Summary
A methodological overview paper (book chapter) that integrates and extends the Kohler-Ortega (KO) framework introduced in Ortega and Kohler (2002). The paper presents a unified toolkit combining tempo adjustment, parity-specific fertility tables, and cohort completion methods, and applies it empirically to Italy and the Czech Republic (1985 and 1995). The core contribution beyond the 2002 working paper is the explicit four-factor decomposition linking the Period Fertility Index to annual birth counts (adding mean generation size G), the "fertility ageing effect" concept, quantified cohort completion scenarios under different postponement assumptions, and numerical calibration showing that the Bongaarts-Feeney (BF) adjustment over-adjusts relative to KO by 7% (Italy) and 17% (Czech Republic) in 1995.
Key Claims
- Childbearing intensities (occurrence-exposure rates conditioned on parity) are strictly preferable to incidence rates for period fertility analysis: incidence rates embed historical parity composition, confounding inference about current timing behavior.
- Bongaarts-Feeney (1998) over-adjusts because it estimates tempo change from incidence-rate mean ages, which shift in response to past quantum/tempo changes even when current behavior is stable. KO adjustment eliminates this compositional bias.
- Italy 1995: Observed total fertility rate (TFR) = 1.18; BF-adjusted TFR = 1.50; KO-adjusted TFR = 1.40; Period Fertility Index (PF) = 1.43. BF overstates adjustment by ~7% relative to KO.
- Czech Republic 1995: Observed TFR = 1.25; BF-adjusted = 1.79; KO-adjusted = 1.52; PF = 1.63. BF overstates by ~17%. Czech Republic's rapid first-birth quantum decline created a strongly unfavourable parity composition (d=−9%) and large tempo effect (r=33%) for first births.
- Czech Republic 1985–1995: Underlying quantum (PF) declined only 17%, but observed TFR declined 35% — the extra 18pp gap is accounted for by the tempo effect (+17.7%) and parity distribution effect (−6.7%) emerging in 1995.
- Four-factor decomposition of birth counts: Births=G×PF×1−r1×1+d1, where G = mean generation size. This links the behavioral quantum (PF) all the way to the actual number of births, adding G to the previously published three-way TFR decomposition.
- Cohort completion: Tempo-distorted observed intensities project ultimate childlessness of ~28% (Italy) and ~25% (Czech R.) for cohorts not yet finished in 1996/1999. KO-adjusted "postponement stops" projections yield ~20% and ~13% — a large upward revision, showing observed intensities systematically understate the proportion of women who will ultimately have at least one child given current quantum levels.
- Fertility ageing effect: If postponement continues indefinitely, it doesn't merely shift births later but progressively reduces higher-order birth probabilities, because women arrive at the age of second/third birth risk later, when progression probabilities are already declining rapidly. Quantified by comparing the "postponement stops" vs. "postponement continues" cohort completion scenarios.
Concepts Introduced or Extended
Entities Mentioned
Quotes
"The reliance of both the BF and KP [Kohler-Philipov 2001] models on incidence rates is unfortunate since the inference about tempo change based on incidence rate schedules is affected by the dynamics of the parity composition of the population. This influence of the parity distribution leads to potentially non-negligible biases in the estimates of tempo distortions."
My Take
This is the most accessible and numerically grounded presentation of the KO framework. The Italy/Czech comparison is particularly instructive: the 17% BF over-correction in the Czech Republic is large enough to matter for policy conclusions about whether TFR < 1.3 in the Czech Republic reflected a genuine catastrophic quantum collapse or a postponement-driven artifact. The fertility ageing effect is a genuinely novel insight — it shows that postponement is not just a timing-neutral shift but can compound into real cohort fertility losses if it continues long enough. The cohort completion scenarios make this concrete by projecting radically different final childlessness rates depending on assumptions about future postponement pace.