14.451: Macroeconomic Theory I Suman S. Basu, MIT Handout 1: Empirics of Economic Growth Welcome to 14.451, the introductory course of the macro sequence. The aim of this course is to familiarize you with the mechanics of growth models, and we anticipate that several of the models will provide knowledge spillovers into other courses you take here. It is difficult to say that you have been properly educated in growth theory without having come across the Lucas quote (although this statement might be a little dated). So let us begin with it. Is there some action a government of India could take that would lead the Indian economy to grow like Indonesia s or Egypt s? If so, what, exactly? If not, what is it about the nature of India that makes it so? The consequences for human welfare involved in questions like these are simply staggering: once one starts to think about them, it is hard to think about anything else. Lucas [JME 1988] This is the fundamental question we are grappling with; but in order to shed light upon this question we need to try to understand the sources of sustained economic growth. That is the object of the course. The purpose of this handout is to explore in further detail some of the empirical aspects of growth covered in the first lecture. 1. Historical Comparisons of Growth Performance The lecture notes highlighted the dispersion in incomes per capita and post-wwii growth rates. To provide further illustration, in Figure 1 below plots the real GNP per capita of several countries in 1992 against the historical performance of the US economy, using data from Maddison (1995). In 1992 Argentina s income per capita was comparable to the US income per capita around World War II, and Pakistan s income per capita in 1992 was below the US level of 18. The differences are the product of many years of sustained growth. 25, Dollars 2, 15, 1, 5, 18 188 189 19 191 192 193 194 195 19 19 198 199 Figure 1 Switzerland Japan Canada Sweden UK Argentina Czechoslovakia Mexico Poland China Pakistan Nigeria Figure 1: Data from Maddison (1995). Adapted from Figure 1.2, p. 3, in Helpman (24). How has the the performance of the world economy fared over time? The answer to this is in Figure 2. Clearly, world growth picked up momentum during the Industrial Revolution, and reached a peak of nearly 3 percent per annum during the 195-193 period, also known as the Golden Age of growth. Since 1998, world growth has been resurgent. Global output rose by 4.9% in 24, led by China (9.1%), Russia (.%), and India (.2%). 1
3. 2.5 Percentage 2. 1.5 1. 1-15 15-182 182-18 18-1913 1913-195 195-193 193-1998 Figure 2 Figure 2: Data from Maddison (21). Adapted from Figure 1.5, p., in Helpman (24). 2. Empirical Results on Convergence Convergence in the unconditional distribution requires that poorer countries get relatively richer over time. As mentioned in the lecture notes, the world income distribution is remarkably persistent. Figure 3 below plots growth rates of countries over 195-85 against their initial income level in 195. Unconditional convergence implies a negative slope. On the contrary, the regression line is virtually flat and, if anything, gently upward sloping. Therefore those countries that are rich in 195 have grown faster in the subsequent 2 year period. 5 Per Capita Growth Rate, 195-85 5 25-25 -5 5 8 9 1 Log (195 Real Per Capita GDP) Figure 3 Figure 3: Data from Summers and Heston (1993). Adapted from Figure 12.2, p. 42, in Barro and Sala-i- Martin (1999). At one level, this is not surprising. If a country differs from another along key parameters such as human capital and physical investment rates, there is little reason to expect them to converge. On the other hand, we would expect regions with very similar underlying characteristics to converge (conditional convergence).one method that has been used in an attempt to control for differences in steady states is to use Barro growth regressions. Another approach has been to restrict attention to countries or regions which are known to have similar characteristics. The restricted subset of countries tends to exhibit convergence. Examples include OECD countries, US states (Figure 4), European regions (Figure 5), and Japanese prefectures (Figure ). 2
For more detail consult Barro and Sala-i-Martin (1999). 25 NC FL VA Annual Growth Rate 2 15 SC GA TN AL AR MS WV TX KY NM KS MD MO MEINWE MN NH UT WI MI DE LA IA TL OH SD ND PA WA OR NJ CT NY MA RI 1 ID WY CO CA AZ NV MT 5 -.4.4.8 1.2 1. 2 Log of 188 Per Capita Personal Income Figure 4 Figure 4: Adapted from Figure 11.2, p. 389, in Barro and Sala-i-Martin (1999). Annual Growth Rate, 195-199 Relative to Country Mean 8 4-4 -8-12 -1 85 3 8 88 81 8 3 9 1 4 9 28 1 58 1 22 5 5 531 55 39 38 3 2 4849 15 54 4 2 82 33 9 229 3 5 4 21 14 91 44 9 2 12 11 53 32 4135 1 34 42 59 83 4 52 45 89 3 18 13 5 9 8 8 5 51 1 83 4 2 43 8 2 24 23 84 -. -.4 -.2.2.4..8 Log of 195 Per Capita GDP Relative to Country Mean 25 Figure 5 Figure 5: Adapted from Figure 11.8, p. 399, from Barro and Sala-i-Martin (1999). 3. Distribution of Income Between and Within Countries Figure below plots the distribution of GDP per worker relative to the United States in 19 and 199. In both periods, countries have been sorted in ascending order of productivity, and each country is treated as a single unit. It is true that the reordering of countries between periods (many countries have moved from one percentile of income to another) makes the figure easy to read but difficult to use in order to derive the dynamics of individual economies. Jones (199) points out that there is a break in the th percentile of the distribution. Above this level of income, there has been a flattening of the income distribution. Below the threshold (i.e. for 2/3 to 3/4 of the distribution), there has been significant divergence. 3
Annual Growth Rate, 193-199 2 58 54 5 4 42 9 13 1 12 3 3 3 1 4395 21 1 11 45 41 8 24 18 32 4 2 25 31 44 23 42 38 19 29 33 34 3 2 43 35 4 22 1 15 2 38 28 34-2. -2.2-1.8-1.4-1 2 -. 14 -.2 Log of 193 Per Capita Income Figure Figure : Adapted from Figure 11.5, p. 395, from Barro and Sala-i-Martin (1999). 1 Relative output per worker (U.S. = 1.).9.8...4.3.2.1 199 19 1 2 3 4 5 Percentile 8 9 1 Figure Figure : Adapted from Figure 1, p. 134, in Jones (199) 4
An interesting issue is the level of inequality between and within countries. World income inequality can be measured by a variety of methods, but the Gini coefficient and the Theil index are most common. Both are equal to when incomes are distributed equally, and rise to 1 as the income distribution gets more and more extreme. The Theil index can be decomposed into between and within contributions to world inequality, and this is the measure employed in the following diagrams (Figures 8 and 9). (Take the early 18s numbers with a pinch of salt.).9.8...4.3.2.1 Between countries Within countries 182 185 18 189 191 1929 195 19 19 198 1992 Figure 8 Figure 8: Data from Bourguignon and Morrisson (22). Adapted from Figure.2, p. 89, in Helpman (24). 1..9.8...4.3.2.1 182 185 18 189 191 1929 195 19 19 198 1992 Figure 9 Figure 9: Data from Bourguignon and Morrisson (22). Adapted from Figure.3, p. 9, in Helpman (24). On this measure, aggregate inequality rose during the nineteenth century, but not so much in the twentieth century. Within country inequality rose up to WWII, then fell until the 19s before rising moderately. Inequality between countries did not diminish in the post-wwii period, a fact consistent with the empirical evidence presented earlier in this handout. The share of within country inequality in overall world income inequality was high in 182, but fell sharply until WWII. Thereafter it stabilized, indicating that within and between trends were roughly in line with each other. 5
4. Does economic growth help the poor? Economic growth figures usually deal with averages. In this section I will try to show that in general, economic growth matters at least in part for the poorest members of society as well. Figure 1 below shows that the share of the poorest in world personal income has shrunk over time, although the rate of decline was markedly lower after WWII. The Figure 11 comes from the work of Dollar and Kraay (22), who showed that the average real income per capita of a country s poorest quintile moves almost one for one with the average real income per capita of the country s entire population. Of course, this is just a correlation rather than necessarily causation, but it is hard to argue that there is no trickle down effect at all. Thus, it is to be hoped that the recent high rates of economic growth in some developing countries should contribute to poverty reduction goals. 5. 4.5 4. Percentage 3.5 3. 2.5 2. 1.5 1. 182 185 18 189 191 1929 195 19 19 198 1992 Figure 1 Figure 1: Data from Bourguignon and Morrisson (22). Adapted from Figure., p. 1, in Helpman (24). 1 Log (per capita income in poorest quintile) 9 8 5 4 3 5 8 9 1 Log (per capita income) Figure 11 Figure 11: Data from Dollar and Kraay (22). Adapted from Figure.9, p. 19, in Helpman (24).
Bibliography [1] Barro, Robert J., and Xavier Sala-í-Martin. 1999. Economic Growth. Cambridge, Mass.: MIT Press. [2] Jones, Charles I. 199. Convergence Revisited. Journal of Economic Growth, Vol 2, pp. 131-153. [3] Helpman, Elhanan. 24. The Mystery of Economic Growth. Cambridge, Mass.: Harvard University Press. [4] Lucas, Robert E. 1988. On the Mechanics of Economic Development. Journal of Monetary Economics, Vol 22(1), pp. 3-42.