Out-migration

We now turn to the results of the event-history analysis of out-migration. We first estimated the effects of logged raw rice prices on the probability of out-migration of both native and non-native individuals and also on the reason for migration of native persons, using different time lags (table not shown).

We found that there were clear and contrasting gender differences in migration responses to rice price fluctuations. The likelihood of male out-migration decreased when local economic conditions deteriorated. Male migration responses were immediate and, though they became weaker as time passed, lasted for a few years. By contrast, female out-migration was in general unaffected by economic variations in the community. This suggests that male out-migration in the two villages, compared to female outmigration, was more sensitive to changes in local economic conditions.

We also found that the migration responses of native men to price fluctuations tended to be more immediate, while those of non-native men were less immediate but lasted longer. Furthermore, although the patterns of migration responses present a contrast between the sexes when we look at out-migration as a whole, the responses of males and females are similar when we focus on natives by reasons of migration. Local economic downturn reduced the likelihood of service-related out-migration of native males and females, while it significantly raised the likelihood of absconding by both sexes. In addition, changes in local economic conditions affected male service migration more quickly than they did female service migration, though the effect on female service migration lasted much longer.

In summary, male out-migration was in general much more responsive to changes in local economic conditions than was female out-migration, and this was due primarily to the fact that a large part of male migration was service-related.

On the other hand, female out-migration was much less responsive to local economic changes, owing in part to the fact that the proportion of female labour-related migration was small, and also to the fact that out-migration due to marriage, which represented the largest proportion of female migration, was little affected by annual changes in local economic conditions. Nonetheless, for both native males and native females, the likelihood of service out-migration declined when local economic conditions worsened, because the deterioration of economic conditions was not limited to their own villages but affected the entire Nihonmatsu domain or even the whole northeastern region. Because the destinations of service migration were mostly villages within the domain (Narimatsu 1985: 103—4; Takahashi 1999), a downturn in local economic conditions negatively affected the local labour market, therefore reducing the opportunities of service outside the villages. By contrast, for both males and females, absconding increased significantly at the times of economic hardship, as expected.

Table 17.6 presents the results of the event-history analyses of out-migration of native men and women, based on the models that include not only price variations

Table 17.6 Estimated effects of Ioggedrice prices and household landholding on the probability of out-migration in the next one year by sex and reason of migration: nativemales and females aged under 75 in Shimomoriya and Niita, 1716-1870

bgcolor=white>0.003"

Covariates	Migration of native males				Migration of native females
	All migration	Migration			All migration	Migration
		Marriage	Service	Abscondⁱⁿg		Marriage	Service	Abscondⁱⁿg
L°gg^ed rice price, in current year	-0.375"	-0.071	-0.672**	0.488**	—	—	—	—
time-lag by 1 year	—	—	—	—	0.008	-0.259#	-0.484*	1.483**
Landholding	-0.055"	-0.146"	-0.024**	-0.018*	-0.066^**	-0.020**	-0.020#	-0.122**
Landholding squared	0.001"	0.001**	—	0.001^**	—	0.001*	0.002**
Time period
1760-99	0.144#	0.686"	-0.250**	1.684**	0.291^**	-0.308*	0.247#	2.087**
1800-39	0.002	0.560*	-0.425**	1.795**	0.017	-0.591**	-0.032	2.134**
1840-70	-0.811"	0.690*	-1.462**	0.532	-0.541**	-0.857**	-1.935**	1.559**
Log-likelihood	-5,642.98	-749.19	-3,895.87	-1,341.47	-4,252.08	-1,765.37	-1,785.79	-838.73
χ²	632.14	250.08	451.92	358.78	797.89	747.47	258.22	213.59
(d.f.)	(23)	(21)	(23)	(22)	(23)	(22)	(22)	(23)
P>χ²	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000
No. of observations	48,807	45,500	48,807	48,807	45,485	45,485	45,241	45,485
No. of events	1,309	131	834	237	959	413	334	138
No. of individuals	2,789	2,690	2,789	2,789	2,789	2,789	2,741	2,789

Significant at 0.01 level.

* Significant at 0.05 level.

# Significant at 0.10 level.

Notes: Estimated withperson-year recorded in local population registers as a unit ofobservation. Estimation of standard errors takes into account the effectsof contribution of more than one observation from the same individualsby using Huber's formula. The analysis above controls for the effectsof age, residing village, number of kin household members, and maritalstatus. In the model for male migration due to marriage, 3,307 observationswere dropped because men whose marital status were unknown or those atage 45-49 did not at all experience the event in question. In themodel for female migration due to service, 244 observations were dropped because none of the women whose marital status were unknown experienced the event in question.

but also household landholding. As shown in the table, household landholding significantly and uniformly lowers the probability of all types of out-migration for both sexes, and the negative effects are often not linear. This suggests that both men and women were less likely to migrate out of their villages if they lived in households with greater resources and wealth. Although a one-unit (koku) increase in landholding does not always have the same effect, there is a large difference in the likelihood of out-migration between individuals living in households with zero or only a small landholding and those in households with large landholding.

Finally, as shown in Table 17.6, similar temporal differences are observed for both sexes in the overall probability of out-migration. Relative to the earliest sub-period (1716—59), the likelihood of out-migration in 1760—99 was significantly higher (this being the case especially for female out-migration), while the likelihood of out-migration became significantly lower in 1840—70. Increases in the probability of out-migration in 1760—99 were due primarily to dramatic increases in absconding at the time of acute and prolonged economic hardship caused by the Tenmei famine in the 1780s (Tsuya 2000). The propensity of absconding was also high in the following sub-period of 1800-40, being propelled by hardships caused by the Tempo famine in the 1830s and a series of years of bad weather preceding it. On the other hand, the probability of out-migration in 1840-70 declined, solely due to a large reduction in service migration, and this was especially the case for male out-migration because a large proportion of male out-migration consisted of labour migration. For females, in addition to declines in service migration, decreases in marriage-related migration contributed to lowering the overall probability of female migration since the out-migration of native women for marriage declined sharply over time.

<< | >>

↑

Source: Allen R.C., Bengtsson T., Dribe M.. Living Standards in the Past: New Perspectives on Well-Being in Asia and Europe. Oxford University Press,2005. - 495 p.. 2005

Out-migration

More on the topic Out-migration: