Household, Lineage, and the Market During the Tokugawa Period

The household and the family system are at the center of the relationship between the demand and supply of labor embodying either high or low levels of population quality during the late Tokugawa period. The reasons are various but can be summarized in terms of the following considerations: for most individuals work meant work in an enterprise managed by one's family or by the household within which one resided; in cases in which household or family members worked outside the home it was the household head, not the individual, who tended to make contract arrangements concerning remuneration and conditions of work. That household and family are the center of the supply and demand for population quality in late Tokugawa has a crucial corollary: those agents demanding workers of a given level of population quality

were also those investing in those workers and therefore supplying them. The incentive to enhance quality was built in—internalized, so to speak—since the family was simultaneously operating on both sides of the market.

During the late Tokugawa period both changes in the supply of labor conditioned by a decline in fertility among rural households and changes in demand for labor conditioned by an increase in the demand for labor that possessed a wide range of skills or could be quickly trained in the requisite skills encouraged an improvement in population quality. As a result, the role of the market in enhancing population quality was strengthened. This trend occurred in many sectors of the labor market dominated by family-run employment; and it also occurred in segments of the market in which bargaining between households by and large shaped labor contracts. Moreover, in the sectors of the labor market where households played a less pervasive role—for instance, in the market in which the samurai supplied labor services—deterioration in fief finances due to the burden of sankin kotai also tended to encourage a favorable trend in population quality.

Since family and household play a crucial role in this argument we must say something about the stem family system that by the late Tokugawa period provided the basic set of rules within which families and households formed.^[5] What is a stem family system? It is a system characterized by the following rules: (1) a spouse is brought into the family by one and only one offspring in each generation; (2) succession of the family headship falls on the offspring who has married within the family (or to the married couple); (3) inheritance, which is unequal, favors the single heir/successor; and (4) the family's organizational form passes through an alternating cycle of conjugal phases followed by a stem phase (in which the junior and senior conjugal units reside together) followed by a conjugal phase, and so forth. The transition from the stem to the conjugal phase is marked by the death of the last member of the senior couple; the transition from the conjugal to the stem phase is marked by the marriage of the heir.

That demography conditions the actual realization of a stem family system ideal is apparent from condition (4), for adult mortality and the timing of marriage play a critical role in lengthening or shortening the two phases of the alternating cycle. But demography conditions the operation of the family system in other ways. What if a family finds itself unable to biologically produce an heir, or produces an heir who is too young to assume the family headship at the same time the older couple

wishes to "retire" and commence the stem phase of the cycle? ^[6] Had there not been a religious ideology buttressing the family system in Japan, families might have been content to calmly contemplate the possibility of their lines dying out. But because ancestor worship was deeply entrenched in Japan—in part because of Buddhism, in part because of long-standing domestic spiritual traditions—families had a strong aversion to the demise of lines (although lines did die out with some frequency among the poorer landless houses). Under the ideology of ancestor worship the Japanese family was conceived of as dynastic as well as stem, with the lineage extending through past and future generations through the worship of one's ancestors. For this reason securing a successor to assume the headship and attend to the family's relics and religious artifacts was a matter of paramount concern. And hence in securing heirs, families resorted to a variety of fictions. For instance, males not related to the family by blood ties were in-adopted as "sons." In this way family lines could be perpetuated even if biology failed. But in practice biology limited most families simply because relying on the market for in-adoption was risky—because supply might be limited at the time a household found itself searching for an heir; because in-adopted heirs might be less inclined to attend to the religious rites of the ancestor cult than children conditioned to do so from birth; and because an individual brought in to the house through in-adoption was not trained in how to operate the family's enterprise (if it had an enterprise to operate) from an early age.

Still families did employ in-adoption on occasion, even bypassing biologically produced putative heirs on occasion.^[7] Why? Part of the problem was demographic and related to the timing of the transition from conjugal to stem phases of the cycle. But economics played a role as well. For it must be kept in mind that the actual working of the system was constrained by more than demographic probabilities. Most families were both demographic and economic units in the sense that they managed family enterprises like farms or craft production shops. If the biologically produced putative heir could not shoulder the task of managing the household economy in a highly competitive economy in which households vied with one another for land and rank, then the household head might be inclined to pass over his own son in favor of a more competent person unrelated by kinship.

Thus economics and demography constrained the family system in practice, if not in theory. But how actual adjustments were made very much depended on the social status, the class status, of the family. Here

I differentiate between two opposing class types: the peasant and the samurai warrior/bureaucrat. The reasons for differentiating in this way between two extreme types are both economic and ideological: economic, because in the peasant case household and enterprise overlapped and in the samurai case they did not; ideological, because in principle the code of behavior applicable to the representative samurai household was at variance with the code of behavior expected of the typical peasant household. I do not consider here merchant and craft families since they are less sharply opposed than the two types I focus on and since they were less numerous and have less bearing on my general argument.

In the case of the samurai the head of the house was invariably male as it was the male—not the female, who concentrated on the raising of a successor and domestic chores—who performed the tasks and maintained the rank and prestige of the house on which its fortunes rested. And among members of the bureaucrat/warrior class the income enjoyed by the house (ie ) was a function of the rank that that male enjoyed with the domain's elite. K. Yamamura (1974) shows that there were marked differences in family support income allotments (measured in koku of rice allotted to the house) according to the position of the head. And while at the beginning of the Tokugawa period the position of the house was hereditary, financial pressures on fief coffers due to the burdens of sankin kotai increasingly forced fiefs to become more efficient in allocation of scarce resources and therefore to seek out talented administrators among those retainers best qualified as signaled by performance in fief schools designed to train the retainer class. During the latter half of the Tokugawa period in many fiefs samurai were promoted upward and downward on the basis of ability and performance on examinations. As fiefs found their coffers depleted, many samurai, especially those of low rank, found their rice stipends inadequate for main-raining a large family in comparative luxury. And since their real incomes fluctuated with the price of rice on the Osaka market, in periods when rice prices declined the marginal retainers found themselves all the more squeezed. Hence is it surprising that family sizes among the samurai class declined and that many samurai found themselves without biologically produced sons or in need of a talented successor to replace a less competent putative biological heir? For these reasons the proportion of in-adopted family heads (yoshi ) was quite high (Yamamura 1974). In short, there was a long slow secular trend among the retainer class toward reduction in fertility—with an attendant substitution of quality for quantity of children—and an increasing emphasis on

ability and training in fief schools. Both factors tended to enhance population quality among members of this class. And this trend was directly related to changes in product markets like the market for rice and in financial markets where fiefs were forced to borrow from merchants to secure funds for their activities.

By contrast, among the peasant class the typical house served as a unit for organizing domestic activities and as a production unit in which all members regardless of sex, except the very young and the very old, were expected to perform the tasks of physical work on which the incomes of the house as a group depended. There was sexual division of labor—women devoted more time to domestic chores than did men and women were likely to engage in by-employments involving spinning and weaving (see Saito 1991)—but this division concerned differentiation within the categories of potential income-generating activities and did not draw a rigid line between those who restricted their activities exclusively to domestic work and those who engaged in market activities. Hence the de facto status of women depended on the actual activities they engaged in. If she possessed sufficient acumen and physical prowess, the mistress of a peasant ie might command greater authority than her husband. For this reason it is not surprising that the headship of the ie might devolve onto the mistress of the house or onto a daughter. Indeed Y. Hayami (1983) shows that male primogeniture, which is the form of inheritance usually associated with the stem family system in Japan, was not as common as was formerly believed. In some areas, for example, families employed ultimogeniture, and in some villages in the Northeast succession passed to the last born child regardless of sex (ane katoku ). The point can now be made by contrasting peasant and samurai forms of the stem family system: because the peasant house tended to be both a kinship and a production unit, it displayed far more flexibility in gender differentiation and inheritance than did the samurai household.

Now, as with many samurai houses, but for somewhat different reasons, there was a secular trend among the peasant household sector toward an improvement in population quality during the late Tokugawa period. The reasons lie with both supply and demand, and since peasant houses tended to simultaneously operate on both sides of the market the conceptual distinction I make here between supply and demand is somewhat artificial. In any case, in many regions of Japan, especially in rice-producing areas, there was supply side drift as fertility declined, encouraging a substitution of quality for quantity. Why did this occur?

Since the occasional extant Buddhist temple register allows us to reconstruct with considerable accuracy for each village covered in the registers the age-specific pattern of fertility for women, we know that in most cases interbirth intervals were long, overall marital fertility was low, and life expectancies at age one were quite high. Findings on fertility and mortality levels for a sample of late Tokugawa villages appears in table 15. As can be seen, fertility levels, although not low by modern Japanese standards, are modest, and life expectancies tend to be moderate to high although variability is considerable. Moreover, since use of the register data also allows us to reconstruct the sex ratios for surviving offspring, we have evidence suggesting that infanticide (or mabiki , literally winnowing out) was practiced in some villages, especially against girls (although there is some debate on whether infanticide was also systematically used against male issue in order to secure sex balance among sibling sets). Since infanticide is an emotionally charged issue, it is not surprising that the issue of why fertility reached low levels in many districts of late Tokugawa Japan is one that generates heated controversy.^[8] Without attempting to settle the issue here, I will merely note that at least three positions have been staked out as to the development of a low fertility/moderate mortality regime in late Tokugawa Japan. T. Smith (1988: chap. 4) argues that the main factor was the desire to limit competition over who would succeed to the headship; S. Hanley and K. Yamamura (1977) argue that peasants motivated by a desire to emulate the conspicuous consumption prevailing among urban dwellers limited their fertility so as to enhance their standard of living; and C. Mosk (1983) attributes the behavior to a desire to maximize survivorship for each child not eliminated through infanticide, low fertility, and a long interbirth interval strategy to reduce the number of infants requiring care at any one time. In any event, regardless of which explanation or combination of explanations is adopted, it is reasonably clear that there was a shift from a high fertility regime with frequent mortality crises and significant population growth during the early Tokugawa period to a regime of moderate fertility and moderate mortality and virtually no population growth during the late Tokugawa period (see Kalland and Pedersen 1984). Not all the regions of Japan experienced this sea change in demographic regime to the same extent, but the general pattern is clear enough. And from the point of view of this study the most important consequence of this secular change was a substitution of quality for quantity among children, as couples who restricted the overall size of their families were able to

devote more resources, more foodstuffs, and more parental time to those whom they chose to raise and train in the techniques of agriculture.

The argument involving fertility concerns the supply of population quality, but secular drift in demand was also important. On the demand side two forces were active: technological changes in agriculture that increased the importance of skill formation, especially in rice cultivation; and the expansion of by-employment opportunities in rural districts. For instance, in rice production new seed varieties were developed especially in the Southwest where experimentation with fresh strains was encouraged by the relatively benevolent climate, making double-cropping of rice possible and hence the risk of total crop failure minuscule. Moreover, the market for soil-enhancing fertilizers like fish cakes and dried sardines and the like expanded as rice cultivation spread and increased fertilizer usage improved the inherent productivity of land. And new threshing machinery like the bamboo semba-koki was introduced. Most of these innovations did not save on labor as did the mechanization of agriculture in nineteenth- and twentieth-century North America. Rather, they increased the demand for skilled labor that could efficiently choose from among a growing variety of seed varieties, fertilizers, and harvesting techniques (Smith 1959, 1988). Care and knowledge increasingly commanded a premium as is attested by the growing prevalence of the practice of transplanting the rice plants to the fields in even rows to maximize absorption of the sun's energy. Moreover, since rice was taxed by fiefs but industrial crops were not, wealthier farmers began to divert their activities to a more diverse set of crops and even to craft production aimed at distant urban markets. The result was a growth in by-employments. Especially common in late Tokugawa was the practice of dekasegi (temporary) migration to a small silk filature, or the like, for a period of a year or two. Girls with a knowledge of silk raising and spinning techniques were especially in demand in regions like Gifu and Nagano prefectures in central Japan where the growing of mulberry leaves on which silk-generating cocoons were fed was common. In short, changes on the demand side of the labor market in late Tokugawa Japan increased the demand for population quality and hence proceeded in tandem with supply changes associated with the secular drift toward moderate fertility. Indeed, A. Hayami has argued that the increase in demand for female labor in by-employments actually worked to depress fertility by raising mean ages of first marriage in many late Tokugawa villages (see Mosk 1995a, 1995b). But as a perusal

of table 15 makes clear, there was substantial regional variability in demographic regimes.

The issue of regional variability brings us to the question of the national and regional standard of living defined in terms of opulence. There has been a long-standing debate about the standard of living that was originally fueled by the question of how Japan was able to become the first country outside of the European cultural zone to successfully industrialize. One answer has been that her preindustrial income level was comparable to that of preindustrial Europe; as we have seen, recent research on late Tokugawa period village demography has suggested that fertility and mortality levels were moderate and some scholars have argued that this demographic regime was either a cause or a consequence (or both) of a reasonably good preindustrial level of income per capita. Backward projection of estimates of income per capita with real income growth rates has been commonly used as a method of coming up with figures for per capita income in the late Tokugawa and early Meiji period. Using this basic approach, S. Kuznets (1971) came up with a figure of $74 in 1965 U.S. dollars; K. Ohkawa, with $172 and $251 in 1970 U.S. dollars, the latter figure adjusted for purchasing power parity; and Y. Yasuba (1987), with $268 in 1970 U.S. dollars. By comparison, N. Crafts (1983) estimates the following for per capita income in 1970 U.S. dollars for various European countries on the verge of industrializing and having reasonably reliable data (date for estimate in parentheses): Great Britain (1760), $399; France (1830), $343; Germany (1850), $418; Italy (1860), $451; and Russia (1890), $276. With the exception of Russia, it can be seen that even the optimistic estimates for Japan like those of Yasuba (1986, 1987) put Japan below European countries in the decade before they began to industrialize. In my opinion this is not as surprising as one might think. For income per capita is an opulence measure of the standard of living, and my thesis is that if we were to measure Japan's income in the 1850s in terms of capabilities or in terms of population quality it would be considerable and that population quality, rather than the standard of living defined in opulence terms, is relevant for successful industrialization. But there is an additional consideration: the regional distribution of income. By establishing fief barriers to the diffusion of best practice technique, the bakuhan system kept pent up in the Southwest a variety of rice production techniques. Had the farmers in the remainder of the country known about these techniques, the national level of per capita income might have been considerably higher.

And of course climate played a role at the regional level both in terms of average levels of per capita output and in terms of the willingness to experiment with new seed varieties and fertilizers.

Hard data on the regional distribution of income are not easily obtained. But it is possible to come up with a few rough indicators. Hanley and Yamamura (1977) provide figures on rice output (in koku) and population for sixty-eight regions (kuni ) of Japan around 1700 and 1870. They are skeptical about the reliability of these figures, especially for the latter year, but it is worth exploring what the figures reveal about variation in output per head. I ranked the kuni by the koku per head figures and then grouped the regions into seven classes, for each of which I calculated the average koku per head. (There are 10 kuni in each group except in that with the lowest level of income for which there are eight regions; the index given over on the far right column below was secured by setting the figure ca. 1700 at 100 and then calculating the 1870 figure relative to it. Kuni in both years are grouped on the basis of the 1700 figures.)

Substantial cross-sectional variation in agricultural productivity is evident from these figures. For what they are worth, the data also suggest that the differentials did not close substantially during the Tokugawa period. However, Hanley and Yamamura (1977) warn that the koku figures for the kuni were not revised frequently so it is possible that labor productivity increased more than we have estimated in the less productive regions. But there is no reason to believe convergence occurred during the late Tokugawa period.

By contrast, rapid convergence took place during the Meiji era (1868-1912). To see this consider the following figures for rice yield per hectare in Saga prefecture in southwestern Kyushu and for rice yield per 10 hectares in Tochigi prefecture, which lies in the northeast of the main island, Honshu, of Japan (Tsuchiya 1976: 60).

How typical the Saga and Tochigi figures given here are for the regions in which they lie, Southwest and Northeast, respectively, is open to debate, but the picture of convergence drawn with these data accords with the descriptive literature on government and landlord promotion of the diffusion of best practice technique during the Meiji era.

That regional variation in climate, soil quality, and methods of farming generated substantial geographic variation in yields can also be demonstrated by examining the capacity of different regions to sustain population increase during the late Tokugawa period. Hence while for the country as a whole population growth over the 1721-1846 period was virtually nonexistent, there was considerable regional variation in growth rates. H. Kito (1983: 17 if.) gives figures for the fourteen major regions of Japan over the period 1721-1746 and I have taken these data and calculated simple unweighted averages of population group rates in percentage terms for groups of these regions classified by an overall index of the warmth of weather (there are five regions in the first two groups and four in the third group, and the northeastern regions rank low in terms of the index while the southwestern districts rank high).

The term "crisis years" refers to the periods of food shortages, especially to the major famine era, including the Kyoho (1732-1733), Tenmei (1783-1786), and Tempo (1836-1838) famines. Note that regional variation in population growth rates is substantial during crisis periods but not during normal years. Thus whatever is checking population growth is related to a realized or potential threat of food shortage. That

this appears to be the case makes one a bit skeptical of explanations for low fertility that emphasize a positive linkage between the level of income per capita and the willingness to restrict fertility.

In sum, the family lay at the center of the adjustment of population quality to changes in labor markets: it was the unit that determined supply, and for the great mass of the people who were peasant producers it was also the unit determining demand for skills, work capacity, and capabilities. In the regions of the country where technological progress in rice farming or the expansion in the demand for by-employments was especially vigorous, there seems to have been a secular trend toward improvement in population quality. This improvement was driven by a decline in fertility and an increase in the amount and quality of training accorded offspring. That the market was as effective as it was in enhancing population quality is one reason why entitlements played a more circumscribed role. However, as we shall now see, entitlements did play a role, especially in the fiefs where food crises occurred throughout the late Tokugawa period.