Empirical analysis of literacy acquisition and the effect of literacy on labour market outcomes has generally been limited to the study of immigrant populations (cf. Kossoudji 1988; Rivera-Batiz 1990, 1996; Chiswick and Miller 1992, 1995; Charette and Meng 1994). These authors have focused on the ability of immigrants to acquire the dominant language(s) of their host country and the effect of dominant language acquisition on earnings. A related literature examines language .and earnings among Hispanics in the United States (cf. Bloom and Grenier 1994; Grenier 1984; McManus 1990). Although most of the latter studies are concerned with the labour market performance of native-born. Spanish-speaking Americans, several also considered Spanish-speaking immigrants. There are a number of Canadian studies that examine the role of 'mother tongue,' 'home language,' or ethnic group (British or French) in the explanation of earnings differentials (cf. Carliner 1981; Chiswick and Miller 1988; Robinson 1988; Shapiro and Stelener 1987). The authors of these studies generally examine the earnings differences across language groups, but not the role of degree of literacy in the given languages.

While motivated by the literature on immigrants, in the present study we make a number of advances over previous work on literacy. First, and most important, the data set used makes available an objective measure of reading ability. Previous studies have generally relied on self-assessed literacy scores, similar to those reported in the American or Canadian census. Generally, these self-assessed measures are limited to speaking ability.¹ Charette and Meng (1994) have shown that there are potentially serious measurement errors in self-reported literacy variables. Second, the data set used in this study also includes a numeracy test score. Little research has been conducted on the determinants of numeracy or its role as a component of human capital. Third, the model specification includes a number of socio-economic background variables not found in other studies. In particular, we are able to control for factors such as parental education and immigration status, and childhood learning disabilities. Fourth, we consider a sample of native-born Canadians. Finally, except for the work of Rivera-Batiz, previous studies of literacy have tended to focus on males. Men and women might acquire language and numeracy skills differently, while literacy and numeracy might affect the labour market activity of men and women differently.

Anticipating the empirical results reported below, we find a differing role for the formal education system in the literacy and numeracy acquisition of males and females. Both literacy and numeracy play a significant role in expressions explaining labour market outcomes. even after years of schooling are accounted for. Male and female experiences also differ with respect to the impact of literacy and numeracy on labour market outcomes. Our results indicate the potential for serious omitted variable and measurement error bias in parameter estimates on years of schooling terms in specifications that do not control for a wider set of human capital characteristics and the quality of years of schooling.

¹ Rivera-Batiz (1990, 1992) uses a test score similar to the score used in this study for a sample limited to individuals 21 to 25 years of age. Ishikawa (1993) makes use of a similar test score in processing the impact of literacy on the labour market activity of young, unemployed workers in United States.