首页 | 学科·学界 | 社会现象 | 文章·争鸣 | 读书 | 学者家园 | 文献服务 | 数据服务 | 中心网刊 | Blog | WIKI | 网上调查 | 社会论坛
  当前位置:首页 >> 读书 >> Re-Drawing Boundaries: Work, Households, and Gender in China >> 正文
分类导读
社会学
人类学
社会工作
社会科学综合
15. Household Economies in Transitional Times

图书名称:Re-Drawing Boundaries: Work, Households, and Gender in China
图书作者:Barbara Entwisle and Gail E. Henderson    ISBN:
出版社:Berkeley: University of California Press    出版日期:2000年

Barbara Entwisle, Susan E. Short, Zhai Fengying, and Ma Linmao

The shift towards a market economy, the rise of household businesses, and the industrialization of the countryside are all components of economic change currently underway in China. In what ways are households participating in these shifts? Does it depend on the size of the household, or gender composition? The answer is: we know very little. In China, as everywhere, statistics on work are generally reported for individuals, not for households. It is easy to know the fraction of the labor force in agriculture, for example, but more difficult to know the fraction of households exclusively in agriculture, exclusively outside of agriculture, and straddling this important economic divide. Similarly, studies of household labor allocation mostly look at the involvement of individual members in farm and offfarm activities in rural areas (e.g., Benjamin and Brandt 1995; Nee 1996; D. T. Yang 1997), yielding only indirect information about the degree to which households might be involved in both.

The same is true for other key divides in the Chinese economy. Employment statistics show the involvement of individual workers in household businesses (getihu) as compared to economic activities organized outside of the household, and the involvement of individual workers in private as compared to state and collective sectors of the economy (e.g., see table 2.1 in chap. 2). However, analyses of which households start up household businesses—whether getihu or household-based activities more broadly defined—have considered this outcome independently of other economic activities in which the household and its members might be engaged (e.g., Entwisle et al. 1995; Short and Zhai 1996). The determinants of participation in the state, collective, and private sectors have been investigated for individuals only (e.g., chap. 7). Little is known about the relationship between households and work in the context of economic change (also see Moen and Wethington 1992).

Households are of particular interest now because, in contrast to a waning influence most everywhere else in the world, in China economic reforms renewed and enlarged the role of households as economic units. Rural Chinese households have received particular attention in this respect. Decollectivization transferred land to small holders and returned control over labor power and many aspects of economic production to households (Judd 1994; Walder 1996). For many researchers, rural households are a basic economic unit (e.g., Benjamin and Brandt 1995; Judd 1994; Nee 1996; D.T. Yang 1997). Urban households and their economic activities have not received as much attention, perhaps because they have not been a target of economic policy. Nevertheless, with the weakening of state control over labor allocation and the gradual opening of labor markets in urban areas, household members have more opportunity to coordinate their work choices in urban areas as well. Further, expanding markets for consumer products and services provide a context in which household-based entrepreneurs can succeed. Mann (chap. 1) comments that of all historical work structures, the household has been the one most resistant to change.

Many aspects of the relationship between households and work are of potential interest. In this chapter we investigate the degree to which households specialize or diversify their work activities, with particular attention to the consequences of household size and gender composition. Household economic pattern need not reflect an explicit household strategy or deliberate action on the part of household members. It may simply be the aggregation of individual outcomes, independently chosen. But specialization and diversification at the household level could reflect coordination among members, possibly a jointly chosen economic strategy. For example, members may diversify the types of work they do to minimize risk to household income. Alternatively, they may maximize household income by all working in the most lucrative sectors, regardless of risk. Of course, households may also specialize or diversify to meet other, non-income-related goals. Little is known about households' involvement in this type of coordination, or the degree to which households are specialized or diversified. Until now, attempts to measure household economic patterns and to quantify their distribution have addressed small samples or local populations (e.g., Johnson 1993; Judd 1994; Ritchie 1994).

We use China Health and Nutrition Survey (CHNS) data, covering many households in diverse settings, to examine the extent to which households specialize or diversify. We document household economic patterns in three ways, orienting around whether or not households bridge three major dimensions of economic structure and change:occupation and industry (agricultural or not); mode of production (household-based or not); and employment sector (private, or state and collective). We describe urban/rural differences in patterns of specialization and diversification, and consider household size and gender composition as potential determinants of these patterns. Of particular interest is whether household involvement in diversified or specialized economic activities reflects coordinated activity, or even explicit household strategy. Our approach to this question is indirect. Whereas the CHNS has detailed data on income-earning activities of all household members, it has no information on the goals, strategies, and decision making that might lie behind these activities. Instead of examining the question of coordination directly, we simulate what household economic patterns would look like if household members made independent choices and simply obtained the best possible job given their individual circumstances, and then compare the simulated patterns with those observed in the data. If observed patterns follow the simulated patterns, then it would appear that household members are acting independently and that there is little if any collective decision making. If observed patterns depart from the simulated patterns, then the evidence suggests that there might be some coordination in workchoices, possibly a jointly chosen strategy. As we will show, the latter is the case in the CHNS data.

HOUSEHOLD ECONOMIC PATTERNS

The context for our analysis is China in 1989. Economic reforms had created a climate characterized by new, diverse, and dynamically changing economic opportunities (Davis and Harrell 1993; Judd 1994). Whereas the collective regime had enforced a uniformity in household activities and arrangements, by 1989 it was possible for households to organize themselves in diverse ways (Croll 1994; Odgaard 1992). Administrative practices binding the rural population to agricultural pursuits and to specific villages and households had loosened (chaps. 11 and 12; Johnson 1993). Households in rural villages could diversify by starting up a small business (e.g., a small restaurant, a cycle repair shop, a clothes shop, a tractor transport business, a noodle-making shop; Judd 1994, 138), having one or more members obtain employment in a local industry, or sending one or more members to some other locale for nonfarm employment (e.g., chap. 10) while continuing to farm the assigned plots. Urban households could diversify by gardening or continuing to farm assigned plots, starting up a household business such as a small shop, or finding employment for one or more household members in a private company or joint venture while continuing to keep at least one member in a state or collective sector job.[1] After rapid economic expansion in the early 1980s, growth had fallen off by 1989 in all sectors of the economy, in both urban and rural areas (Perkins 1994). As it was the year of the Tiananmen Square pro-democracy movement, 1989 was also a politically uncertain year.

Our first goal is to develop a broad description—to use survey data to characterize economic patterns in a large and diverse sample of urban and rural households. We do so with respect to three overlapping dimensions of economic structure and change. The first is occupational mix—involvement in agricultural versus nonagricultural jobs. The second is mode of production—whether work activities are organized by the household, such as household-based farming or household businesses, or organized outside of the household. The third is employment sector—dependence on the state and collective sectors versus the private sector. There is overlap between these three dimensions, of course. Private sector employment, for example, may be agricultural or not, household-based or not. Employment by the state, in contrast, precludes a household-based mode of production, and since the decollectivization of agriculture, has almost always implied nonagricultural work as well.

Our consideration of specialization and diversification with respect to multiple dimensions of economic structure and change is a departure from standard practice. Studies of household economic behavior (in China and elsewhere) have typically looked at the involvement of farm households in nonfarm activities (e.g., Benjamin and Brandt 1995; Besteman 1989; Judd 1994; Quataert 1985; Ritchie 1994; Rosenfeld 1985; Rozelle and Jiang 1994; D. T. Yang 1997). A few have examined diversification with respect to mode of production, particularly in urban areas (e.g., Hareven 1991; Tiefenthaler 1997). None to our knowledge has considered specialization or diversification with respect to employment sector.

There are several reasons for considering households in relation to employment sector in a study of contemporary China. The Chinese economy has a dual structure right now, with redistributive elements coexisting with market elements, and at times the boundaries are unclear (Perkins 1994; Walder 1992a). Workers may labor in the state or collective sectors, the private sector, and even both private and either state or collective(though this last possibility has not been examined much). In parallel fashion, households may derive their income exclusively through jobs held by members in the state and collective sectors or in the private sector; or they may derive their income through jobs held by members in both. If we are to understand household behavior in relation to the changing structure of the larger economy, then, it is important to look at employment sector. Risks and rewards differ across these sectors, with state and collective jobs generally more secure but paying less than private sector jobs. Diversifying across employment sector thus has some potential benefit for households, just as it does for larger economic units (Stark 1996).

Our data come from the 1989 wave of the China Health and Nutrition Survey (CHNS89 ), a collaborative effort between the Carolina Population Center (University of North Carolina at Chapel Hill) and the Institute of Nutrition and Food Hygiene (Chinese Academy of Preventive Medicine). Data were collected from eight provinces in China—Liaoning, Jiangsu, Shandong, Henan, Hubei, Hunan, Guangxi, and Guizhou. Sampling was based on a stratified multistage cluster design. Two cities in each province were selected for the sample, (usually) the provincial capital and a lowerincome city. Within these cities, urban neighborhoods and suburban villages were selected randomly. Also, in each province counties were stratified by income and four counties selected. Villages and town neighborhoods were sampled in each county selected. A sample of 20 households was drawn from the official household registration system (hukou) for each village and neighborhood. Household membership, determined at the beginning of the interview, included individuals formally registered with the household (including those working elsewhere at the time of the interview) and others who either lived in the household on a long-term basis or had economic relations with the household, such as dependent students or those who worked out of town and shared their income. Our analyses are based on data for 3792 households—1275 urban and 2517 rural.

Our approach is aggregative. Households are classified based on the work of their members. The survey includes sections with detailed questions on the involvement of each household member in a variety of activities, including wage work, farming, gardening, fishing, livestock raising, and household businesses. The questionnaire is designed to identify multiple work activities, and indeed, many people engage in multiple activities (Entwisle et al. 1995; also see table 15.4 in this chapter). For every individual in the sample we classify every type of work reported as either agricultural or nonagricultural, public or private, and household-based or not. We take account of any income-earning activity, whether formal or informal, even if the income earned is small or some of the product is for home consumption.

To create household-level measures of specialization and diversification, we first classify the economic activities of household members and then characterize households according to the joint pattern of activities.[2] Consider employment sector. For each household member, we determine involvement in the state or collective sector (including township and village enterprises), private sector, both, or neither. Then, aggregating over all household members, we see whether there is employment in the public sector only, the private sector only, both, or neither. A specialized-household isone in which all household members work in a single sector, whether state, collective, or private.[3] A diversified household is one in which some members work in the public sector and others in the private, or one or more members

 
<COL align=left> <COL align=right> <COL align=right> <COL align=right>
Table 15.1 Household Diversification across Employment Sector, Mode of Production, and Occupation/Industry: Observed Patterns
  Employment Sector Mode of Production Occupation/ Industry
SOURCE: China Health and Nutrition Survey, 1989.
Urban Households (N = 1275)      
 Specialized household (%) 69.7 82.0 69.6
 Diversified household (%) 30.3 18.0 30.4
  Not internally specialized 4.7 9.9 7.2
  Internally specialized 25.6 8.1 23.2
Rural Households (N = 2517)      
 Specialized household (%) 79.0 65.4 61.2
 Diversified household (%) 21.0 34.6 38.8
  Not internally specialized 11.4 21.6 25.4
  Internally specialized 9.6 13.0 13.4
work in both. Among diversified households, we also consider the division of labor within the household and distinguish between those in which at least one member works in the public sector and in the private sector (not internally specialized), and those in which each working member works exclusively in either the public or private sector (internally specialized). Measures of specialization and diversification with respect to occupational sector and mode of production are formed analogously.

As shown in table 15.1, significant fractions of households are diversified. Between 21 and 39 percent of the rural households in the CHNS sample are diversified, depending on whether employment sector, mode of production, or occupation/industry is under consideration. Between 18 and 30 percent of urban households are diversified. Through the work patterns of individual members, whether their work is coordinated or not, households are participating in and contributing to major shifts in the economy. Indeed, it may surprise some readers that so many urban households in our sample bridge the agrarian and industrial economies. This is because the CHNS sample design disproportionately represents the experience of town households.[4] Nevertheless, it is still noteworthy that so many of these urban households span occupational sectors. As noted, previous studies of household diversification with respect to occupation and industry have focused on rural households exclusively (e.g., Besteman 1989; Quataert 1985; Ritchie 1994; Rosenfeld 1985; Rozelle and Jiang 1994).

The extent of diversification differs between urban and rural areas. Diversification across mode of production shows the most pronounced urban/rural differences in the CHNS sample: 35 percent of rural households obtain income from a combination of household-based and nonhousehold-based activities, compared to 18 percent of urban households. Diversification across occupation and industry follows the same pattern, although the differences are less striking: 39 percent of rural households span agricultural and nonagricultural occupations, compared to 30 percent of urban households. If we had a representative sample of urban households, undoubtedly this difference would be wider and probably would be associated with mode of production also. In contrast, differences with respect to employment sector are reversed: 30 percent of urban households obtain income from private sector and state or collective sector activities, compared to 21 percent of rural households. Patterns of economic activity among households thus reflect the nature of change occurring within the larger economic context in which they are embedded: in rural areas, industrialization; in urban areas, the shift towards a market economy.

There are also marked differences in how households diversify across urban and rural areas. When urban households diversify, it is because household members specialize in different types of economic activity. When rural households diversify, it is because the activities of at least one member span a major economic divide. Take occupation, for example. In urban areas, households gaining income from agricultural and nonagricultural activities are three times more likely than not to have a division of labor such that some members specialize in agricultural activities exclusively and other members specialize in nonagricultural ones. In rural areas, such complementarities are only about half as common, relatively speaking. The “parttime peasant” is thus a rural phenomenon. Although the pattern is not quite so clear for employment sector and mode of production as it is for occupation, the basic point still holds.

COORDINATED ACTION OR NOT?

Do patterns of economic activity observed at the household level reflect a household strategy, with coordination among household members? When asked “Overall, what do you think is the best job to have? If all the adults in your family had this job, would it be best?” participants in focus group interviews conducted in towns and villages in Hubei province in 1995 gave a variety of responses. Several participants identified a particular type of work, such as state sector work, as best for all family members. Still, many recommended that family members do different types of work. For example, one man commented, “Running an enterprise yourself is very dangerous, so everyone in the family shouldn't be doing this together. Don't put all your eggs in one basket.” Despite differing opinions, one thing was clear: On the whole, participants were comfortable responding in terms of what a family or household should do. Only two participants seemed to think in terms of individuals making their own choices. But even of these, one did not neglect to mention the family: “Everyone in my family should do their own thing. That way they can complement each other. It is secure.”

Patterns of economic activity observed at the household level may reflect a household strategy, with coordination among household members. Indeed, those who write about rural economic life, in China and elsewhere, commonly assume that households are economic units and theorize about household economic strategies (e.g., Benjamin and Brandt 1995; Chayonov 1966; Hareven 1991; Judd 1994; Nee 1996; Rosenzweig 1980; D.T.Yang 1997). It is not clear how one would go about proving the existence of a household strategy, however (Moen and Wethington 1992). Household members may not always be able to describe and explain a household strategy, even if one exists. Further, a claim by household members that they are acting in accord with a household strategy might be nothing more than a post hoc explanation of choices made for other reasons. The focus group respondents who were just quoted seemed perfectly comfortable with the notion of household strategies, but of course without more observation and analysis, we cannot know whether they are reflecting cultural myths or everyday realities. Our approach to the problem of interpretation—is there really coordinated action?—is indirect. As explained in the next several paragraphs, we compare empirical patterns with those predicted assuming coordinated economic activity among household members, and with those predicted assuming that members act independently of one another.

If household members do coordinate their economic activities with respect to jointly held goals and strategies for reaching them, then it should be possible to make some predictions about patterns observed at the household level. We are particularly interested in household response to risk. Risk reduction is a relevant consideration throughout China, and in urban as well as rural areas. Profound economic transitions are underway—a situation that creates an environment of uncertainty for households as well as for larger economic units (Guthrie 1997; Stark 1996). Adding to this economic uncertainty are the political upheavals of the past 50 years, and the fact that certain kinds of economic activity—especially nonagricultural household businesses—have at times had serious negative political consequences for those engaged in them. Nee and Young (1991, 298) describe it this way: “Imagine what peasants must think when they listen to the exhortations of local cadres, who only a short time before preached the virtues of collectivism but now urge peasants to enrich themselves by pursuing private entrepreneurship oriented to the free market. It must surely cross the minds of some that such a reversal could be changed yet again, possibly in the near future.” To the extent that households do have strategies, this is an environment that would encourage strategies oriented towards risk reduction.

Diversification is a risk-reducing strategy (Guthrie 1997; Odgaard 1992; Rozelle and Jiang 1994), although not always (Fligstein 1991; Harrell 1993). Even a one-person household can be diversified, but the ability of households to minimize risks by diversifying will depend on their size: larger households are in a better position to diversify their “labor force” than smaller households. Because of the greater risks inherent in agriculture (Judd 1994; Rozelle and Jiang 1994), diversification may be especially appealing for rural households, and particularly with respect to occupation/ industry and mode of production. Further increasing the risks to peasants is the fact that they do not own the land assigned to them (Nee and Young 1991; Rozelle and Jiang 1994). In many contexts, migration is a potential element of a diversification strategy (chap. 10; Moen and Wethington 1992). In China, though restrictions have eased and in some cases the volume of migration is quite large (chaps. 11 and 12), migration rates still remain low by international standards. Diversification is more likely to be a strategy implemented in situ.[5]

But suppose there is no household strategy and no coordination among household members: each member simply obtains the best possible job, without regard to the economic activities of other members. If this is so, then the pattern of economic activity observed at the household level is simply an aggregation of individual outcomes. As an economic unit, the household is no more than the sum of its parts. We should be able to reproduce the aggregate patterns by appropriately clustering predictions made for individuals. As we will demonstrate formally, assuming household members make their work choices independently of one another, we would expect diversification to be more common in rural than in urban areas and among larger rather than smaller households. Note that the household strategy argument also predicted these effects. As a consequence, we cannot select between the household strategy hypothesis and the independent action hypothesis based on predicting a positive effect of household size on diversification, or predicting more diversification in rural than urban areas, and simply checking conformity to these predictions. We need to make our predictions more precise.

Our first step is to simulate what the distributions would look like if households were simply aggregations of individuals, with the work activities of each household member independent of the activities of other members. We do this by calculating separately for each of the three dimensions (occupation, mode of production, and employment sector), for urban and rural areas and for each given household size, the probability that a household of that size will be specialized or diversified; and among diversified households, we distinguish between those with and without internal specialization. The probabilities are based on the assumption of independence—that

 
<COL align=left> <COL align=right> <COL align=right> <COL align=right>
TABLE 15.2 Simulated Patterns of Household Diversification across Employment Sector, Mode of Production, and Occupation/Industry: Assuming Independence
  Employment Sector Mode of Production Occupation/Industry
SOURCE: China Health and Nutrition Survey, 1989, and formulas in chapter appendix.
Urban Households      
 Specialized household (%) 59.2 68.7 58.5
 Diversified household (%) 40.8 31.3 41.5
  Not internally specialized 6.1 11.8 8.8
  Internally specialized 34.7 19.5 32.6
Rural Households      
 Specialized household (%) 72.0 51.6 54.7
 Diversified household (%) 28.0 48.4 45.3
  Not internally specialized 14.3 26.5 26.5
  Internally specialized 13.7 21.9 18.9
the type of work done by any member of the household is independent of the type of work done by any other member. In these calculations we assume the likelihood that any individual engages in any particular activity is the sample mean, given place of residence. The chapter appendix presents the formulas that we use.

Table 15.2 presents the distributions of household economic patterns under the assumption of independence. In fact, the general patterns are fairly similar to those observed in the data, shown in table 15.1. But there are some interesting deviations. There is less household diversification and more household specialization observed in the data (table 15.1) than we see in the economic patterns predicted under the assumption of independence (table 15.2). This is true for urban and rural households, but a bit more so in the urban ones. It appears, then, that household members might coordinate their work activities, but to the extent they do, it is to concentrate effort in one sector of the economy rather than to diversify. Given the climate of risk in China, we expected to see diversification, not specialization, so this result comes as somewhat of a surprise. To explore these patterns further, we now turn to household size and its effects.

SIZE EFFECTS

Labor supply is central to economic models of labor allocation (Chayonov 1966; Rosenzweig 1980). Correspondingly, household size and structure are central to social and historical studies of households and economic change (e.g., Goode 1963; Hareven 1991; Kriedte, Medick, and Schlumbohm 1981). In many different studies of households in rural China, based on a variety of theoretical approaches and statistical specifications, household size affects the participation of individual members in nonfarm work activities, whether they are based in the household or not (e.g., Benjamin and Brandt 1995; Entwisle et al. 1995; Nee 1996; D. T. Yang 1997). None of these studies has considered that individuals might participate in multiple economic activities, however. Nor has any study modeled household economic pattern directly. Rather, to the extent that attention is given to the interrelated nature of labor allocation decisions, it is treated as a problem of endogeneity in predicting what individuals do (Benjamin and Brandt 1995; D. T. Yang 1997). In other words, in making a decision about work, one household member might be influenced by what another household member is doing, but there is no attempt to consider the household as a whole. Our approach is quite different. We treat the household as a collectivity and attempt to predict specialization and diversification directly.

Table 15.3 reports results from multinomial logistic regressions of household economic pattern (an unordered three-category variable that distinguishes two different ways that households can diversify from household specialization) on household size and type of place of residence. The coefficients reported in the table for “rural” show the effects of rural context on the log-odds of one household economic pattern versus another, controlling for household size. The coefficients reported for size represent the effect of each additional household member aged 15 or older, controlling for type of place of residence. Earlier, we predicted that larger households would be more diversified than smaller households, and the data confirm this prediction. Size has a consistently positive effect on household diversification versus household specialization (comparing diversified households having no internal specialization versus specialized households [column 1], and comparing diversified households having internal specialization versus specialized households [column 2]). Larger households are not only more likely to be diversified, they are also more likely to have all members specializing in one type of economic activity or another (comparing diversified households having internal specialization versus those with no internal specialization [column 3]). This pattern is clear for all three dimensions of economic activity.

The size effects just described are irrespective of whether the household is urban or rural. We wondered whether the effects might differ—whether size might make more of a difference in one context as opposed to another. There is good reason to expect this. For one thing, the degree to which households act as decision-making units may well vary. In urban areas, until recently, individuals and households had little direct influence over their

 
<COL align=left> <COL align=middle> <COL align=middle> <COL align=middle>
TABLE 15.3 Multinomial Logistic Regressions of Household Economic Pattern on Household Size and Place of Residence
  Diversified (no Specialization vs. Specilized Diversified (Internal Specilizaton) vs. Specialized Diversified (Internal Specialization) vs. Diversified (No Specilization)
p =.05 (two-tailed test).
Mode of production      
 Size 0.13∗ 0.33∗ 0.21∗
 Rural 1.02∗ 0.74∗ −0.29
 Constant −2.52∗ −3.41∗ −0.89∗
 χ2 211.4 with 4 df    
Occupation/industry      
 Size 0.15∗ 0.31∗ 0.16∗
 Rural 1.39∗ −0.42∗ −1.81∗
 Constant −2.72∗ −2.09∗ −0.63∗
 χ2 325.9 with 4 df    
Employment sector      
 Size 0.14∗ 0.30∗ 0.16∗
 Rural 0.76∗ −1.12∗ −1.87∗
 Constant −3.12∗ −1.94∗ 1.17∗
 χ2 272.8 with 4 df    
work assignments (chap. 7). Work registration bureaus matched individuals and their qualifications to specific jobs and their requirements. Initial job assignments were crucial. Opportunities for household members to coordinate their economic opportunities were thus quite limited (Whyte and Parish 1984), other than using connections (guanxi) to obtain the best initial placement, which often meant a job in a state-run work unit (danwei). Rural households, by contrast, have a tradition of coordinated action that even collectivization did not entirely undermine (Judd 1994; Parish and Whyte 1978; Stacey 1983).

A further reason for expecting urban and rural households to behave differently is that in urban areas individuals are more likely to be remunerated for their own work. To the extent that coordinated action on the part of household members is the result of one or two individuals making decisions about economic activities, a coordinated economic action could be more difficult to effect when individual members control significant shares of resources. Whereas in rural areas it is often assumed that the eldest males represent the household and make decisions on behalf of the household, especially concerning economic production, this assumption holds less sway in urban areas. Common budgets and joint decision making are less common in city and town families (Croll 1987; Siu 1993).

If the size effects are because of households acting as economic decisionmaking units, we might expect the size effects to be stronger for rural than for urban households. The evidence does not support this expectation, however (results not shown). The effects of size on diversification are the same for rural and urban households with respect to mode of production and with respect to occupation. There is some suggestion that size effects differ when it comes to diversification with respect to employment sector. Interestingly, in this instance, size effects are weaker in rural areas. The differences are not very pronounced, though, and achieve only marginal statistical significance.

As discussed earlier, interpretation of size effects depends on the proper comparison. Size effects are expected, regardless of whether or not household members coordinate their work activities. We need to compare the size effects in our data with those expected when household members act independently and do not coordinate their work choices. When we do so, what we find confirms our earlier results: Household diversification is less common than we would expect under the assumption of independence. With a few exceptions for very small households (one or two members aged 15 and older), the patterns observed earlier for all households also apply to households of every size. Further, differences between what is observed in the data and what is predicted assuming no coordination of economic activity increase with household size. For every dimension of economic structure and change, and for rural and urban households, the effects of size on diversification are weaker, not stronger, than predicted based on independent action. This is the reverse of what we expected. We thought that households would prefer to diversify and would be especially likely to do so when they are large. Instead, what we find is that households might prefer to specialize, especially when they are large.

GENDER COMPOSITION

Although we find specialization when we expected to find diversification, the results do suggest that household members might coordinate their income-earning activities. Of course, the latter inference can only be made if we have properly predicted the consequences of independent action on the part of household members. Assessing the validity of the inference requires us to consider the circumstances under which these predictions might be wrong. These predictions treat household members as interchangeable with respect to their participation in various income-earning activities. We know that this assumption is not realistic, and in this section, we give particular attention to gender composition.

In China and elsewhere, men and women often engage in different types of work. Historically, the ideal pattern for peasants was for men to work “outside” in the fields and for women to weave “inside” the house (chap. 1). With respect to the contemporary categories of interest in this chapter, “outside” work translates into nonfarm work, especially outside of the household; “inside” work translates into household-based work, including agricultural fieldwork to an increasing extent. “Outside” includes the state and collective sectors and some but not all of the private sector (e.g., it does not include getihu ). Corresponding to the general alignment of male with “outside” work, men are more likely than women to hold jobs in the state and collective sectors, to have jobs outside of agriculture, and to work outside of the household. Interestingly, men are also more likely than women to do more than one type of work, bridging major divides in the economy, most notably in rural areas (see table 15.4). This may be for very practical reasons revolving around women's many other responsibilities in the household: cooking, cleaning, doing laundry, taking care of children, and looking after the elderly. Or, it may reflect a greater pressure on women to align their gender and work identities. Or, it may reflect constraints on the range of work opportunities available to women.

Gender composition thus may be very relevant to household economic pattern. If household members plan together, then it should not surprise us that these plans take into account the economic opportunities available to specific members and preferences regarding appropriate work. To the extent that opportunities and preferences are specific to gender, opportunities to specialize or diversify at the household level will depend on gender composition as well as household size. Indeed, studies of labor allocation typically take into account the gender as well as the number of adult workers (e.g., Benjamin and Brandt 1995; Entwisle et al. 1995).

Earlier we found that household size increases the odds of household diversification as opposed to household specialization. If we rerun the analysis, distinguishing the effects of men from women, we find that increases in the number of men largely explain the effect of size on diversification (results not shown). Households with more men are more likely to diversify. However, the way in which households diversify depends on the number of women. Households with more women are more likely to specialize internally.[6]

To illustrate the size and nature of these effects, table 15.5 simulates household economic pattern, based on estimates of the effects of household size and gender composition derived from the empirical data, for households with four adult members, varying the gender makeup. We show results for households composed of one woman and three men, two women and two men, and three women and one man. Altogether, households of this size and these gender compositions comprise 20 percent of all urban

 
<COL align=left> <COL align=right> <COL align=right> <COL align=right> <COL align=right>
Table 15.4 Percentage Participation of Men and Women in Different Types of Economic Activity
  Urban Rural
  Men Women Men Women
SOURCE: China Health and Nutrition Survey, 1989.
Employment sector        
 State and collective only 24.8 18.7 7.2 4.0
 Private only 47.2 40.0 69.0 73.3
 Public and private both 1.8 2.0 6.0 3.3
 No income-earning activity 26.1 39.3 17.8 19.3
Occupation/Industry        
 Agricultural only 19.8 21.5 52.7 66.7
 Nonagricultural only 51.0 36.3 14.8 7.9
 Agricultural and non-agricultural both 3.1 2.9 14.8 6.1
 No income-earning activity 26.1 39.3 17.8 19.3
Mode of production        
 Household-based only 7.7 12.2 52.1 65.8
 Outside only 61.3 44.8 16.8 9.0
 Household and outside both 4.9 3.7 13.3 5.8
 No income-earning activity 26.1 39.3 17.8 19.3
N (Members aged 15 and older) 1901 2028 3806 3846
CHNS89 households, and 16 percent of all rural CHNS89 households—a minority, but a considerable one.

Table 15.5 shows that shifting from a household composed of three men and one woman to a household composed of one man and three women does affect overall level of diversification or specialization. In rural areas, although an increase in women relative to men tends not to affect household diversification overall, or even decreases it, it increases household diversification with internal specialization. Consider diversification over mode of production. A significant minority of rural households containing three men and one woman are likely to be diversified (38.1%). Reversing the sex composition to one man and three women has little effect on household diversification (39.1%). The change is more noticeable if we consider forms of diversification. A sex composition favoring men is almost twice as likely to produce diversification without internal specialization as diversification with internal specialization (24.0% vs. 14.1%). A sex composition favoring women is only a little more likely to do so (21.8% vs. 17.4%). Looking at it

<COL align=left> <COL align=right> <COL align=right> <COL align=right>
TABLE 15.5 Gender Composition and Household Economic Pattern for Households with Four Adults: Simulations Based on Multinomial Regression Analysis
  One Man, Three Women Two Men, Two Women Three Men, One Woman
SOURCE: China Health and Nutrition Survey, 1989, and estimates from a multinomial logistic regression analysis similar to that in Table 15.3 but distinguishing numbers of men and women.
Urban households      
 Mode of production (%)      
  Specialized household 81.8 79.1 75.7
  Diversified household 18.2 0.9 24.3
   Not internally specialized 8.1 11.0 14.6
   Internally specialized 10.0 9.9 9.8
 Occupation/Industry (%)      
  Specialized household 62.7 63.8 64.9
  Diversified household 37.3 36.2 35.1
   Not internally specialized 8.1 7.8 7.5
   Internally specialized 29.2 28.4 27.6
 Employment sector (%)      
  Specialized household 66.0 62.8 59.3
  Diversified household 34.0 37.2 40.7
   Not internally specialized 4.8 5.8 7.1
   Internally specialized 29.2 31.4 33.6
Rural households      
 Mode of production (%)      
  Specialized household 60.9 61.4 61.9
  Diversified household 39.1 38.6 38.1
   Not internally specialized 21.8 22.9 24.0
   Internally specialized 17.4 15.7 14.1
 Occupation/Industry (%)      
  Specialized household 57.8 57.2 56.3
  Diversified household 42.2 42.8 43.7
   Not internally specialized 24.7 27.2 29.8
   Internally specialized 17.5 15.7 13.9
 Employment sector (%)      
  Specialized household 77.9 76.7 75.5
  Diversified household 22.1 23.3 24.5
   Not internally specialized 10.7 12.0 13.3
   Internally specialized 11.4 11.3 11.2

277
another way, shifting the sex composition from one that favors men to one that favors women decreases diversification with no internal specialization (24.0% vs. 21.8%), and increases diversification with internal specialization (14.1% vs 17.4%).

In urban areas, there tends to be a tradeoff between specialization at the household level and specialization within diversified households. A shift in the gender composition of the household tends to affect these two patterns in opposite ways. For example, again consider household economic pattern with respect to mode of production. Urban households containing three men and one woman are likely to be specialized (75.7%). Reversing the sex composition increases household specialization by six percentage points (to 81.8%). The same positive effect is observed for household specialization with respect to employment sector, although this is counter-balanced to some extent by a negative effect on diversification with internal specialization (from 33.6% to 29.2%). The patterns are opposite for occupation/ industry, with increasing representation of women in the household decreasing overall specialization but increasing diversification with specialization. For urban households, then, a sex composition favoring women also tends to favor some type of specialization.

As noted before, an empirical relation between gender composition and household economic pattern does not by itself constitute evidence of a household strategy or of coordination between household members. This is because some relationship might be expected even if households are no more than aggregations of their members. Earlier we used observed participation rates for individuals to predict the amount of diversification at the household level that we would expect if household members behaved independently of one another. We do this again, now taking gender as well as size into account. We use observed participation rates for each gender to predict household economic pattern, assuming independence. Just as we found earlier that the effect of household size was weaker than expected based on the assumption of independent action, the effect of gender composition is also weaker than expected.

To demonstrate and illustrate this result, table 15.6 shows economic patterns for households containing four adults predicted under the assumption of independence. It can be compared to table 15.5, the patterns derived from the empirical data. In rural areas, shifts in gender composition lead to greater changes in household economic pattern under the assumption of independence than we find in the data. A change in gender composition from three men and one woman to one man and three women should increase the relative number of specialized households by 8 to 14 percentage points in rural areas (looking at all three dimensions of economic structure and change). The actual effects of gender composition, as summarized in table 15.5, fall short of this (1 to 4 percentage points). In

<COL align=left> <COL align=right> <COL align=right> <COL align=right>
TABLE 15.6 Gender Composition and Household Economic Pattern for Households with Four Adults: Simulations Based on a Model of Independence
  One Man, Three Women Two Men, Two Women Three Men, One Woman
SOURCE: China Health and Nutrition Survey, 1989, and formulas in chapter appendix.
Urban households      
Mode of production (%)      
Specialized household 55.0 56.0 57.4
Diversified household 45.0 44.0 42.6
Not internally specialized 15.1 16.1 17.2
Internally specialized 29.9 27.9 25.4
Occupation/Industry (%)      
Specialized household 42.0 40.7 39.8
Diversified household 58.0 59.3 60.2
Not internally specialized 11.3 11.5 11.7
Internally specialized 46.7 47.8 48.5
Employment sector (%)      
Specialized household 44.9 41.5 38.2
Diversified household 55.1 58.5 61.8
Not internally specialized 7.6 7.4 7.2
Internally specialized 47.5 51.2 54.6
Rural households      
Mode of production (%)      
Specialized household 43.7 36.1 30.0
Diversified household 56.3 63.9 70.0
Not internally specialized 27.5 33.3 38.6
Internally specialized 28.8 30.5 31.2
Occupation/Industry (%)      
Specialized household 45.3 37.3 30.0
Diversified household 54.7 62.7 69.2
Not internally specialized 29.5 36.0 41.9
Internally specialized 25.2 26.7 27.2
Employment sector (%)      
Specialized household 69.0 64.7 60.6
Diversified household 31.0 35.3 39.4
Not internally specialized 15.0 17.4 19.7
Internally specialized 16.0 17.9 19.7
 
urban areas, differences between the effects of gender composition assuming independence and the effects observed in the data are quite small. Taking gender into account thus clarifies the results found earlier. Although the number of adult men and women has similar implications for household economic pattern in urban and rural areas, these effects differ more from those expected under the assumption of independence for rural areas than for urban areas. This result confirms the initial hypothesis that household size and gender composition would matter more in rural areas.

The particular way in which the observed data depart from what we would expect under the assumption of independence is also quite interesting. In both rural and urban areas the income-earning activities of household members are more likely to be in the same broad occupational category, in the same employment sector, and in the same relation to the household than we would otherwise expect based on household size, gender composition, and gender-and place-specific participation rates. In both rural and urban areas the effect of increasing household size in moving households into the diversified household categories is less than we would expect. In rural areas the effect of increasing the number of men relative to women in moving households into diversified categories is also less than we would expect. There is some interdependence among household members that works against the diversifying effects of increasing numbers of men in rural areas. We speculate about the nature of this interdependence in the concluding section of the chapter.

IMPLICATIONS

Transitional economic times are often characterized by more potential and more risk. If one goal of household economic activity is to maximize income while limiting risk to acceptable levels, one means of accomplishing this goal is to diversify across activities. This chapter has documented patterns of household diversification with data from the 1989 wave of the China Health and Nutrition Survey. Significant fractions of CHNS89 households are diversified: 20 to almost 40 percent, depending on the particular dimension of economic structure that is of interest. Indeed, there is ample reason to expect Chinese households to diversify. The reform period has brought a more diversified economy—industrialization and major shifts in mode of production in the countryside and a growing private sector in towns and cities. Furthermore, the uniformity in household strategies enforced by the collective regime has been lifted. Even if diversification were not an explicit strategy, diversified households nevertheless would benefit from the reduced risks it implies.

Economic transitions have proceeded differently in urban and rural areas, and we expected patterns of diversification to differ as well. We expected to see more diversification according to occupation/industry and mode of production in rural areas, where industrialization is key. We expected to see more diversification according to employment sector in urban areas, where shifts towards a market economy are key. These expectations were borne out to some extent, although the differences were not as sharp as anticipated. In China, at least, household diversification is pervasive throughout the economy.

Household economic involvement in diversified or specialized economic activities may reflect coordinated economic activity, or even explicit household strategy. Or it may not. It is difficult to infer from an outcome the social processes that give rise to it. To address the issue of coordination among household members, we simulated economic patterns assuming that the type of work performed by each household member is independent of every other household member and compared the results to what was found in the data. We expected household size to have a positive effect on household diversification. It did, but the effect was weaker than we expected. We expected that economic pattern would reflect gender composition as well as household size. It did, but again, the effect was weaker than expected. Rather than diversifying, households specialize. Perhaps it was the very instability of economic change in the late 1980s that encouraged a short-run strategy focused on gain (Judd 1994).

But there are other interpretations. For one thing, the operational definition of household might not be right. Perhaps households as defined for the CHNS89 are not the correct unit of analysis. We have considered throughout this chapter that individuals rather than households make the key decisions about their work. It is also possible that economic decision making spans several households. In urban areas, the functions of daily life may link together geographically distinct household units (Davis and Harrell 1993; Unger 1993). In rural areas, households may cooperate economically, eventhough from the perspective of the definition we employ they are distinct units. For example, Croll (1987) describes “aggregate families,” groups made up of kin-connected households or neighboring households cooperating for economic or political reasons. It is possible that we might see more diversification if we considered these larger units, although Judd (1994), for one, has argued that boundaries between households have hardened as a result of economic reforms—even boundaries between households linked by family ties.

We might also see more diversification if we did not focus on the spanning of major economic divides. In rural areas, farming households can diversify the crops they grow or pursue agricultural sideline businesses as well as cultivate field crops. Such households may be pursuing explicit diversification strategies, but they would appear specialized according to the measures used in this chapter. In urban areas, it is possible that households diversify between different units in the state sector, between the state and collective sectors, between large and small collectives, or between different kinds of private sector activities. Again, such diversification strategies would not be detected in our measures. At the same time, there is no reason to expect that our hypotheses about household size and gender composition should hold differently for diversification within than for diversification between major sectors of the economy.

Possibly, similarities in the characteristics of individuals living in the same household (because of marital selection, intergenerational status transmission, and mutual influence), common locational advantages, some inherited expertise or physical capital, or shared connections might make similar types of economic activity for household members likely, even if there is no attempt by household members to coordinate their activities. This could explain why we find so much household specialization—although if this is the case, it is not clear why larger households would be more likely than small households to specialize when we compare observed and predicted patterns of economic activity.

Another interpretation is that, at least to some extent, household economic pattern is a result of deliberate planning. It might be a household head setting goals and strategies for the entire household. It might be household members discussing and agreeing on these goals and strategies (Moen and Wethington 1992). In either case, in this interpretation the observed pattern is a matter of intent, a matter of choice. In rural areas, one can imagine explicit labor allocation decisions being made by the household head, perhaps in consultation with others (M. Wolf 1985). In urban areas, one can imagine household members with desirable jobs using their influence to get similar jobs for other household members. It appears that if these patterns do reflect coordinated activity, whenever possible households prefer not to diversify but rather to specialize in some type of economic activity. Such specialization is often correlated with industrialization and economic development (Boserup 1970).

APPENDIX

Using the example of mode of production, the probability that a household will be diversified but not specialized within is simply the probability that any member works inside as well as outside the household. This is calculated using the binomial formula, which simplifies to:

formula
where q is the probability that no individuals in the household do household and nonhousehold activities and n is the number of household members. The probability that a household is diversified and specialized is calculated by:
formula

where p is the probability that an individual does household-based activity, q is the probability of doing non-household-based work, and r is the probability of doing neither. The household has n members, of whom k do household-based activity, and l do non-household-based activity. Finally, the probability of not being diversified is necessarily the probability of each type of diversification subtracted from one.




NOTES

* This research was supported by National Institute for Child Health and Human Development (NICHD) grant P01-HD28076. We also owe a considerable debt to the China Health and Nutrition Survey. Funds for parts of the survey design, data collection, and computerization were provided by the Chinese Academy of Preventive Medicine and the Carolina Population Center. The National Science Foundation (#37486) and the Ford Foundation funded collaborative training and development work. The survey is a collaboration between the Chinese Academy of Preventive Medicine (Ge Keyou as principal investigator, and Zhai Fengying and Jin Shuigao as coprincipal investigators) and a group from the University of North Carolina at Chapel Hill (Barry M. Popkin as principal investigator, and John S. Akin and Gail E. Henderson as coprincipal investigators). We thank Gail Henderson, Nan Lin, and an anonymous reviewer for helpful comments on an earlier draft, and Grant Izmirlian for his assistance with probability computations.

1. In fieldwork conducted during 1993 in Shandong Province and again in 1995 in Hubei Province in CHNS sampling units, we encountered urban neighborhoods that had been villages incorporated by the growing town. Households in these neighborhoods still farmed plots of land and engaged in other agricultural pursuits.

2. Measures of economic pattern are based on data for 11,584 household members ages 15 and older. We restrict age because we are interested in the organization of household labor and want to include only those most eligible for income-producing activities.

3. A small minority (5%) of households containing only retired persons and others not engaged in any type of income-earning activity are also considered specialized households.

4. The CHNS sampled equal numbers of households in county towns as in cities. In the absence of weighting, the urban sample disproportionately represents the experience of town households. Further, because of the way the urban sample was drawn, even with weights, it is not representative of the urban population. The sampling frame included all towns, but not all cities.

5. Given the definition of household employed by the CHNS, temporary migrants may still be included as household members if their registration is with the household, or if they remain economically tied to the household.

6. We tested for urban-rural differences in the effects of size and gender composition. The results of this test showed no differences for occupation and mode of production, and were marginal for employment sector (P2 − 11.2 with 4 df).

【字体: 】【打印】【关闭

附 件:


| 中心简介 | 网站介绍 | 版权声明 | 服务条款 | 站点导航 | 请您留言 | 网站地图 | RSS聚合资讯