The Demographic Transition

There is a pattern so consistent across societies that demographers gave up hedging about it. It goes: high birth rates and high death rates, in rough balance, for millennia. Then death rates fall while birth rates stay where they were, and population grows explosively for two or three generations. Then birth rates fall as well. Then both are low and growth stops.

That middle stage is the one that produced most of the twentieth century's alarm, and the essential point about it is that it is not a departure from the pattern. It is the pattern. The surge is the gap between two curves falling at different times, and it closes by itself once the second one moves.

Why does the death rate fall first? Because reducing mortality does not require anybody to change their plans. A sewer works on everyone whether or not they understand it. Clean water, a more reliable food supply, and later vaccines and antibiotics all act on a population without asking permission. Fertility is different in kind: it is the aggregate of millions of decisions taken inside families, and those decisions respond to conditions that take a generation to become evident.

And the conditions that eventually move it are reasonably well established. Children surviving, so that fewer births are needed for the same surviving family. Schooling, which converts a child from a contributor to household income into a long and expensive investment. Pensions and savings replacing sons as old-age security. And above all women's education and paid employment, which is the single most consistent correlate of falling fertility across countries with almost nothing else in common. Contraception makes a smaller family achievable; it does not by itself make one desirable, and fertility fell substantially in several places before modern methods existed.

What the model does not do is predict. It is descriptive: it tells you the order things happen in and is silent about speed and about where fertility stops. Both silences turn out to matter enormously.

Speed first. Britain took roughly a century and a half to run the sequence. South Korea did it in about forty years. A slow transition spreads a large cohort across decades; a fast one produces one enormous generation followed by a much smaller one. For twenty or thirty years that is a gift (a large working population supporting few dependants, which is where a good deal of East Asian growth came from) and then it becomes a pension and care problem of a kind no previous society has faced.

Then the floor. The model was built by people who assumed fertility would settle at roughly replacement, because that seemed like equilibrium. In a great many countries it has gone well below and stayed there, and nothing in the framework says why or where it stops. That is the honest limit of a descriptive model, and it is why the 1970s managed to get the twenty-first century wrong in both directions at once: alarmed about an explosion that was already self-limiting, and entirely unprepared for the shortage that followed.