land. Tenancy arrangement would raise income only when income derived from tenancy cultivation is higher than the opportunity costs of labour and working capital used in land under tenancy. The productive power of the family labour depends on education and training levels of the labour. Besides, the development of infrastructure increases the price of produced output at household level, reduces input prices and helps market orientation.Â
Noticeably also, despite a reduction in the number of agricultural workers in the household, the contribution of workers in influencing household income has increased. This means, productive power or productivity of agricultural labour goes up over time. But, at the same time, the contribution from non-agricultural workers almost doubled from about 13 per cent in 1988 to 23 per cent in recent times. Â By and large, the contribution of labour in household income now accounts for roughly 40 per cent compared to only 25 per cent in earlier period. The policy implication of this change points to the need for enhancing access to education, training and health for workers as these would raise their productivity. Besides that, employment generation also requires expansion of infrastructural facilities.
In changing household income, the contribution of schooling marginally declines. Undoubtedly, it is a cause for concern as education to us is one of important determinants of household income. But we reckon, possibly the impact of education is felt the other way. Â The impact might have been through migrating members whose contribution increased four times. In other words, in educated households, the number of migrating members is large; so income is also higher. However, agricultural and non-agricultural capital of households has increased overtime and their contribution also has increased. And as argued before, remittance income rose its share as migrant members of the households increased over the years. The contribution of access to electricity is apparent from the fact that the access increased its share from about 5.0 per cent to 7.0 per cent overtime. It should also be mentioned that 4 per cent of the household income in rural areas flow from NGOs.
The writer is Professor of Economics at  Jahangirnagar University.
 abdulbayes@yahoo.com
Â