Let’s start with the facts. During the First World War the fertility rates of European nations collapsed – see Figure 1a. They then recovered, often overshooting their pre-war trend for a limited period of time. The deficit of births during the Great War is estimated, for France, to be about 1.4 million (Figure 1b) while military casualties are about 1.4 million too. This means that the French population was 2.8 million people ‘short’ at the end of the War, and that half of this shortage is accounted for by the collapse of fertility rates. Germany was affected in quite a similar way by the war. Its military casualties are estimated at two million and its deficit of births during the war is about 3.2 million.
Figure 1a. European birth rates
Figure 1b. Number of births in France and Germany (the shaded area corresponds to the date of the war shifted by nine months)
Measures of lifetime fertility, such as Total Fertility Rates and Completed Fertility, convey a message similar to that of Figure 1a. In addition, they show that the generations who reached their twenties during the war had fewer children over the course of their lives than other generations. Thus, even though these generations postponed some births until after the war was over, they did not fully compensate the forgone births of the war.
The conventional view
The conventional view is that the deficit of births resulted from the absence of men who, because of the war, were unable to achieve their desired fertility, rather than from a change in the desired level of fertility (Huber 1931, Vincent 1946, Festy 1984). In support of this view one has to acknowledge that the mobilisation was massive. In France, for example, a total of 8.5 million men served in the army during the war, out of 8.7 million men aged 20-50 in January 1914.
Yet, the view that mobilised men were physically unable to conceive children, and that this was the sole reason for the collapse of fertility is problematic for several reasons. First, and as shown in Figure 1b, the number of births in France was, in the early 1920s, above its pre-war level even though 1.4 million men did not come back from the war. This would not be possible if the absence of men was the sole reason for the fertility collapse. Second, not all the mobilised men were at the front. In fact, at any point during the war between 30% and 50% of the mobilised men were in the 'rear', that is in contact with the civilian population (Huber 1931). Third, men at the front did not stay there for four years in a row, and leave policies became more systematic and generous after the first year of the war.
In a recent paper (Vandenbroucke 2012) I proposed to evaluate the effect of the war on the desired level of fertility for couples that reached their childbearing years on the eve of the First World War. The main idea is that the war is tantamount to a large income shock for two reasons. First, the expected lifetime income of a couple where the husband is mobilised is reduced by the likelihood that he gets killed during the war: a negative permanent income shock. Second, during the war a mobilised husband is unable to work. The government may partially compensate this loss, but it remains a negative temporary income shock for the household.
To assess the quantitative importance of this mechanism I built a model where couples live for a finite period of time and can choose to have children at two points in their lives, say age 1 and 2. Children provide utility but they are costly. In particular, they require that a wife spends time raising them for a number of periods before they move out of the household. A couple faces three unanticipated shocks when the war breaks out:
(i) an increase in the probability that its wife remains alone after the war;
(ii) a partially-compensated loss of its husband's income; and
(iii) a decline in productivity.
These shocks induce an increase in saving and a reduction of consumption. The resulting increase in the marginal utility of consumption raises the cost of having a child, i.e. an income effect. Once the war is over an age 2 couple that reduced its fertility during the war has a high marginal utility for children and, therefore, experiences above-trend fertility.
First, I calibrate the model to fit the downward slopping time series of fertility until the eve of the war. With exogenous wage growth (for men and women) this implies restrictions on preferences such that the substitution effect from wages dominates the income effect. Second, I compute the optimal fertility of couples reaching their childbearing years during and after the war. I use:
(i) the ratio of military casualties to mobilised men (16%) to calibrate the probability that a wife is alone after the war;
(ii) income data to calibrate the proportion of uncompensated income loss by mobilised husbands (50%);
(iii) output per worker to calibrate the drop in wages during the War (30%).
Figure 2a shows the path of fertility predicted by the model versus the actual path at the five-year frequency. (A model period is five years, thus the war lasts for one period in the model.) The model implies that fertility declines by 54% during the war, instead of 49% in the data, and that it increases by 154% after the war, instead of 118% in the data. In summary the model implies a strong effect of the war represented by these three shocks on fertility. These changes occur despite the fact that there are no ‘physical separations’ of couples during the war in the model. Figure 2b reveals that age 2 households have an above-trend fertility after the war due to their low fertility at age 1, that is during the war.
Figure 2a. French Fertility Rate, Model and Data
Figure 2b. French Fertility Rate by Age in the Model
Counterfactual experiments reveal that, among the three shocks representing the war, the shock to expectations, that is the increased probability that a wife remains alone, is the largest contributor to the changes in fertility during and after the war. I calibrated this probability to the ratio of military casualties to mobilised men (16%). This may exaggerate the likelihood that a women remains alone after the war since remarriage was possible. When this probability is set at 10% to mitigate this concern the model predicts a 41% decline in fertility during the war (versus 49% in the data).
The First World War ravaged Europe to an extent that had not been seen until then. Half of its demographic impact is accounted for by the collapse of fertility during the war. But why did such a collapse even occur? In a recent paper (Vandenbroucke 2012) I argue that this was the optimal reaction of rational, forward-looking individuals facing a shock to their expectations. Contrary to the conventional view, I argue that this shock was enough to account for the data even when abstracting from the physical separation of couples during the war.
The case of the First World War was not unique. Caldwell (2004) shows that fertility declined in 13 different episodes of crises such as wars and revolution in various countries and periods of time.
Caldwell, John (2004), “Social Upheaval and Fertility Decline”, Journal of Family History, 29(4):382-406.
Festy, Patrick (1984), “Effets et répercussion de la première guerre mondiale sur la fécondité française”, Population, 39(6):977-1010.
Huber, Michel (1931), La population de la France pendant la guerre, Les Presses Universitaires de France.
Vandenbroucke, Guillaume (2012), “Optimal Fertility During World War I”, Manuscript.
Vincent, Paul (1946), “Conséquences de six années de guerre sur la population française”, Population, 1(3):429-440.