James A. Brander and Steve Dowrick investigate the effects that population growth and birth rates have on economic output in their study published in 1992. In “The role of fertility and population in economic growth”, Brander and Dowrick use data from 107 countries from 1960-1985 to strengthen results from an earlier study of theirs in 1991. The report is based on a regression model they create from this plethora of data. The primary variables that are incorporated into the model are; population, annualized 5 year per capita growth rate, annualized 5 year population growth rate, crude birth rate, adjusted birth rate, share of the population working age, investments as a % of GDP, relative price of investment and the relative productivity of the country.[1] The general hypothesis for this study is that countries with high birth rates will reduce economic growth while those countries with lower birth rates, or declining birth rates, will have a steady per capita income growth over time.
This regression model itself is a highly complex representation of the data. Thus, the specific variables, significance and structure tests will not be examined in great detail other than what is necessary in this paper. The data itself is what is important and where the data itself originates from must first be investigated.
Brander and Dowrick stress the idea that reliable, standardized data is essential in achieving empirical results. Their data comes from two sources. The first source is the income data from Summers and Heston (1991). Their data is contained in the Penn World Table – Mark 5 (PWT5) which is based upon purchasing power rather than standard income data, thus offering better comparisons across countries and time. The second data set, the demographic data, is from the United Nations World Population Prospects (1992).[2] The two groups of data were standardized to compensate for any discrepancies, adjusted to 1985 prices, and then compiled. It should be noted that both data sets are time-series data, which is the best class of data to explain certain trends in specific countries. Individual countries can use resulting information from this study to help understand trends in their own country and possibly influence future policies as well.
After the data was compiled, adjusted and standardized, the regression model was created using the aforementioned variables as a function of y, or per capita real output in 1985 U.S. dollars.[3] The results from the model varied, and in some cases the variables coefficients were statistically significant while many others were not. Of particular interest was the results stemming from the Chow test, a test for break points or any significant structural changes in the model. In this case, the test highlighted a break between the least developed 31 countries and the remaining 76. Thus, the regression model was rerun for these two groups (31 least developed and the rest) and some interesting results were observed.
In the 76-country model, the results seemed to be more precise and statistically significant on the whole. For one, the relationship between the changes in the birth rate and the growth in the working age proportion are much more pronounced and stronger in this particular data set. What this means, as generally predicted by Brander and Dowrick, is that every 1% decrease in the birth rate will yield a 1% growth in the working age 15 years later. This solidifies their main prediction that a negative correlation exists between the birth rate and per capita growth if you consider there to be a positive relationship with per capita income and the size of working class as a percentage of the total population. This model also provided the evidence that investment plays a significant role in determining per capita output growth. The investment variable in this set was statistically significant as it was in model for the 31 least developed countries and in the overall model. Brander and Dowrick, in their conclusion, once again supported the idea that it is fertility that affects investments and thus if fertility is high, investments are low and less economic growth ensues than when fertility rates are low.
In the model for the lesser developed countries, there was found to be a negative coefficient on population growth. This means that there is an inverse relationship between fertility and per capita income growth from the data in these countries. Unfortunately however this model has little explanatory power because the population growth coefficient is not statistically significant. Brander and Dowrick conclude that this may be due in part to lower quality data that comes from these countries.
The bottom line here is that Brander and Dowrick are only halfway there to proving their hypothesis that a negative correlation exists between birth rates and economic growth across all countries. In more developed countries, their “neo-Malthusian” interpretation of fertility and economic growth is proved; sinking birth rates in these countries from 1960 to 1985 have tended to result in economic prosperity. Their overall model also supports the “relative labor supply effect” of declining birth rates; as birth rates fall, there is a period when the working-able population rises significantly more than the dependant population resulting in higher per capita income. In lesser developed countries, Brander and Dowrick’s model does not prove that these countries have a lower per capita income due to high fertility rates. Future data from these countries that is of higher quality may lead to reliable empirical results.
[2] Ibid, 3
[3] Ibid, 10