Conclusion

View the most recent version.

In this paper, we examine literacy both from the perspective that it is an important entity in its own right and in terms of its impact on earnings. We find that literacy is considerably more equally distributed than income and that the level of literacy inequality has decreased between the time of the 1994 literacy survey and the more recent, 2003, survey. A closer examination reveals that there have been improvements in literacy at the low end of the literacy distribution but deterioration at the top end. The improvements partly reflect improved literacy among high school drop outs but appear mainly to result from improvements in the education level of the population. The declines at the top end are evident among high school graduates, university graduates and graduates of non-university post-secondary educated institutions. We find that successive birth cohorts have had poorer literacy outcomes at the top of the distribution, potentially pointing to an education system that is doing better for those at the low end but doing a poorer job of generating literacy for those at the top. We also find that literacy tends to decline with age after leaving formal schooling and that schooling itself is the prime driver of literacy. Parental education is also strongly related to literacy, but in an interesting way: having parents who are high school drop outs has a strong negative impact on literacy but there is almost no added impact from parental education rising above high school graduation. There is some evidence of a correlation of literacy with the use of literacy skills at work, suggesting (along with the declines with age) a "use it or lose it" model of post-schooling literacy, but these effects are not large.

In terms of earnings generation, literacy plays a substantial role. According to our Ordinary Least Squares estimates, a 25 point increase in the average literacy score (the equivalent of about 1/2 of a standard deviation in the literacy score distribution) has an impact equivalent to an extra year of schooling. When we introduce the literacy variable into a standard earnings regression, the coefficient on the schooling variable is reduced by approximately 20%, implying that about one-fifth of the typically measured impact of schooling on earnings arises because schooling generates higher levels of literacy. Our instrumental variables results seem to imply a more substantial role than this for literacy in accounting for returns to schooling but we need more investigation to be sure of this result. The fact that the coefficient on schooling remains strong and positive indicates, further, that schooling also affects earnings through its impact on non-cognitive traits such as leadership skills and reliable attendance at work.