Mortaiity
Using different time lags (current year up to five years before), we first examine the effects of logged rice price in the local market of Aizu on the mortality of males and females at different life stages.
We find considerable differences in the magnitude and significance of mortality responses to fluctuations in rice price by gender and life stage (table not shown).26 Rice prices are in general associated significantly andTable 17.1 Means of the covariates used for the discrete-time event-history analysis of mortality responses to shortterm economic stress: the villages of Shimomoriya and Niita, 1716—1870
| Covariates | Infant | Child | Adult | Old age |
| Male | ||||
| Logged rice price in Aizu, current year | -0.236 | -0.229 | -0.240 | -0.252 |
| Landholding (in koku) | 12.447 | 12.102 | 12.673 | 10.492 |
| Time period (1716-59) | ||||
| 1760-99 | 0.247 | 0.253 | 0.287 | 0.321 |
| 1800-39 | 0.239 | 0.254 | 0.234 | 0.233 |
| 1840-70 | 0.172 | 0.162 | 0.139 | 0.081 |
| Female | ||||
| Logged rice price in Aizu, current year | -0.218 | -0.228 | -0.235 | -0.247 |
| Landholding (in koku) | 12.340 | 12.554 | 12.566 | 11.266 |
| Time period (1716-59): | ||||
| 1760-99 | 0.242 | 0.247 | 0.295 | 0.327 |
| 1800-39 | 0.244 | 0.258 | 0.245 | 0.266 |
| 1840-70 | 0.194 | 0.173 | 0.148 | 0.090 |
Notes: In the analysis of mortality, infant refers to age 1, childhood to ages 2—14, adult to ages 15—54, and old age to ages 55—74.
For the definition and measurement of age, see the text.Table 17.2 Means of the covariates used for the discrete-time event-history analysis of responses of marital fertility, first marriage, and out-migration to short-term economic stress: the villages of Shimomoriya and Niita, 1716—1870
| Covariates | Marital fertility | First marriage | bgcolor=white>Out-migration|||
| Male | Female | Male | Female | ||
| Logged rice price in Aizu, current year | — | — | -0.240 | — | |
| Time lag by 1 year | -0.245 | — | — | — | -0.244 |
| Time lag by 3 years | — | -0.244 | -0.229 | — | — |
| Landholding (in koku) | 12.469 | 11.359 | 12.668 | 12.166 | 12.348 |
| Time period (1716-59) | |||||
| 1760-99 | 0.287 | 0.281 | 0.256 | 0.285 | 0.288 |
| 1800-39 | 0.229 | 0.259 | 0.289 | 0.2380 | 0.251 |
| 1840-70 | 0.149 | 0.159 | 0.197 | 0.135 | 0.145 |
Notes.
The analysis of marital fertility looks at married women aged 15—49; the analysis of first marriage looks at never-married males aged 10—49 and never-married females aged 5-49; and the analysis of out-migration looks at males and females (of all marital statuses) under age 75. For the definition and measurement of age, see the text.positively with female infant and child mortality, while the association with male childhood mortality is, though mostly positive, statistically insignificant. This suggests that female infants and children were more highly responsive to shortterm economic stress caused by annual fluctuations in agricultural output, but that the chances of survival for males were largely unaffected by such variations.
By contrast, we find that local rice price variations had a significant positive effect on the mortality of adult and elderly men, and their mortality responses to these local economic fluctuations were immediate and short-lived. On the other hand, the mortality effects of rice price variations on adult and elderly women were mostly insignificant, with the possible exception of elderly women, whose mortality increased significantly with increases in rice prices with a one- year time lag. This suggests that, in contrast to mortality at earlier years of life, the mortality of men in adulthood and old age responded immediately to short-term economic stress, while the mortality of women in adult and elderly years was largely unaffected by such stress.
Table 17.3 presents the results of discrete-time event-history analysis of male and female mortality rates at different life stages, including not only local rice prices but also household landholding as the covariates. As shown in the table, the nature of the mortality effects of local rice price fluctuations does not change when household landholding is added to the model. With regard to the effects of household landholding (per kin household member), three things stand out in the table.
First, there are significant and negative interactions between rice price and landholding, as well as between landholding and time periods in female infant mortality. This suggests that, while infant girls were highly vulnerable to short-term economic stress caused by changes in local economic conditions, their likelihood of death was much lower if they lived in households with resources and wealth. This tendency was especially strong in the period from 1760—99 and, though to a lesser extent, in the period from 1800—30. These two sub-periods include the decades of the two most serious and prolonged famines (1780s and 1830s). Thus, it may have been the case that the health benefits that infant girls enjoyed by living in a wealthy household were especially important when the economic condition in the community deteriorated seriously due to large-scale famines. Putting it differently, the survival chances of infant girls in economically deprived households were much worse when local economic conditions deteriorated, and this was especially the case in the times of acute economic difficulty.Second, household landholding significantly reduced the mortality of adult and elderly men as well as the mortality of elderly women, while adult female mortality was unaffected. In the case of elderly men, the mortality-reducing effect of landholding is also curvilinear. This suggests that, except for adult women, adults and the elderly were likely to benefit from the resources and wealth of their households in countering health risks. Such protective health effects of household resources were especially strong for elderly men.
Third, there are clear but opposite time trends in female child mortality and old-age mortality. Female child mortality decreased over time. By contrast, the
Table 17.3 Estimated effects of logged rice prices and household landholding on the probability of dying in the next one year by sex: Shimomoriya and Niita, 1716—1870
| Covariates | Infant | Child | Adult | Old age | ||||
| Male | Female | Male | Female | Male | Female | Male | Female | |
| Logged rice price, in current year | -0.201 | 2.416** | 0.057 | 0.317# | 0.392* | -0.006 | 0.730** | 0.076 |
| Landholding | 0.014 | 0.018 | 0.007 | -0.001 | -0.022** | 0.006 | -0.545** | -0.015* |
| Landholding, squared | — | — | — | — | — | — | 0.001* | — |
| Time perioc | ||||||||
| 1760-99 (P2) | -0.544# | 1.575** | -0.207 | -0.285 | -0.151 | -0.388** | 0.117 | 0.025 |
| 1800-39 (P3) | -0.103 | 0.441 | -0.333* | -0.336# | 0.573** | -0.264# | 0.163 | 0.191 |
| 1840-70 (P4) | -0.574 | 0.398 | -0.276 | -0.525** | 0.166 | -0.554** | 0.430* | 0.506** |
| Price*land- holding | — | -0.104** | — | — | bgcolor=white>—— | — | — | |
| P2*land- holding | — | -0.145** | — | — | — | — | — | — |
| P3*land- holding | — | -0.079# | — | — | — | — | — | — |
| P4*land- holding | — | -0.031 | — | — | — | — | — | — |
| Log-likelihood | -305.8 | -276.0 | -1,171.6 | -964.1 | -1,474.7 | -1,680.5 | -1,524.4 | -1,414.0 |
| χ2 | 6.48 | 35.94 | 177.12 | 176.49 | 94.31 | 50.46 | 158.55 | 171.33 |
| (d.f∙) | (7) | (11) | (11) | (11) | (18) | (18) | (15) | (14) |
| p>χ | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| No. of observations | 1,129 | 1,062 | 11,592 | 10,904 | 28,142 | 26,800 | 8,952 | 8,107 |
| No. of events | 88 | 83 | 266 | 216 | 270 | 313 | 389 | 366 |
| No. of individuals | 1,129 | 1,062 | 1,487 | 1,481 | 2,148 | 2,041 | 849 | 746 |
Significant at 0.01 level.
* Significant at 0.05 level.
# Significant at 0.10 level.
Notes: The analysis above controls for current age, residing village, and the number of kin members in household. For adult and old-age mortality, the effect of marital status of index individual is also controlled. The coefficients were estimated by using 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.
mortality of elderly men and women increased as time passed. We speculate that the loosening of rigid social (and intra-household) hierarchies by age and gender during the last years of the Tokugawa regime might have worked in favour of the health of female children, who were the most powerless members of the family and household, while it worked against the elderly, who had enjoyed respect and esteem under the traditional family system.
5.2