• Print •  • E-mail Alerts • Search Articles

1996 Annual Report Essay

Breaking Down the Barriers to Technological Growth

Another view about economic growth

As we mention in the accompanying essay, there are some economists who do not place such emphasis on how institutions influence the incentives to use and adapt world knowledge. For example, in a recent article, N. Gregory Mankiw argues that the state of technology plays a minor role, if any, in explaining differences in labor productivity across countries. He states, “Put simply, most international differences in living standards [author's note: labor productivity] can be explained by differences in accumulation of both human and physical capital” (1995, p. 295). (See also Chari, Kehoe and McGrattan 1996.)

What evidence does Mankiw have for his claim? His evidence is from his work with Romer and Weil (Mankiw, Romer and Weil (MRW) 1992). MRW perform an accounting exercise, like that by Denison (1967) decades earlier, but find that nearly all differences in labor productivity are due to differences in capital, both physical and human.

We are not persuaded by the MRW evidence due to two types of considerations. One is a critique of the MRW study itself. The other is a demonstration that counterfactual implications follow from the assumption that all countries have the same technologies.

First, as Klenow and Rodriguez-Clare (KRC) show, the MRW result is fairly fragile. For example, KRC take exceptions to the measure of human capital employed by MRW. One objection is that MRW use secondary school enrollment rates to construct their measure of a human capital stock for each country. A more comprehensive measure would use, for example, primary and post-secondary school enrollment rates to construct stocks. Primary enrollment rates, not surprisingly, vary much less than secondary, so a more comprehensive measure of human capital turns out to vary much less across countries than does MRW's. Hence, human capital is less likely to have a big role in explaining differences in labor productivity. And this is what KRC find. That is, when KRC use the expanded measure of human capital, they find that differences in technology once again play a big role in cross-country productivity differences.

KRC have other problems with MRW's treatment of human capital. In particular, they object to the nature of the accumulation technology for human capital. Fixing this accumulation technology adds still more to the role of the state of technology, and less to capital, in explaining labor productivity differences across countries. After these changes in the treatment of human capital, KRC are essentially back to the old view about the key role of the state of technology.

A second consideration for why we don't accept MRW's findings is based on recent work of Prescott (1996). He supposes that all countries use the same state of technology and asks whether differences in inputs of physical and human capital can, in plausibly calibrated growth models, account for differences in cross-country labor productivity levels. He finds that they cannot. Furthermore, he argues that such models have two other counterfactual implications:

    1. They imply capital-output ratios should be relatively higher in rich countries than in poor ones. Prescott argues that evidence, such as Kuznets (1967), suggests, if anything, the opposite is true.
    2. They imply that the real rate of return on capital in rich countries should be lower than that in poorer countries because richer countries have relatively more capital. Prescott argues that there appear to be only small differences in real rates of return between rich and poor countries.

Top