Accueil » 30-2 ( 1975) » Facteurs affectant l’offre de travail : les personnes à faible revenu

Facteurs affectant l’offre de travail : les personnes à faible revenu

Bertrand Belzile et Viateur Larouche


Le présent article rend compte en partie d'une recherche visant à évaluer l'impact sur l'offre de travail des parents de familles à faible revenu des régimes publics de sécurité du revenu au Québec. Dans un premier temps, les auteurs font une revue de la littérature sur les taux d'activité, en insistant plus spécialement sur les principaux facteurs qui ont été utilisés dans quelques recherches empiriques sur le sujet. (Ce dernier examen pivote autour d'une excellente opérationnalisation faite récemment par Bowen et Finegan de l'analyse traditionnelle de l'offre du travail.) Dans un deuxième temps, ils s'attardent à quelques facteurs susceptibles d'affecter plus spécifiquement les parents de familles à faible revenu. Ils suggèrent d'abord une définition opérationnelle de chacun de ces facteurs, puis une mesure possible dans le cadre d'une enquête sur le terrain. Enfin, ils formulent l'hypothèse attendue entre chaque facteur et le taux d'activité. En conclusion, ils rapportent quelques résultats de recherches empiriques relativement aux effets de revenu et de substitution.


Analysis of Factors Capable of Affecting Labour Supply : The Case of Parents of Low-Income Families.

The present article presents in part a progress report of a research which tries to estimate the impact of public programs of economic security in Québec on the work effort of parents of low-income families. Our interest in such a research comes from the recent but increasing effort of our society to fight poverty, and more particularly from the passing of the Social Aid Act by the Québec Government in 1969. The question of incentive to work takes a particular importance, since, according to this Act, benefits to families are calculated as the difference between income and needs, whatever the labour force participation of their members.


The essential of the traditional analysis of labour supply is retained in empirical studies of labour supply, often called studies of labour force participation rate : the main emphasis is on the estimate of the income and substitution effects. Theoretically, the income effect is negative, whereas the substitution one is positive. But, the net effect is an empirical question.

Bowen and Finegan have published, in 1969, a quite interesting operationalisation of the traditional analysis of labour supply. They have identified four (4) classes of variables as influencing the decision of the family in the allocation of its time. The classes are : tastes, expected market earnings rates, expected non-market earnings rates, and family resources. It is important to underline the third use of time introduced through non-market earnings and the fact that the family is considered as the relevant decision-making unit.

Before proceeding to a brief analysis of the main factors used in many empirical studies, it is worth mentionning the model published by Cain, in 1966, to explain the labour force participation rate of married women. The variables in his general model were : potential family income, market wage for the wife, home wage for the wife, market wage of other family members, home wage of other family members, wife's tastes for market work relative to homework and leisure, and an error term. For empirical studies of disaggregated data for cross-sections, Cain proposed a model including the following variables : permanent income, transitory income, market earnings for the wife, presence of children, education and age of the wife. The variables examined in the third section of the present article have been in large part suggested by this last model.

For the present research, the low-income family is defined as one with total annual income under 150% of maximum Social Aid plus family allowances. A pilot field survey has been done mainly to validate the research instruments. But, no results are presented here, since the sample used was not representative of the population, among other reasons.


The effort here is to interpret the factors used in many empirical studies in terms of the four classes of variables reported above. For instance, it seems very significant to identify the variables used as proxies and also to investigate the rational for their use. Such an investigation may greatly help to specify a particular model as well as to interpret empirical results. The proxy variables examined here are : education, age, number and age of children, sex, marital status, work status, occupation, health, ethnic origin, residence and housing. The other variables covered are direct ones, that is they are not used as indirect measures for other variables. But proxies may have to be used for direct variables. For instance, tastes for market work have to be approximated by other variables, such as education, age, residence, health, housing, etc...

The variable « other family incomes » is defined as the total family income minus the income of the parent concerned. The variable gives a very good estimate of the pure income effect that is negative on the work effort of a person.

The expected market earnings rate is certainly the most influential variable. Theoretically at least, the relation is positive with the labour supply. In a field survey, one can use the actual rate as an excellent proxy for the expected one, of course, for the working people. For those not working, they can be asked what their perceptions are. The same is more or less true for the other direct variable, expected non-market earning rate. The last direct variable here, tastes for market work, has already been discussed.

Education is very often used as a proxy variable in studies of labour force participation rate. It certainly represents an excellent approximation for personal job prospects and for tastes for market work. More or less the same comments can be made about age, except that in addition to education, age affects labour supply through the family cycle, or in other words the number and age of children. This last variable serves as a good proxy for tastes for home goods as well as for monetary income. The rational for the use of health and sex is somehow similar as for the use of the three proceeding proxies. In a similar vein, this is more or less true for their effect on work effort.

Of course, marital status also is generally used in empirical studies of labour supply. Moreover, most of such studies and probably the best of them, have been made for married women. It is evident that their expected home wage rate generally is far greater than the one for their husbands. The same can also be said about tastes for home tasks. The husband's (wife's) work status is another proxy in large part related to the marital status, and it affects the labour supply through the other family incomes.

The occupation of a person and her ethnic origin serve as good approximations for tastes for market work and, among others, for personal job prospects. Residence affects work effort through proximity to job opportunities. The last proxy considered here, housing, is ambiguous, since it can affect the tastes for monetary income as well as being affected by it.


We briefly examine thereafter some factors that we consider more particularly relevant for an empirical study trying to explain the work effort of parents of low-income families, using disaggregated data collected for cross-sections in a field survey. Some of the variables have been suggested by the model used by Gain and mentionned above.

For the sake of brevity, and taking into account what has been said earlier, it seems sufficient only to mention here the following variables : expected market earnings rate, expected home earnings rate and other family incomes.

Transitory income calls for a short additional comment. It is defined as the difference between permanent income and actual income for a family, the permanent income being Friedman's concept presented earlier.

The last three factors considered here deal with the public programs of economic security. Two distinctions are made in this respect. First, the benefits received by a parent are taken separately from those by his family, since we feel that the strength of the relationship differs from one case to the other. Second, we consider apart the social aid from the other programs, mainly because of its nature of last resort program. Moreover, we operationalize this last variable by defining it as the difference between the total benefits his family would have received from social aid if the parent concerned had not been working at all, during the last year, and the benefits his family would have received if the same parent had been working full time, during the same period.


Given the economists' interest for the income and substitution effects, we report a few results from empirical studies. Mincer found that for married women the substitution effect in terms of elasticity was twice larger than the income effect. Cain confirmed such results. Rowlatt recently reported interesting results for persons living in low-income families in Alberta, substantiating the proceeding ones.