The difference in the speeds of convergence between developed and emerging countries in Figures 2 and 3 merits a few comments. Starting with fertility rates, we note that family size is primarily driven by choice, as socioeconomic situations change over time. The impact of plunging birth rates on support (or dependency) ratios is felt very quickly, as a rapid drop in fertility ratios immediately affects the number of children. Fast-rising life expectancy, on the other hand, takes decades to affect our support ratios, for the obvious reason that it takes decades for a child to become a senior citizen.
Medical advances are the most important contributor to the sharp increases in longevity we have seen over the last century. These changes impact our support ratios at a slow pace, as more and more people survive to become senior citizens. On the other hand, medical advances have a modest effect on child-support ratios: Infant mortality in developed countries was already at very low levels a half-century ago, and fertility treatments still have limited applications.12
Let's go back to Figure 1 and take a look at the 1950 U.S. population profile (Phase IIa, the dark gray line). We can clearly see a high fraction of children aged 0-4—in fact, almost as high as Halley's profile—followed by a sharp drop (a local valley) between the ages of 5 and 19, as a consequence of the Great Depression and World War II. Consequently, in the decades after 1950, the United States received a generous demographic dividend: The baby boomers—and later their children—entered the work force with little need to provide for previous generations and little competition for jobs from older age groups.
To see this strong tail wind at work, the 2010 U.S. profile (Phase IIb, the yellow line) shows that we have fewer children and fewer young adults (as a share of the overall population) than we did in 1950; and we have more mature adults and more senior citizens than in 1950 (or ever before, for that matter). Indeed, for people age up to 54, this demographic profile looks an awful lot like a steady state: There's about 7 percent of the population in each of these 11 age cohorts.
In the coming decades, the still-low fraction of senior citizens aged 65 and above will swell. The baby boomers born right after 1945 are just reaching retirement age, which will put strong upward pressure on support ratios in the next two or three decades. Just imagine the 2010 curve moving to the right with the passage of time. First, the baby-boom bump boosts the senior citizen roster. Then, three decades later, the echo of the baby boom (the smaller bump from age 15-30) boosts the roster of senior citizens a bit further. In the second half of the paper, we show that the past strong tail wind becomes a comparable head wind—for some countries, like Japan, a severe head wind—over the coming decades. This next transitory period (Phase III) will likely impose a "demographic tax" before we can reach a new steady state.