First marriage

We next examine the effects of logged raw rice prices on the probability of first marriage of single men and women by type of marriage, using different time lags.

Table 17.4 Estimated effects of logged rice price and household landholding on the probability of having recorded marital birth in the next one year: women aged 15—49 in Shimomoriya and Niita, 1716—1870

Significant at 0.01 level.

Significant at 0.05 level.

Significant at 0.10 level.

Notes: Estimated with person-year recorded in local population registers as a unit of observation. Estimation of standard errors takes into account the effects of contribution of more than one observation from the same individuals by using Huber's formula. The analysis above controls for the effects of age, residing village, and the number of kin members in household.

For both men and women, the effects of rice price variations are almost entirely negative (i.e. in the theoretically expected direction), indicating that the likelihood of first marriage declined at times of economic downturn (table not shown). We also find that, for both sexes, rice prices with a three-year lag have the strongest effect. Therefore, we use logged raw prices with a three-year time lag in our extended model.

Female first marriages responded more quickly to price fluctuations than did male first marriages. This suggests that female marriages were more vulnerable to changes in local economic conditions than were male marriages. This in turn implies that as economic conditions in the community deteriorated, daughters' marriages were forgone or delayed before those of sons. Given the small offspring size prevalent in the two villages (Tsuya and Kurosu 1999), it is likely that a considerable proportion of men were the primary family heir.²⁸ It may therefore be the case that male first marriages were more resilient to local economic downturn than were female first marriages because families would have put a higher priority on their male heirs' marriages than on their daughters'.

We also note the different effects of rice price variations on first marriages by marriage type.

Table 17.5 presents the results of discrete-time event-history analysis of first marriage by sex that includes not only price variations but also household landholding in the model. As shown in the left panel of the table, even after landholding is introduced into the model, the effects of rice price fluctuations remain significant and strongly negative for male marriages, especially male virilocal marriages.

We can also see that household landholding (net of number of kin members) significantly increased the likelihood of male first marriages. Since a large majority of male first marriages were virilocal (i.e. men brought their bride into their household), this suggests that household economic resources enhanced the chances of first marriage among inheriting sons. Furthermore, the effect of household landholding on male virilocal marriages was not only positive but also curvilinear. This means that, whereas landholding did not make much difference between male heirs of landless peasants and male heirs of peasants with only a small landholding, differences were larger between inheriting sons in households with a medium landholding and inheriting sons in households with a large landholding.

There are clear temporal differentials in the likelihood of male first marriage regardless of marriage type. The probability of male first marriage became significantly lower in 1800—70, relative to 1716—59 (see the left panel of Table 17.5). Given the nearly universal male marriage in the two villages, this finding indicates a delay in the timing of male first marriage in the nineteenth century.

Turning to the probability of female first marriage by marriage type (see the right panel of Table 17.5), three things stand out. First, a downturn in economic conditions in the community as measured by local rice prices significantly reduced the likelihood of female intra-village virilocal marriage, that is, marriage of a woman with a man residing in the same village. This mirrors the negative effect of prices on male virilocal marriage. These findings suggest that the deterioration of local economic conditions reduced the likelihood of first marriages within the villages for men and women alike.

Second, household landholding is positively associated with the likelihood of female uxorilocal marriage. This again mirrors the positive effect of landholding on male uxorilocal marriage. We interpret these findings together to mean that the chances of obtaining a marriage partner for inheriting sons and daughters were enhanced by household wealth and resources in pre-industrial Japanese agrarian villages.

Third, there is a very strong and clear trend of decline in the age-specific probability of female first marriages of all types. The likelihood of female first marriage declined significantly and linearly as the time passed. Our earlier study found the

Table 17.5 Estimated effects of logged rice prices and household landholding on the probability of first marriage in the next one year by sex and type of marriage: never-married males aged 10—49 and never-married females aged 5—49 in Shimomoriya and Niita, 1716—1870

Significant at 0.01 level.

* Significant at 0.05 level.

# Significant at 0.10 level.

Notes: Estimated with person year recorded in local population registers as a unit of observation. Estimation of standard errors takes into account the effects of contribution of more than one observation from same individuals by using Huber's formula. The analysis above controls for the effects of age, residing village, and the number of kin members in household. The results for 75 males who married out of the villages are not shown because the results are statistically unstable due to a small number of events.

increasing delay of first marriage among women in these two villages (Tsuya and Kurosu 2000b), and this result confirms that finding.

5.4