In a recent post,
I used a Mundlak within–between mixed model to ask: when cohorts within the same country grew up in safer and richer times (higher life expectancy and log GDP), did they complete fertility at lower levels? The answer was yes, robustly so across multiple childhood exposure windows and placebo tests.
Subscribe free to get weekly insights that don’t make it into mainstream coverage. Paid members unlock deep dives + exclusive apps for exploring the data yourself.
This time, I refine the setup rather than switching designs outright:
👉 I extend the childhood window by splitting it into 0–5 (early childhood) and 6–18 (school ages), to see whether shocks in survival and development matter differently across formative stages. I also add child mortality alongside life expectancy, since mortality varies more year to year and lets me disentangle its effects from later survival improvements. Finally, I introduce an 18–45 window, capturing exposure during the reproductive years themselves, to contrast developmental scarring with period replacement/insurance effects.
The true Δ-model-first-differencing across adjacent cohorts to capture pure short-run shocks-is something I will bring in next. For now, these richer Mundlak within–between specifications already allow us to distinguish childhood vs adult exposures, and early vs later childhood conditions.
So the new models ask: within countries, when childhood or adult conditions (mortality, life expectancy, GDP, education ratios) shift, how do those shifts map onto completed fertility outcomes?
So what do the results show?
Across childhood windows (0–14, 0–18, 0–25), the within-country deviations line up clearly: cohorts exposed to higher child mortality in childhood went on to complete fewer births. The GDP effect is also consistently negative in childhood windows, reinforcing the idea that upward shifts in living standards during development tend to coincide with smaller eventual family size.
When I split childhood into 0–5 vs 6–18, the picture sharpens: life expectancy effects concentrate in the 6–18 school years, while child mortality matters more in 0–5. This suggests different mechanisms-early scarring vs later educational/partnership pathways-rather than one unified childhood effect.
By contrast, in the 18–45 window the signs flip. Child mortality is now positively associated with fertility (consistent with replacement or insurance behavior), while GDP during adulthood becomes pro-cyclical with fertility. Life expectancy, however, remains negatively associated, pointing to structural modernization effects that keep fertility declining even when economies expand.
Key takeaways
Childhood shocks → lower fertility (scarring).
Within countries, higher child mortality and higher life expectancy shifts during childhood both predict lower completed fertility.
GDP gains in childhood also predict lower fertility, consistent with modernization and opportunity-cost channels.
Timing inside childhood matters.
0–5: child mortality matters most here.
Higher child mortality in infancy/early childhood → lower completed fertility.
Likely reflects a direct demographic channel (children dying before reproductive age reduces average cohort fertility), with some possible scarring among survivors (poorer health, delayed marriage).
6–18: life expectancy matters most in these school-age years → consistent with education, partnership, and opportunity pathways.
GDP in childhood: consistently negative.
Cohorts growing up during economic improvements → lower completed fertility, fitting the modernization/opportunity-cost story.
Adult exposure (18–45) flips the story.
Child mortality turns positive → higher mortality during reproductive years is linked to higher fertility, reflecting replacement/insurance behavior.
GDP turns pro-cyclical → better macro conditions in adulthood coincide with higher fertility at the margin.
Life expectancy stays negative, in line with longer-run modernization pressures.
Education ratios are secondary.
Effects are small and inconsistent across windows, with signs sometimes reversing by period, likely reflecting composition effects that we will examine further. In contrast, in the 18-45 window the estimates are highly significant, consistent with evidence that higher female-to-male education is associated with lower fertility.
Bottom line: Early-life exposures reduce fertility, but through different mechanisms. In early childhood (0–5), higher mortality lowers fertility largely because more children die before reaching reproductive age (direct truncation), with some added scarring among survivors. In later childhood (6–18), survival and economic improvements reduce fertility via schooling, health, and opportunity-cost pathways. By contrast, adult exposures (18–45) work through period responses: higher mortality pushes fertility up (replacement/insurance), while GDP gains also raise fertility (pro-cyclical timing). The same forces matter, but the mechanisms flip depending on when in the life course the cohort is exposed.
If you loved this post, please share it with friends or invite someone to check it out!