Fertility-Employment Correlation

Definition

The fertility-employment correlation refers to the empirical association between the total fertility rate (TFR) and the female labour force participation rate (FLP) across countries and within countries over time. The central puzzle is that the cross-country correlation reversed sign between the 1960s and the 1990s — from negative (high-FLP countries had lower fertility) to positive (high-FLP countries now have relatively higher fertility) — yet within each country the TFR-FLP association remained negative throughout. The sign reversal is a cross-sectional aggregation artifact, not evidence of a structural change in the micro-level relationship between work and childbearing.

Key Ideas

Cross-Country Sign Reversal (1960–1999)

Using 21 Organisation for Economic Co-operation and Development (OECD) countries, Engelhardt and Prskawetz (2004) document that the cross-country Pearson correlation between TFR and FLP shifted from approximately $-0.8$ in 1960 to approximately $+0.4$ by 1999. Countries that once combined high female employment with low fertility (the Nordic countries, United States, Canada) now combine high FLP with relatively higher TFR. Meanwhile, Mediterranean countries (Italy, Spain, Greece, Ireland, Belgium, Netherlands, Luxembourg, with the Netherlands later migrating out of this group) moved from moderate-to-low FLP to the global extremes of lowest FLP and lowest TFR simultaneously.

Within-Country Association Remained Negative

The cross-sectional reversal does not hold within countries over time. Kögel (2004) and Engelhardt, Kögel, and Prskawetz (2004) demonstrate using pooled cross-section panel models and error-correction models, respectively, that the within-country time-series association between TFR and FLP was negative throughout — and in some specifications became more negative over time. The sign reversal in the cross-country correlation is therefore explained by:

Unmeasured country-specific fixed effects: high-FLP countries share unmeasured features (institutional, cultural) correlated with sustained fertility.
Country heterogeneity in the magnitude of the negative slope: low-FLP countries experienced steeper declines in TFR per unit increase in FLP, producing the "neither-work-nor-children" outcome.

Dynamic Country Groupings

Engelhardt and Prskawetz (2004) classify 21 OECD countries into three FLP groups tracked across four decades:

High FLP ( $>65\%$ ): Nordic countries, UK, USA, Canada — historically high participation; sustained or moderately declining TFR.
Medium FLP ( $50$ – $65\%$ ): France, Germany, Japan, Australia — intermediate trajectories.
Low FLP ( $<50\%$ ): Mediterranean countries (Italy, Spain, Greece) plus historically Ireland, Belgium, Netherlands, Luxembourg — most pronounced TFR decline. The Netherlands shifted into the high-FLP group across this period primarily via part-time employment expansion.

Drivers of More Pronounced TFR Decline in Low-FLP Countries

Three variables explain why fertility fell furthest in low-FLP countries:

Female unemployment rose most sharply in low-FLP countries from the late 1970s onward, combining with an already weak male economic position to create uncertainty over future income.
Mean age at first birth increased most in low-FLP countries from approximately 1985 — indicating accelerating postponement of fertility, with each year of delay reducing completed cohort fertility.
Family allowances for first children declined in low-FLP countries in the mid-1980s, while remaining stable or rising in high-FLP countries — removing a key financial cushion at the margin.

Variables that do not explain the pattern include part-time employment share, female working hours, and pre-primary enrolment rates. Pre-primary enrolment was in fact highest in Mediterranean low-FLP countries (state-subsidized day care without flexible market childcare), revealing a mismatch between the type of family policy available and the type needed.

Extended Correlation Reversals (Kohler, Billari, and Ortega [KBO] 2006)

Kohler, Billari, and Ortega (2006) extend the correlation reversal finding beyond TFR-FLP to the full family-formation complex across OECD countries from 1975 to 1996:

TFR × FLP: $-0.61 \to +0.81$
TFR × total first marriage rate (TFMR): $+0.50 \to -0.02$
TFR × total divorce rate (TDR): $-0.48 \to +0.27$
TFR × extra-marital birth share: $-0.41 \to +0.61$

The simultaneous reversal of all four correlations reflects the same institutional bifurcation: high-compatibility countries now combine high FLP, delayed but completed marriage, relatively higher divorce, high nonmarital births, and moderate-to-high TFR; low-compatibility countries combine low FLP, early but declining marriage, low divorce, very low nonmarital births, and lowest-low TFR. This is the signature of the Lowest-Low Fertility regime in Southern and Eastern Europe vs. the high-compatibility Anglo-Nordic regime.

"Neither-Work-Nor-Children" Trap

In Mediterranean low-FLP countries, a cluster of reinforcing factors generates a self-reinforcing low-equilibrium:

Traditional gender and family norms requiring male breadwinner role to be established before marriage/childbearing.
Deteriorating male economic position (high male unemployment, widening gender wage gap eroding male relative earnings, declining first marriage rates).
Declining family allowances for first children.
Rising age at first birth, reflecting permanent postponement rather than temporary delay.
Rigid labor markets offering little part-time or flexible employment.

The result is that women in these countries forego both employment and childbearing, rather than choosing one over the other. This directly contradicts the simple role-incompatibility hypothesis, which would predict that low female employment reflects a choice to specialize in childbearing — but instead, both outcomes are jointly suppressed.

Failure of the Role-Incompatibility Hypothesis

The original negative cross-country correlation (pre-1980) was interpreted under the role-incompatibility hypothesis: employment and fertility compete for women's time; countries with high FLP sacrifice fertility (and vice versa). The post-1980 positive cross-country correlation refutes this interpretation at the aggregate level. But more importantly, the Mediterranean trap shows that even the old negative cross-sectional correlation masked heterogeneity: low-FLP countries did not simply have women specializing in fertility — they had institutional and normative structures that depressed both.

How It Works

The mechanism linking institutional context to the dual outcome involves sequential effects:

Weak male economic position delays household formation (marriage, cohabitation).
Strong traditional family norms condition childbearing on stable household formation (low out-of-wedlock births in Mediterranean countries, very low compared to Nordic or Anglo-Saxon countries).
Limited flexible or part-time employment makes combining work and family structurally difficult once children arrive.
Declining family allowances reduce the financial feasibility of the first child.
Women respond by postponing — and ultimately foregoing — both employment entry (waiting for stable positions) and childbearing (waiting for stable household conditions).

Why It Matters

For Cross-Country Fertility Research

The cross-sectional sign reversal is frequently cited in policy discussions as evidence that "work-family compatibility" policies raise fertility — but the micro-level within-country evidence consistently shows negative associations. This distinction matters enormously for policy evaluation: interventions that raise female employment may not causally raise fertility, and the positive cross-sectional association reflects selection of high-institutional-quality countries into both outcomes.

For the Second Demographic Transition

The below-replacement fertility of OECD countries is not uniform in its distribution across the country spectrum. Mediterranean lowest-low fertility is the extreme manifestation of the second demographic transition where ideational change (secularization, individualization) interacts with institutional rigidity to produce a dual constraint. See Demographic Transition.

For Family Policy

The finding that pre-primary enrolment does not track the FLP-fertility pattern — and that flexible childcare provision, family allowances at first birth, and male labor market quality do — reshapes what family-policy interventions should target.

Open Questions

Is the within-country TFR-FLP association causal, and in which direction? Engelhardt and Prskawetz (2004) is purely descriptive; the error-correction models of Engelhardt, Kögel, and Prskawetz (2004) establish cointegration but not clean causal direction.
To what extent did the Mediterranean countries' position change post-2000 as rigid labor markets partially reformed and FLP rose?
Does the "neither-work-nor-children" trap have a stable equilibrium or does it eventually shift as cohorts socialized under new norms replace older cohorts?
What is the micro-level mechanism: are individual women making a joint opt-out decision, or are aggregate patterns the result of heterogeneous individual decisions summing to a corner solution?