Welvaartsfuncties in de landbouw : ontwikkeling en toepassing

D. Kingma

Research output: Thesisinternal PhD, WU


An individual's well-being (also referred to as welfare or utility) is intimately linked to the extent to which the desires and wishes of that individual are fulfilled, and pain and distress avoided. These desires can relate to various aspects of daily life. Some pertain to economics, others are more sociological in nature. The economic aspect that has received the most attention is income. In this respect income has a very broad meaning: command over commodities. It covers elements such as income distribution, wealth and its distribution, poverty, etc. Closely related to income and income distribution is the division of time between labour and leisure (in some instances also time spent in the household). To a large extent the distribution of time between labour (paid time) and leisure (unpaid time) determines the amount of income an individual can earn, In addition, the choice of a particular job (at least if one has the option of choosing) also has implications regarding the amount of income. Some aspects of life usually considered as belonging to the field of sociology such as health, education, safety, social relations and feelings of fear, have nonetheless received an increasing interest of economists (Becker, 1965; Rosen, 1986).

This research concerns income and labour in agriculture. Literature on agricultural economics devides income research into three general categories:
-income formation and income distribution in agriculture
-comparison of incomes earned inside and outside agriculture - research on the incidence of poverty in rural areas.
Research on labour is mainly concerned with:
-the quality of jobs: working conditions, education, on-the job-training, etc.
-the amount of work done: for example, how many hours a day does a farmer work.

In this study a special kind of utility function (Van Praag, 1968) is used to describe farm families' preferences with respect to income and labour. This socalled individual welfare function is a cardinally measurable utility function. This function was originally derived as a lognormal distribution function. Up to now most applications of this special utility function have been aimed at deriving the welfare function of income.

The univariate lognormal distribution function is determined by two parameters, μand σ. In order to measure the welfare function of income, that is, to estimate the parameters μand σthe 'Income Evaluation Question' was developed. Under the assumption that utility can be measured directly, respondents to surveys have to specify the amounts of income they associate with verbal qualifications such as 'very bad', 'bad', etc. The parameters of the welfare function can be estimated using those evaluations, making some special assumption about the way those surveyed provide these evaluations (the so-called 'equal interval' assumption). Differences in estimated parameter values can be explained by the differences in economic and demographic background among the people being surveyed.

The welfare function of labour is measured in a way that is analogous to the welfare function of income. In the labour evaluation, respondents express their feelings about two different, yet related aspects of labour. First, working on the farm provides 'psychic income' for the workers involved. Therefore, to a certain extent, labour will be evaluated positively. Second, the provision of labour reduces the amount of leisure time available. This aspect of labour provision is evaluated negatively. According to the theory of the welfare function, the evaluation of both aspects of labour can be approximated by a bivariate lognormal distribution function (Van Praag, 1968; Van de Stadt, 1983). The four parameters of this specification are measured similarly to the welfare function of income. Survey respondents are asked to answer the 'Labour Evaluation Question', which indicates a range of hypothetical working hours. The respondent is supposed to give a (numerical) evaluation for the various working hours relevant to him or her. As the bivariate lognormal distribution function is rather demanding as far as the number of parameters to be estimated is concerned, additional specifications such as a gamma and a beta specification were also estimated. Just as with the welfare function of income, differences in parameters can partly be attributed to differences in the respondents' economic and demographic circumstances.

Both the welfare function of income and that of labour should be considered partial welfare functions. When evaluating working hours, income is assumed to be constant, whereas the amount of working hours is constant when evaluating different amounts of income. According to the theory of the welfare function, the simultaneous evaluation of income and labour should be specified as the product of the partial welfare functions (at least whenever the aspects evaluated, in this case income and labour, are independent). It will probably be very difficult for a respondent to evaluate an amount of income that is very high (or very low), compared with that respondent's usual income, without (even if implicitly) changing the amount of labour involved. The same is true when evaluating labour. Respondents might relate different amounts of working time to changes in income. This means that any simultaneous evaluation of income and labour based on such partial evaluations can only be valid when conducted within the scope of respondents' own actual labourlincome combinations. In addition to the (theoretically correct) specification as a product of partial welfare functions, the simultaneous evaluation of income and labour has also been specified as the (convex) weighted sum of the partial evaluations.

The parameters of the welfare function are estimated using survey results. The answers to the specific evaluation questions (the 'Income Evaluation Question' and the 'Labour Evaluation Question') are particularly relevant in the estimation procedure. For the purpose of estimating the welfare function of farm families in the Netherlands, a small sample of farmers (arable, dairy, intensivelivestock, and others) was chosen from among the Dutch farm population. In addition to answering the evaluation questions, respondents were also requested to provide information about the technical characteristics of their farms and the demographic characteristics of their farm families. Contrary to most previous applications, the evaluation questions were put to both the farmers and their wifes (if present). Although the sample used is certainly not a perfect reflection of the Dutch farm population, the farms selected seem to have provided a reasonable picture of farm type subgroups within the population.

As far as the preferences regarding income are concerned, the results broadly correspond to those for the Dutch population. (For results concerning the population at large, see, e.g. Kapteyn, 1977). It generally takes a higher income to satisfy a farmer than it does to satisfy his spouse. This fact has been explained in Kapteyn et al., 1986.
Past incomes do affect perceived satisfaction with certain amounts of income in the present. This is the so-called 'preference drift'. Because of the highly fluctuating incomes of farmers, this effect is less pronounced for the farm population than for the average Dutch citizen. Using some kind of permanent income measure that fluctuates less than actual farm incomes, substantially reduces the differences between farmers and nonfarmers.
The number of persons in a family can be considered an indication of the costs involved in feeding and clothing the family. Although family size does have a comparable effect on the farmer's income preferences, as in studies covering the Dutch population, family size has a much greater effect on the preferences of the farmer's wife. This is due to the fact that previous research only addressed the income evaluation question to the head of the family, in most cases a man.
The welfare function of income can be used to calculate parity incomes for different groups of farmers. Group A earns a parity income compared with group B when both groups are equally satisfied with their incomes. Based upon the (small) sample, intensive-livestock farmers seem to be the most satisfied with their incomes, whereas arable farmers are the least satisfied. In general, farmers seem to be less satisfied with their incomes than the average Dutch citizen. It should be kept in mind however, that this observation is related to the period of the survey. In different circumstances the comparison result could very well be reversed.

The evaluation of labour is not monotonically decreasing in time, as is almost always assumed in neoclassical economics. If only a small number of hours is worked on the farm, the perceived sense of well-being increases. If the amount of labour exceeds a certain number of hours, perceived welfare decreases. The increase in perceived welfare is the result of 'psychic income' derived by farmers working their farms. The welfare perception eventually starts decreasing as the amount of leisure time is reduced due to increasing labour time. Various specifications of the welfare function of labour have been estimated. The lognormal specification fit the data best.
As with the parameters of the welfare function of income, habit formation determines to some degree the estimated values of the parameters of the welfare function of labour.

The combination of the welfare function of income and the welfare function of labour results in the welfare function of income and labour. If both partial evaluations are independent, then the combined welfare function is the product of the partial welfare functions. The cardinality of the welfare function makes it possible to determine the unique welfare levels of the indifference curves.
The shape of the indifference curves is different than in the neoclassical case because of the 'psychic income' derived from working, whether it is on the farm or elsewhere.
The amount of income needed to persuade a farmer to work an extra hour (the so- called marginal rate of transformation) can be calculated from the estimated welfare function of income and labour. The interpretation of the calculated values is rather difficult. Compared with the values in other research (based on neoclassical assumptions), the calculated rates of transformation based on the welfare- function are very high.

Given that farmers would have the same preferences regarding on-farm and off-farm labour, calculations based on neoclassical assumptions show that the on-farm amount of labour for individual farmers is not compatible with utilitymaximizing behaviour. Because farmers are working too many hours, their wage rate, measured by the value of the marginal product, is low. Nonfarming wage rates in 'comparable' jobs are higher. Here farmers could earn a higher income with presumably less labour. One reason why farmers do not leave agriculture is that in the above comparison an incorrect value for the opportunity cost of farm labour is used in calculating off-farm income opportunities. Nonagricultural wage rates for current farmers would be even lower than their agricultural wage rates. There is also the chance that the right kind of jobs (part-time, close to the farm) are not available. Another explanation is that the income measure is too restricted. In addition to providing money, farming also supplies other kinds of income (e.g. housing, home- grown goods, etc.). From this point of view, farmers' incomes would not be (relatively) low as long as all components were adequately assessed. Another explanation takes preferences towards labour into account. It could be that farmers have such preferences (with respect to farm labour), that they are satisfied with a relatively low reward.

The results based on the welfare function of income do not confirm explanations concerning income, as such. Farmers are less satisfied with their money income than workers outside agriculture. Obviously non-money income does not fully compensate for the low money reward. When preferences regarding labour and income are simultaneously taken into account, in order to justify an additional hour of work, the wage rate should be substantially higher than the marginal wage rate calculated in previous (neoclassical) research. So, even when preferences towards farm labour are taken into account, it is still impossible to explain the low wage rates in agriculture.

The high values for farmers' transformation rates, based on their preferences regarding income and labour, do not fit with the incidence of low agricultural wage rates. One explanation could be that the welfare function should contain other arguments besides income and labour. Or perhaps attention should be paid to nonagricultural circumstances in order to discover why farmers are so reluctant to leave agriculture. Despite the fact that circumstances in agriculture do not concur with farmers' preferences, farmers seem to be trapped in agriculture without any opportunities for leaving and taking a job elsewhere. This inquiry into the preferences of farm families cannot, by itself, provide the ultimate answer to the question of why farmers remain farmers.

Original languageDutch
QualificationDoctor of Philosophy
Awarding Institution
  • Oskam, A.J., Promotor, External person
  • Kapteyn, A., Promotor, External person
Award date23 May 1995
Place of PublicationS.l.
Print ISBNs9789054854029
Publication statusPublished - 1995


  • agriculture
  • agricultural policy
  • income
  • farm income
  • human population
  • poverty
  • family farms
  • agricultural economics
  • netherlands
  • property
  • income policy
  • wealth distribution
  • farm labour

Cite this