Seasonality of birth rates in agricultural Iceland

ABSTRACT

The seasonal pattern of birth rates in nineteenth-century agricultural Iceland, peaking in late summer and early autumn, gradually disappeared when the population migrated to fishing villages in the last decades of the nineteenth century and the first three decades of the twentieth century. We describe how this pattern is consistent with changes that have occurred in other countries and discuss some possible causes.

KEYWORDS:

• Fertility
• urbanisation
• seasonality

JEL CLASSIFICATION:

• J13
• D02

Acknowledgement

We thank Asgeir Jonsson, Ron Smith and two anonymous referees for useful comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 The population was around 40,000 between AD 900 and 1100; 55,000 from 1000–1300; 60,000 from 1300–1500; and 55,000 from 1500–1900. It started to grow in the 1890s, reaching 77,967 in 1900; 84,528 in 1910; 120,264 in 1940; 173,855 in 1960; 226,948 in 1980; and 279,049 in 2000. As of 2016, it is 332,529. Source: Gunnarsson (2000) and Statistics Iceland (2016).

2 For surveys of the literature, see Lam and Miron (1991).

3 The population was 71,981 in 1880. There were about seven sheep for every person alive; that is, 515,364 sheep on the island, 23,337 cows, and 41,342 horses. By 1990 the population had increased to 253,784, with around two sheep per person (548,707 sheep), while the number of cows had increased in proportion to the population, to 74,903 cows.

4 Our main data source is Statistics Iceland (2016). In 1990 the agency published Icelandic Historical Statistics, which includes annual data on the proportion of the Icelandic population living in urban nuclei from 1889 to 2014, the number of births by month from 1853 to 2015, and data on the share of labour employed in agriculture in the period 1801–1990.

5 We also plotted the average absolute deviations of the monthly ratio of births from the annual average and the plot looked almost identical to the two plots shown in Figures 1 and 2.

6 We do not have historical time series for the number of women per month nor for the number of women in each age group. Therefore it is not possible to calculate the fertility rate by month going back to 1853.

7 Similar changes have also occurred in recent year in Denmark, Finland, Germany and the Netherlands. Data provided by authors upon request.

8 So, to take an example, the February number for 1901–1948 is the average of the February values of b over this period.

9 A more detailed look at each period can be found in the Appendix.

10 Similar changes have also occurred in recent years in Denmark, Finland, Germany and the Netherlands. Data provided by authors upon request.

11 A unit root test (augmented Dickey–Fuller test) that included both a constant and a trend term rejected the null hypothesis of a unit root for August births. (ADF = −4.86, where critical values were −4.03 for significance of 1% and −3.45% for significance at the 5% level). The hypothesis is also rejected when allowing for a constant and NOT a trend term for the urbanisation variable (ADF = −6.79 where −3.48 is the critical value at the 1% level and −2.88 is the critical value for the 5% level).

12 Data on urbanisation do not exist before 1889.