The changing nature of the link between infant mortality and fertility in Finland 1776-1978

Abstract

Short-term and long-term variations in data from the long Finnish time series were exploited for possible information on the nature of the relationship between child mortality and fertility. Various statistical models were applied to estimate the importance of different paths of causation between infant mortality and from short-term variations in those time series. All models of analysis--cross-lagged correlational analysis for absolute values and for annual percentage changes distributed lag regressions for total and for age-specific rates and the consideration of child mortality age 1-2 -- gave essentially the same result. In premodern (before 1917) Finland high tends to increase infant mortality in the following year; the infant mortality experience of the previous years does not appear to be an important determinant of current fertility. In the modern period (after 1917) the structure of determination seems to be reversed and high infant mortality experience in a certain year tends to increase in the following years. In terms of the interpretative model given it can be concluded that there is no evidence for a replacement strategy or for a strong lactational effect in premodern Finland. Evidence for a significant effect of fertility stress and possible direct or indirect infanticide appears. Once the practice of family limitation became accepted there was no more need for reducing the number of living children but still some less effective involuntary factors could have been present. The most significant change is that in modern Finland where fertility is within the calculus of conscious choice there seems to be clear evidence for a replacement strategy. To check the plausibility of the assumptions made and to shed more light on the determining influence of long-term development focuse moves to the impact of the secular infant mortality decline. It seems reasonable to see 1860-70 as a threshold since infant mortality enters its steep decline which continues almost linearly until 1960 when the curve which approaches zero levels off. Thus the steep and unreversed decline in infant mortality started in Finland 40-50 years before that of fertility. 40-50 years seems to quite long for the assumed perception lag which should link directly to infant mortality. The development of 5-year averages from 1751-1857 reveals significant declines even before 1880. A regression line drawn through the plot gives a slope coefficient of - .462 implying a decline in the infant mortality rate by 23.1 points in 50 years. Comparing this pattern of early decline in infant mortality to the time series of no effect of that early decline on was found. In fact even increased during the 2nd half of the 19th century.