NOTES

1.

2.

3.

4.

5.

6.

7.

The EAP is a comparative project dealing with historical demography from a household perspective in five European and Asian populations: Belgium, China, Italy, Japan, and Sweden.

For a more detailed definition, see Section 4 below.

As Jorberg points out, using the rye price instead of a cost of living index (containing a more realistic bundle of consumption goods) will give a reasonably good approximation until the final decades of the nineteenth century when the real wage gain will be somewhat overestimated (Jorberg 1972: vol. II, 335).

See Section 4 for a discussion of the relationship between harvest outcome and grain prices at the local level, and how grain prices can be assumed to have affected the landless.

For example, in nineteenth-century Sweden, spending on food was 83% of the typical household budget (Myrdal 1933: 115), and of that sum, 59% was spent on grain (Jorberg 1972: vol. II, 182).

The proportion of the population that got poor relief in Sweden in 1829 was only 2.1%. It was even smaller in Scandia (Skoglund 1992). For a more detailed discussion, see Bengtsson (2004).

The Scandian Demographic Database is a collaborative project between the Regional Archives in Lund and the Research Group in Population Economics at the Department of Economic History, Lund University. The source material is described in Reutersward and Olsson (1993).

Mantal is an old tax unit used to measure the productive potential of the farm and as a basis for the taxes to be paid to the crown (see for example Dribe 2000: 26—7).

We also tried analysing the truly landless group separately, which yielded practically identical results.

See Bengtsson and Dribe (1997) for details on the construction of the series.

¹¹ With a Hodrick-Prescott (HP) filter a smooth curve is fitted through a time series, rather than a deterministic trend (e.g linear or polynomial) or a moving average, which have been shown to have undesirable effects on the data (e.g Harvey and Jaeger 1993). It is also a less crude way of de-trending than the use of first differences, leaving more of the medium-term cycles. However, Harvey and Jaeger (1993) argue that the HP filter might create spurious cyclical patterns (medium-term), which could imply that observed correlations between series, de-trended by a HP filter, could be overestimated. However, despite these problems the HP filter is used in this study to de­trend the aggregate series to be used as communal covariates in Cox regressions, since the same and even larger problems are associated with the alternative approaches available to us. The smoothing parameter was set to 100, which is the value normally used for annual data.

¹² For a more detailed discussion and analysis of this problem, see Dribe (2000: ch. 7).

¹³ For the basic concepts of event-history analysis, see any standard textbook on the topic, for example, Collett (1994).

¹⁴ Estimations of models for infants and the elderly not reported here also support this conclusion.

¹⁵ For a deeper analysis of the fertility response to economic stress in the same area, see Bengtsson and Dribe (2002). That paper discusses in more detail social differences in the response, seasonal patterns, linearity, and threshold effects, as well as whether the response is intentional or not.