3.3  Growth in the working-age population and growth in labour productivity

The previous section discussed the working age population’s proportional share; this section discusses its absolute size. Many economists argue that the size of the working age population can affect labour productivity, the term in Equation 1.

Changes in the size of the working age population are shown in Table 3. In all countries and regions, growth slows over time. The projected values for sizes and growth rates are quite sensitive to variations in assumptions about migration, and, over the long run, assumptions about fertility. Given these uncertainties, the most sensible way to interpret the table is probably that New Zealand, Australia, the RDCs, and North America belong to one relatively fast-growing group, while Japan and Europe belong to another relatively slow-growing group.

Source: Calculated from data from the UN Population Division’s World Population Prospects online database.

Many models of endogenous economic growth and of agglomeration effects posit a positive relationship between population size or growth and labour productivity. “Population” is used implicitly to mean “working age population”, so these models may shed some light on the economic implications of the trends in working age population apparent in Table 3.

Some endogenous growth models assume that labour productivity is positively related to population size or growth simply because more people means more ideas (Romer 2001). These models are sometimes cited in discussions of population ageing and countries’ relative economic prospects. But it is not clear that such models are in fact relevant. After setting up one such model, for instance, Jones (1997: 21) warns that “as with many ideas-based growth models, this is a model of growth for the world economy as a whole…The Belgian economy does not grow solely or even primarily because of ideas invented by Belgians, so the model does not predict that Belgium’s per capita growth rate should be related to its population growth rate.’

There are other, more plausible, mechanisms through which increased population size could raise a country’s productivity. An increase in the pool of workers permits increased specialization, thicker markets, and greater potential for knowledge spillovers. Such effects may be particularly salient for small, distant economies such as New Zealand {Treasury, 2002 #3231}.

However, not all the proposed effects of growth in the working age population are necessarily favourable. Economists have long argued that if labour becomes relatively abundant, its price falls, and employers have less incentive to provide workers with capital or new labour-saving technologies. If labour becomes relatively scarce, the converse is true. A notable formal model of this process has been developed by Romer (1990). In this model, a rise in the supply of labour draws human capital from away from the production of research into the production of final goods, which lowers the rate of technical change and hence of economic growth.

There is a large literature investigating the empirical applicability of these models.[5] The remainder of this section looks briefly at some of the macro-level studies of the relationship between growth in the working age population and growth in per capita productivity or incomes.

Several studies focussing directly on the relationship between growth in productivity and growth in the working age population have found the relationship to be negative. Romer (1990: 354-7) graphs movements in the working-age population and output per hour worked in the United States over the period 1839-1979 to show that, over periods of 20-40 years, increases in the growth rate of the working-age population seem to induce a one-to-one reduction in the growth rate of labour productivity. He also points out that the incomes of US applied scientists and engineers declined relative to the incomes of doctors, lawyers, and managers during the period of rapid growth in the working-age population in the 1970s and 1980s, just as his model would predict. Little and Triest (2002: 145-9) carry out a simple regression analysis on US data for the period 1904-1999, and find a similar effect from growth in the population in the working ages to growth in labour productivity and multifactor productivity. Cutler, Poterba, Sheiner, and Summers (1990: 38-45) carry out a regression with fixed country effects on data for 29 high-income countries in the period 1960-85. The relationship between growth in the working-age population and growth in labour productivity varies across specifications, but under all except one specification an increase in the growth rate of the working-age population of 1% is associated with a decrease in the growth rate of labour productivity of between 0% and 1%.

Findings such as these imply that the relationship between growth in labour productivity and growth in the working age population can be approximated by an equation of the form

(3)    

where and are constants, and measures the extent to which growth in the working age population depresses growth in labour productivity. Substituting Equation 3 and the identity into Equation 2 gives

(4)    

This equation implies that if the productivity-depressing effects of growth in the working age are large, then regressions of growth in per capita GDP on growth in the working age and total population should obtain coefficients for the working age population that are close to zero or negative.

This is not what happens. In their influential study of demography and economic growth in East Asia between 1965 and 1990, Bloom and Williamson (1998) regress growth rates for GDP per capita on growth rates for the working age and total population, and on other possible determinants of economic growth. Under a range of specifications, the coefficient on growth in the working age population is slightly higher than 1. This suggests that growth in the working age population did nothing to depress productivity growth. The relevance to the OECD of results from developing countries could perhaps be questioned. Brander and Dowrick (1994), however, carry out a similar analysis on data for 67 middle and high income countries over the period 1960-1985. In their preferred specification, the coefficient on growth in the working age population is also slightly higher than 1 (Brander and Dowrick 1994: Table 7).

It seems, then, that neither the empirical nor the theoretical literature provide clear guidance on how growth in the working age population affect growth in labour productivity. The significance for New Zealand’s labour productivity of the trends shown in Table 3 is therefore uncertain.

3.4  Fertility rates and female labour force participation

New Zealand and the other countries with relatively young age structures owe these age structures to relatively high fertility rates. High fertility rates might be expected to depress the participation term in Equation 1, on the grounds that parenthood keeps people out of the workforce. Given the prevailing division of labour between genders, the effect on participation would presumably be most apparent for women.

Figure 6 presents 2001 OECD data on fertility rates and female labour force participation to test these ideas. The two countries with the highest fertility do indeed have the lowest participation rates. But these two counties, Turkey and Mexico, are clear outliers. Among the remaining countries, there is in fact a mildly positive association between fertility and labour force participation. This was not always the case: in the 1970s the association was negative (Ahn and Mira 2002). Scholars generally explain the current positive association by pointing to institutional features, such as the availability of childcare, that reduce the incompatibility of childrearing and paid work, and raise both fertility and participation (Rindfuss and Brewster 1996). The reason that Mexico and Turkey fail to conform to these generalizations is probably that they are significantly poorer than any other OECD countries, and have rather different economic and social structures.

The international literature on fertility rates and labour force participation is vast and inconclusive. Given that there is no clear relationship between fertility rates and participation, and that fertility rates have in any case been converging across the OECD (Figure 3), it seems unlikely that differences due to fertility in participation rates will be an important source of cross-country variation in per capita incomes.

Figure 6 – Female labour force participation rates versus total fertility rates, OECD 2000

Source: OECD Health and Labour Force Indicators online databases.

Note - The female labour force participation rate is defined here as the female labour force as a percentage of the female population aged 15-64.