当前位置:首页 >> 其它课程 >>

Labor allocation in transition Evidence from Chinese rural households


China Economic Review 18 (2007) 287 – 308

Labor allocation in transition: Evidence from Chinese rural households☆
Xiaobing WANG a,?, Thomas HERZFELD b , Thomas GLAUBEN a
a

Leibniz Institute of Agricultural Development in Central and Eastern Europe (IAMO), Theodor-Lieser-Str. 2, D-06120 Halle, Germany b Department of Agricultural Economics, University of Kiel, Germany

Abstract Empirical models are developed in this paper to quantitatively analyze households' participation in decisions on hiring labor and supplying labor off the farm, hired labor demand and off-farm labor supply of rural Chinese households. Econometric estimates use micro-level data from Zhejiang province over the period 1995–2002. The main results suggest that the decisions to hire labor and participate off the farm are made jointly and are positively correlated. A household's labor demand decreases with increasing wages for hired labor, whereas the effect of the wages of off-farm workers on a household's labor supply differs significantly depending on the household's kind of labor market participation. The results also indicate that the accumulation of productive assets, the development of livestock production and agricultural prices have increasing effects on labor demand but reducing effects on a household's off-farm labor supply. Land market integration enhances participation significantly but has no significant impact on time allocation. Finally, the results suggest non-separability between hired labor demand and household characteristics, indicating the rural labor market in Zhejiang province is still functioning imperfectly. ? 2007 Elsevier Inc. All rights reserved.
JEL classification: J22; J23; J31; P23; Q12 Keywords: Rural China; Labor demand; Labor supply; Panel data

1. Introduction Rural labor markets enjoy high rankings on political agendas in many countries. China earns special international public interest due to the size of its rural labor force and potentially migrating
☆ The authors acknowledge helpful comments from Stephan Brosig, Martin Petrick, Scott Rozelle, and one anonymous referee as well as financial support of the Deutsche Forschungsgemeinschaft (DFG). ? Corresponding author. Tel.: +49 345 2928 214; fax: +49 345 2928 299. E-mail address: wang@iamo.de (X. Wang).

1043-951X/$ - see front matter ? 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.chieco.2007.02.004

288

X. Wang et al. / China Economic Review 18 (2007) 287–308

population. Almost two-thirds of all employed Chinese people live in rural areas. 318 million inhabitants, currently active in agriculture, are expected to move to other sectors in the future; around 150 million of them are expected to migrate from their current residence. By the year 2020, this figure is expected to rise to 250 million potential migrants (Agra-Europe Germany 42/ 05; p. 14). Before 1978, rural employment in China was predominantly in agriculture and was organized into collectives. Self-employment was almost non-existent, and household business as a demand factor in rural labor markets was negligible. At that time, the Chinese government was interested in securing agricultural production and limiting demand for subsidized food in urban areas, resulting in a strict segmentation between rural and urban labor markets (Dong & Putterman, 2000; Waldman, 2004). When the Household Responsibility System (HRS) was introduced in rural China, the decision on time allocation was transferred from collectives to households, thus offering new channels of employment. In particular, households started to supply labor off the farm and began demanding labor for household business. Whereas employment in agriculture declined from 93% to 64% of total rural employment between 1978 and 2003, the exchange of on-farm labor among households existed marginally in the busy season (Benjamin & Brandt, 2002; SSB). Based on author's survey data, Rozelle (1994) describes, albeit in relatively small numbers, that households also began to hire labor for agriculture and self-employment business. Despite the progress, it is clear from the literature that since the 1990s, rural households still face restrictions in their labor allocation. For example, although the property rights of cultivated land have been greatly improved, land rental markets are still thin in some regions (Kung, 2002). Therefore, households that would prefer to look for work off the farm may find themselves tied to their land. These households may fear completely abandoning agricultural production due to fears of losing their assigned land or receiving land of an inferior quality in future reallocations by village authorities. In addition, education rates in many villages remain low, a factor that has been identified as a primary determinant of accessing the labor market (Yang, 2004). Because of these restrictions, the opportunity cost of farm labor is not equal to non-agricultural wages, as would be expected following economic theory (Zhai, Hertel, & Wang, 2003). Thus, the questions arise: how do households allocate their labor time between farm and off-farm work and how do they develop this allocation over time? What are the driving forces behind a household's time allocation? In particular, what is the impact of land market development on a household's labor allocation? To answer these questions, we use a rich data set from farm household surveys over a long period. The analytical focal point of this paper is the estimation of a series of labor demand and supply functions for the hired work force and off-farm employment. In order to understand how the emergence of the on-farm labor market affects the flexibility of households in their labor market decisions, we divide the sample into three types of households according to their participation in on-and off-farm labor markets. In particular, households hire farm labor, supply off-farm labor or simultaneously act in both markets.1 Thus, we estimate two on-farm labor demand equations: one for households that only hire farm labor and the other for households that simultaneously hire and supply labor. Likewise, we estimate two labor supply functions: one for households that exclusively work off the farm beside their agricultural production and the other
For the demand of hired labor in households which participate in both markets, we cannot explicitly separate hired labor into on-farm hired labor or those who work in the household's non-agricultural business, including industry, construction, transportation and service. Therefore, our incidence of labor demand differs from estimates by Rozelle (1994) as well as Benjamin and Brandt (2002).
1

X. Wang et al. / China Economic Review 18 (2007) 287–308

289

for those households that participate in both labor markets. We expect that households with access to an additional market should be able to respond more flexibly to market signals. Several papers analyze the determinants of Chinese households' participation in off-farm occupations or hiring additional labor using discrete choice approaches (e.g., Chen, Huffman, & Rozelle, 2004; de Brauw, Huang, Rozelle, Zhang, & Zhang, 2002; Glauben, Herzfeld, & Wang, 2005, 2006; Hare, 1994; Rozelle, Guo, Shen, Hughart, & Giles, 1999; Tuan, Somwaru, & Diao, 2000; Zhang, Rozelle, & Huang, 2001). Only Bowlus and Sicular (2003), Meng (2000), Yang (2004), Zhang et al. (2001) and Zhang, Huang, and Rozelle (2002) quantitatively analyze households' time allocation. Whereas Meng (2000) analyzes total working days of household members, irrespective of their type of occupation, Bowlus and Sicular (2003) analyze total agricultural labor use to test separability assumptions. Yang (2004) examines the determinants of households' labor demand for the non-agricultural part of household business. The remaining papers by Zhang et al. (2001, 2002) focus on households' time allocation in agricultural production. This paper adds to the literature the analysis of time allocation of hired and off-farm supplied labor, taking into account the possibility of simultaneous participation in both markets. The applied variable of off-farm participation includes non-agricultural household business as well as wage work. Secondly, this paper uses recent panel data from 1995–2002 to explore the development of the labor market in rural China.2 Thirdly, recent work highlights the importance of land market regulations on households' time allocation (Hertel & Zhai, 2006).3 This paper picks up these hypotheses and especially focuses on the impact of the integration of the land rental market on households' labor market participation and labor time allocation. The rest of our paper is structured as follows: in the next section, we describe our data set and focus on the development of labor allocation within the sampled households. The theoretical framework and econometric specifications are presented in the third section. The following section centers on the results of the econometric analysis and the final section concludes. 2. Data description and the emergence of labor markets 2.1. Data description This study uses fixed-point rural survey data series from Zhejiang province conducted by the Research Center for Rural Economy (RCRE), a large-scale panel survey from 1995–2002.4 The sample is based on a multistage, random-cluster process. Counties, which are below provincelevel administrative units, were stratified by income level and selected based on a weighted

2 Generally, the literature mentioned relies mainly on cross-section household data and does not control for household unobserved characteristics. Two exceptions are the papers by Bowlus and Sicular (2003) and Yang (2004). Whereas the former applies panel estimation techniques to household data from one county in Shandong province covering the period 1990–1993, the latter analyzes data from Sichuan province from 1986 to 1995 broken down into two sub-periods. 3 The aspect of tenure security and investment in agriculture is closely related to this topic. Jacoby, Li, and Rozelle (2002) show that use of organic fertilizer in Northern China is significantly and negatively affected by the risk of expropriation. 4 The rural survey teams of the Research Center for Rural Economy (RCRE) in Beijing conducted the primarily trial survey at the beginning of 1983 in nine provinces. After 1984, the survey was extended to 28 provinces and, finally, now includes 31 provinces and is conducted annually. The survey questionnaire was revised in 1992, 1995 and 2003. For more information see http://www.rcre.org.cn or the discussion in Fang, Wailes, and Cramer (1998) and Benjamin, Brandt, and Giles (2001).

290

X. Wang et al. / China Economic Review 18 (2007) 287–308

sampling scheme. Then, the villages within the counties were randomly chosen according to geographic characteristics (plain, hilly, or mountainous area), location (suburb of city or not), and economic features defined as mainly agriculture, forestry, husbandry, fishery or others. The village survey provides information on resource endowment, employment, and production activities, as well as welfare and other socio-economic indicators. Subsequently, the households are randomly selected from the respective villages. Data based on record dairies are collected quarterly. Households receive a payment between 50 and 200 Yuan (around 6 and 24 US$) from the local government for their efforts. The individual household data contain detailed information on family, farm and other household business characteristics. The survey particularly provides information on households' time allocation of labor to several occupations, such as self-employment off the farm or wage work, as well as households' demand for additional hired labor in household production activities, measured in days. In addition, their related income and expenses are reported. The sample covers around 500 households per annum. The ten villages in the survey are marked with black dots in Fig. A1 in the Appendix. Zhejiang is a rapidly developing province on China's southeast coast and today is one of the richest.5 Its labor market underwent a tremendous transition and seems to be the most suitable for the following empirical analysis. The sectoral composition of the province's economy has changed dramatically as compared to other provinces over the course of economic reforms. Agriculture accounts for only 33% of provincial employment compared to a national average of 44%. Tertiary industry accounts for 33% (SSB, 2004). Regarding rural areas of Zhejiang province, employment in agriculture dropped by 1.6% annually between 1978 and 2003, whereas rural non-agricultural employment grew at the impressive rate of 8.9% annually. However, there is also great heterogeneity within the province, e.g. per capita income between the richest 10% of the counties and poorest 10% of the counties differs more than 90% and off-farm employment rates lie in a range of more than 64% to 32% (SSB, 2004). In short, although our study examines only one province, we believe our choice of province will offer both informative and interesting results and may portend what will happen in the rest of China in the coming years. A sample of 566 households is used in the econometric analyses. On average, 70.2% of households supply family labor off the farm; only 1.4% hire additional labor and 12% participate in both markets.6 Participation rates as well as hired and supplied labor time show no clear trend over time. Fig. A2 in the Appendix displays the participation rate of households in the hired labor market and the average demand of working days of hired laborers at the household level during the survey period. The contributed working days of hired laborers in a household fluctuated substantively between 150 to 520 days in this period, representing the uncertain trend of the household's demand for hired labor. Fig. A3 presents households' participation rates in the offfarm labor market and the working days of off-farm workers at the household level over the period 1995–2002. Off-farm labor supply in a household is on average more than 400 days, which indicates that once a household supplies labor off the farm, on average, the number of offfarm workers in a household is at least two. Table 1 presents a summary of data definitions and some descriptive statistics. The dependent variables in the demand and supply functions are the working days of hired labor within household
The growth rate of nominal GDP per capita is 16.8% annually between 1986 and 2003 (SSB). Unfortunately, we are not able to track different occupations of individual household members. Therefore, we are not able to control for personal characteristics of off-farm workers as in other studies (Huffman & Lange, 1989; Sumner, 1982).
6 5

X. Wang et al. / China Economic Review 18 (2007) 287–308 Table 1 Data definitions and descriptive statistics Labor markets Participation rate (%) Hired labor 13.38 (34.05)

291

Off-farm employment 82.23 (38.23)

Regime h Regime oh Regime o Regime oh Labor demand (days per household) Labor supply (days per household) Labor markets Participation (1 = Yes, 0 = No) Household characteristics Share of labor graduated from elementary school (%) ELEMENTS Share of labor graduated from secondary school (%) Share of labor graduated from high school (%) Share of labor with special abilities (%) No. of male labor (person) No. of female labor (person) No. of dependents (person) Net transfer per capita (1000 Yuan/person) SECONDS HIGHS SKILLS MLABOR FLABOR DEPEND TRANSFER Symbol Hired labor Ih = 1 39.10 (32.01) 40.44 (30.77) 11.78 (21.86) 15.10 (24.00) 1.45 (0.65) 1.18 (0.60) 1.20 (0.91) ? 0.30 (1.96) 0.16 (0.37) 0.26 (0.44) Ih = 0 43.06 (33.82) 33.85 (31.68) 7.33 (18.76) 7.98 (18.25) 1.35 (0.63) 1.19 (0.68) 1.16 (0.98) ? 0.23 (2.55) 0.08 (0.27) 0.13 (0.34) 199.13 (468.54) 305.43 (517.32) 452.46 (294.25) 495.63 (280.93)

Off-farm employment Jo = 1 42.49 (33.27) 35.51 (31.73) 8.25 (19.62) 9.61 (19.87) 1.39 (0.64) 1.21 (0.68) 1.19 (0.97) ? 0.27 (2.41) 0.08 (0.27) 0.14 (0.34) Jo = 0 42.71 (35.15) 31.15 (30.97) 6.42 (17.43) 5.81 (15.88) 1.24 (0.60) 1.08 (0.62) 1.07 (0.94) ? 0.08 (2.75) 0.14 (0.34) 0.21 (0.41)

Any household member village cadre (1 = yes, 0 = no) CADRE Communist party membership (1 = yes, 0 = no) PMEMBER

Farm characteristics Household's production assets per capita at 1995 constant prices (1000 Yuan/ person) Land per capita (mu/person) a

7.28 (14.85) LANDPC 6.75 (10.88) Animal husbandry (log of output in quantity) LIVESTOCK ? 1.27 (4.62) Share of vegetable sown area (%) VEGETABLE 0.14 (0.24) Agricultural output value divided by inputs value (%) TOT 24.02 (289.53) Value of output at 1995 constant prices (Yuan) OUTPUT 7919.38 (17221.4) Land (mu) a LAND 22.94 (31.21) Family labor (days) FAMILYL 121.28 (231.60)

ASSET

3.06 (13.56) 2.35 (5.05) ? 1.94 (4.39) 0.10 (0.19) 17.06 (136.28) 8700.75 (2585.7) 7.88 (14.37) 161.90 (195.22)

3.32 (11.13) 2.98 (6.62) ? 1.85 (4.41) 0.10 (0.19) 10.96 (69.67) 6461.23 (14643.7) 10.14 (19.04) 140.24 (168.90)

0.51 (2.23) 2.74 (4.78) ? 1.85 (4.52) 0.11 (0.21) 50.55 (360.56) 18478.28 (37223.7) 8.77 (14.45) 231.54 (297.38)

(continued on next page)

292 Table 1 (continued ) Labor markets

X. Wang et al. / China Economic Review 18 (2007) 287–308

Symbol

Hired labor Ih = 1 Ih = 0 0.00 (0.00) 11484.22 (48833.4) 304.11 (353.4) 5819.95 (20512.4)

Off-farm employment Jo = 1 44.63 (225.08) 12503.54 (41566.5) 312.95 (598.5) 3601.46 (13606.8) Jo = 0 15.27 (138.91) 18155.97 (75195.9) 323.69 (349.6) 14927.43 (35712.5)

Participation (1 = Yes, 0 = No) Farm characteristics Hired labor (days) Capital at 1995 constant prices (Yuan) Expenses on fertilizer (Yuan) Sum of intermediate inputs at 1995 constant prices (Yuan) HIREL CAPITAL FERTILIZER INTERMED

294.61 (513.08) 26610.49 (50199.5) 384.50 (1246.0) 4278.01 (15609.0)

Village characteristics Share of households that lease land in the village (%) LRENT Unemployment rate (%) Annual net income per capita in the village at 1995 constant prices (1000 Yuan/person) Population density (inhabitants/mu) UNEMP ANIPC POPDEN

12.44 (12.80) 14.19 (12.32) 5.47 (2.36) 0.92 (1.19)

8.58 (10.94) 12.71 (12.48) 5.06 (2.29) 1.22 (1.17)

9.54 (11.83) 13.10 (12.60) 5.24 (2.36) 1.20 (1.22)

7.05 (7.98) 12.01 (11.84) 4.54 (1.92) 1.05 (0.96)

Time trend Time Observations Source: RCRE survey for Zhejiang 1995–2002. Note: standard deviation in parentheses. a 1 mu = 0.0667 ha.

TIME

4.71 (2.26) 442

4.55 (2.22) 2862

4.61 (2.23) 2717

4.41 (2.22) 587

business and working days of all family members in any off-farm occupation, respectively. Three groups of factors explain the observed labor demand and supply: household characteristics, farm characteristics, and local village features. Household characteristics include educational attainment and skill of laborers (ELEMENTS, SECONDS, HIGHS, SKILLS), gender composition of the family work force (MLABOR, FLABOR) and number of dependents (DEPEND) and the net transfers from relatives or government (TRANSFER).7 The proxies to represent social networks in a household include whether any family member is a village or township cadre (CADRE) or member of the communist party (PMEMBER).8 The econometric analysis controls for following farm characteristics as follows: size of production asset holdings (ASSET), farm size per household member (LANDPC), scale of labor-intensive production activities (LIVESTOCK, VEGETABLE) and the relative development of agricultural prices (TOT). Three variables at the village level are used to proxy for the variation in the external labor market conditions, including unemployment rate (UNIMP), average income per capita (ANIPC), and population density (POPDENS). Finally, several authors highlight the importance of land markets on the households' time allocation (Benjamin, 1992; Hertel & Zhai, 2006; Kung, 2002; Yang, 1997b). Because the outcome of land
7 Dependents are defined as children less than 16 years, full-time students, and the elders above 65 years who are not classified as labor force in the survey. 8 The importance of social networks for Chinese labor markets is discussed by Knight and Yueh (2002).

X. Wang et al. / China Economic Review 18 (2007) 287–308

293

markets depends crucially on local decision makers, the share of households that lease land (LRENT) at the village level is used as a proxy of a land market's development. The following sections theoretically and econometrically explore the determinants of these above-mentioned developments. 3. Theoretical background and econometric specification 3.1. A theoretical framework To illustrate the farm household's decision problem, we constructed a household model assuming labor markets are imperfect. 9 It is well known that under perfect labor market conditions, the allocation of family labor between on- and off-farm work and the demand for a hired work force in agricultural production could be determined separately. However, it is jointly determined in the case of imperfect labor markets (Sadoulet, de Janvry, & Benjamin, 1998). The model particularly allows households to simultaneously hire laborers and supply labor off the farm. The farm household is assumed to maximize utility (U ) derived from the consumption of goods (Cm) and leisure (Cl) subject to a technology constraint (Eq. (2)), a time constraint (Eq. (3)), and a budget constraint (Eq. (4)). Therefore, farm households solve the following maximization problem: max U ?Cm ; Cl ; zu ?
c

?1?

subject to G?Y ; Xv ; Lf ; Lh ; K; zg ? ? 0 Tl ?L ? Lh ?Lo ?Cl z0 Pm Cm V Py Y ?Pv Xv ?g?Lh ; zh ? ? f ?Lo ; zo ? ? E ?2? ?3? ?4?

Here, U is the farm household's utility function, which is assumed to be well-behaved. C is a vector of consumption goods consisting of market commodities (Cm) and leisure (Cl), and zu represents exogenous utility shifters, e.g. household characteristics. G represents a well-behaved production technology (Eq. (2)). The farm household is assumed to produce agricultural products (Y ) using variable inputs (Xv), family labor (Lf) and hired labor (Lh), and the quasi-fixed factors (K) ? capital and land ? while zg is an exogenous production shifter. The farm household faces a time constraint (Eq. (3)), where Tl is the total time available and L = Lf + Lh is the total on-farm labor time subdivided into family labor (Lf ) and hired labor (Lh). Furthermore, Lo indicates offfarm family labor. The farm household's budget constraint (Eq. (4)) states that the household's expenditures must not exceed its monetary income. Here, Pi (i = m, y, v) represents the exogenous consumer and producer prices. Conditional on the labor market participation regimes noted below, the farm households might generate revenue from farming PyY ? Pv Xv ? g(Lh;zh), where g(Lh;zh) denote the cost of hiring onfarm labor, and labor income from off-farm work f (Lo; zo), as well as exogenous transfers (E ). To consider labor market imperfections, revenues from off-farm employment f (Lo; zo) and hired
This approach dates back to Becker's (1965) theory of household production, further developed by Singh, Squire and Strauss (1986) and applied extensively in several papers (Findeis & Lass, 1994; Glauben, 2000; Glauben et al., 2005, 2006; Key, Saduolet, & de Janvry 2000; Sadoulet et al., 1998). Therefore, we present only a very short description, which relies mainly on the latter sources.
9

294

X. Wang et al. / China Economic Review 18 (2007) 287–308

labor cost g(Lh;zh) are conceptualized as functions of supplied (Lo) and hired labor (Lh) time (Glauben, 2000; Glauben et al., 2005, 2006) and the exogenous shifters zo and zh, respectively, that may include differences in external wage levels, skills and human capital, or transaction cost. Under perfectly competitive labor markets, the functions are both linear, with f (.) = wl Lo or g(.) = wlLh. Hence, marginal off-farm income or marginal costs of hired labor are equal to the exogenous wage rate (wl), in which case the farm household model is said to be separable. In contrast, when labor markets are imperfectly competitive, both supplied and hired labor 2 functions become nonlinear with the following properties:?f (.) / ?Lo N 0; ?2f (.) / ?Lo ≠ 0 and ?g(.) / 2 2 ?Lh N 0; ? g(.) / ?Lh ≠ 0, respectively. Now, off-farm income is a nonlinear function of supplied labor time. Analogously, the cost of hiring labor is a nonlinear function of hired labor time. In this case, the price of labor and leisure (wl) is endogenously determined, and thus the farm household model is non-separable. The production and consumption decisions are simultaneously determined by the stationary solution of the equation system (1) (2) (3) (4). This framework is applicable to several kinds of labor market imperfections.10 In particular, it accounts for those that lead to an upward-sloping or backward-sloping price, effectively received for each further unit of off-farm employment and paid for each further unit of hired labor time. Hence, the per-unit cost of accessing labor markets can be increasing or decreasing. Increasing per-unit cost associated with working off the farm may be caused by increasing heterogeneity between on- and off-farm family labor. With increasing migration, household members are first transferred to the ‘best jobs’ followed by the ‘next best jobs’ and so on (Low, 1986). However, offfarm networks within a household or village prove to reduce the information cost for potential offfarm workers by providing the information on off-farm posts (Rozelle, Brandt, Guo, & Huang, 2002; Zhao, 2002). With an effective and expanding network, the marginal cost of family members' off-farm employment will decrease. Thus, the effects of internal wage on off-farm labor supply could present in opposite directions. Increasing per-unit cost of hired labor may result from increasing search activities. These increases may stem from the growing difficulty in finding the ‘right’ staff for the different and often farm-specific areas of production. Similarly, the existence of land-specific experience may lead to a decreasing substitutability between family and hired labor. Thus, hired labor could become less productive and the cost for a standardized hired labor unit increase. However, familiarity between the hired laborers and hosted households, could decrease the supervision and monitoring cost of hired laborers with the increasing contributed time of hired labor. Thus, the marginal cost of hired labor could also possibly decrease given the more efficient work of the hired laborer.11 The first order conditions of the maximization problem (1) (2) (3) (4), that is, Ui ?:? ? uGi ?:? ? k fl ?:? ? gl ?:? ? wl , determine a complete set of demand and supply functions (Huffman, 1980; Huffman & Lange, 1989; Lass & Gempesaw, 1992). Here, λ, u, μ N 0 are Lagrangian multipliers
10 In general, the literature points to several reasons why labor markets may be imperfect, leading to non-separation of consumption, production and labor-supply decisions. For example, binding hour constraints in off-farm employment may prevent a complete adjustment in agricultural labor markets (Benjamin, 1992). Family and hired labor may be imperfect substitutes in agricultural production (Deolalikar & Vijverberg, 1987; Jacoby, 1993). Also, farmers may have preferences towards working on or off the farm (Lopez, 1994). In addition, costs associated with labor market transactions can explain why households have different relationships to the labor markets (Sadoulet et al., 1998). 11 Note that the approach could additionally incorporate fixed costs of transactions that are invariant to the traded quantity, but also could affect the farm household's decision to participate in markets (Sadoulet et al. (1998) for the labor markets; Goetz (1992) as well as Key et al. (2000) for food markets; Skoufias (1994) and Carter and Yao (2002) for the land market). Fixed transaction costs may include bargaining and negotiation efforts and transportation costs, often taking place once per transaction, and are invariant to the level of transaction.

X. Wang et al. / China Economic Review 18 (2007) 287–308

295

associated with the budget, the technology, and the respective time constraints, and Ui, Gi, fl and gl represent the first derivatives of the corresponding utility, production, and labor market functions. Further, wl = μ / λ denotes the internal wage rate that results around the optimum as an implicit function of all exogenous variables mentioned above: wl ? k?Pc ; Pv ; K; E; Tl ; zh ; z0 ?. In the theory, all marginal contributions of labor and leisure should be equal to the internal wage rate. However, from the empirical point of view, there might be differences between the internal wage rate for hired labor (wlh = gl (.)) and the wage rate for supplied labor time off-farm (wlo = fl (.)), in particular because of unobserved transaction costs. Four possible regimes of households' labor demand and supply may arise: (1) exclusive demand of hired on-farm labor (Lh N 0, Lo = 0) further denoted with regime h; (2) exclusive supply of off-farm labor (Lo N 0, Lh = 0) further denoted with o; (3) simultaneous demand of hired labor and supply of off-farm labor (Lh N 0, Lo N 0) further denoted with oh and (4) neither demand of hired labor nor supply of off-farm labor (Lh = 0, Lo = 0). As mentioned before, the appropriate reduced form of hired-labor demand and off-farm labor supply functions are simultaneously determined by solving Kuhn–Tucker conditions. The demand for hired labor is not conditional on the supply of off-farm labor when none of the family members takes off-farm employment as in the regime h: Lh ? Lh ?wh ; zh ? l ?5?

When some family members participate in off-farm employment, their wage or income will influence the demand of hired labor. As mentioned before, although according to theory there should be no difference between the internal wage for hired labor (wlh) and wage rate for labor supplied offfarm (wlo) optimally, empirically there are differences between these two wage rates (Zhai et al., 2003). Thus, because the demand for hired labor is not independent from the wage of off-farm labor when the household simultaneously hires labor and supplies family labor off the farm, we also included the wage rate for off-farm labor in the demand function for hired labor in regime oh: Lh ? Lh ?wh ; wo ; zh ; zo ? l l ?6?

The off-farm labor supply function is defined similarly. In the absence of hired labor, households' off-farm labor supply does not depend on the cost of hired labor for households in regime o: Lo ? Lo ?wo ; zo ? l ?7?

In the regime (oh) off-farm labor supply is influenced by the cost of hired labor and off-farm wages: Lo ? Lo ?wh ; wo ; zh ; zo ? l l ?8?

Assuming a binding non-negativity constraint of households' time allocation implies nonseparability between farm production, household consumption, and off-farm labor supply; in other words, the same variables determine the optimal amount of time spent on household business and off-farm work (Singh et al., 1986). 3.2. The econometric specification The empirical model of households' labor time allocation consists of the demand function for hired labor (Lhi), the supply function for family off-farm labor (Loi), and two participation

296

X. Wang et al. / China Economic Review 18 (2007) 287–308

decision rules (Findeis & Lass,1994; Heckman, 1974, 1979; Huffman & Lange, 1989; Lass & Gempesaw, 1992; Tokle & Huffman, 1991). In general, labor participation rules are determined by comparing the utilities of households participating in the different labor supply and demand ? regimes.12 Unobserved indicators (Ih ) are assumed to represent utility differences between households that hire labor and households that do not hire labor. Positive working days of hired laborers will be observed if potential utility earned by a household from hiring laborers is greater than the utility of a household that does not demand extra laborers. The participation decision rule to hire labor for the ith household is as follows:  ? N0 if Ihi ? zhi ah ? ehi N0 Lhi i ? 1; N ; N ?9? ? ? 0 if Ihi ? zhi ah ? ehi V0 Regarding the household's decision to participate in off-farm labor markets, the household compares the utility earned from its supply of labor off the farm to the utility in the case of no participation in off-farm occupations. If the utility earned from off-farm employment exceeds the utility without off-farm workers, the household will supply off-farm labor (Loi N 0). In this case, the unobserved utility difference between households with and without off-farm workers is ? represented by Jo . In other words, the participation decision rule to supply off-farm labor for the ith household is:  ? N0 if Jo i ? zoi ao ? eoi N0 ?10? Loi ? ? z a ? e V0 i ? 1; N ; N ? 0 if Jo i oi o oi Observed differences can be represented by dichotomous dependent variable models. Assuming that Ih and Jo each equals one if the household participates in the respective markets, this decision could be analyzed econometrically using binary choice models for estimation in the first step. That means Ih = 1 if Lh N 0 and Ih = 0 if Lh = 0. Similarly, Jo = 1 if Lo N 0 and Jo = 0 if Lo = 0. Given that households' decisions on hiring labor or supplying labor off the farm are assumed to be joint within a household-optimizing framework, the probability of households' hiring labor is affected by the probability of households' supplying labor off-farm, and vice versa. Also, these decisions are affected by random or unmeasured shocks to labor demand and supply functions, and these shocks likely occur for both hired labor and off-farm workers. Thus, a bivariate probit model is appropriate for the first step estimation (Lass & Gempesaw, 1992; Tokle & Huffman, 1991). If the correlation coefficient of the error terms in the bivariate probit equation is significantly different from zero, this indicates that the household decisions to hire labor and supply labor offfarm are not statistically independent. This could indicate that the rural labor market is still imperfect. Because labor demand and supply functions are conditional on participation in the respective labor market, there is the possibility of sample selectivity bias in the error terms of the labor supply and demand equations (Heckman, 1974, 1979). Thus, the inverse Mill's ratio is calculated from the bivariate probit model as the sample selection terms and consistent tests of sample selectivity bias will be conducted (Heckman, 1974, 1979; Sumner, 1982).
It should be noted that the assumed labor market conditions can create non-convexities of the budget set. Thus, the simple story of reservation wage in the neighborhood of zero “marketed” labor hours does not hold. Both the net wages and the time interval, over which the story of reservation wage holds, must be considered in assessing labor market participation. In other words, the purely local considerations of reservation wage models are no longer sufficient to determine whether a household chooses to participate in labor markets when non-convexities are present (Hausman, 1980).
12

X. Wang et al. / China Economic Review 18 (2007) 287–308

297

It is of direct interest to understand if the households' labor allocation changes when households are participating in rural labor markets that allow them to both hire labor and supply labor off the farm. In other words, are households more responsive to changes in internal wages and other factors when they have more labor decision-making options? Thus, the second step is to estimate the two hired labor demand functions for households that purely hire labor in Eq. (5) and for households in both markets in Eq. (6), respectively; and the two off-farm labor supply functions including Eq. (7) for households exclusively supplying labor off farm and Eq. (8) for households in both markets. Much of the literature proves that the internal wage, rather than the observed market wage, determines the household's labor allocation (Benjamin, 1992; Huffman & Lange, 1989; Jacoby, 1993; Skoufias, 1994; Sumner, 1982). Similar to the procedure in Sumner (1982), as well as Huffman and Lange (1989), wage functions are modeled for hired labor and off-farm workers. The resulting values enter the demand and supply function as predicted endogenous variables.13 Under the assumption that hired labor and family on-farm labor are substitutes—albeit imperfect substitutes—the anticipated wage of hired labor could be expressed by the marginal return to hired labors' farm work according to the first order conditions of hired labor derived from the agricultural production function: wh ? l ?G ?2 G and b0 ?Lh ?L2 h ?11?

This means the wage function of hired labor is the change in net farm returns of household production resulting from a marginal increase in hired labor input. Regarding the aggregated return to households' off-farm activities, the anticipated wage of the off-farm worker is assumed to be influenced by the accumulation of a household's human capital and local labor market characteristics (zo) (Tokle & Huffman, 1991): wo ? wo ?zo ? l l ?12?

Labor demand and supply are measured in working days and wages in Yuan per day. The bivariate probit equation and the labor demand of households which exclusively hire on-farm labor are estimated as a pooled cross-section. All the other specifications are estimated as panel models.14 This procedure controls for unobserved household characteristics such as management ability or inherent preferences for farming. The exogenous variables in the empirical estimations include characteristics of the household, farming activity, and village. 4. Empirical results A bivariate probit model is estimated for households' participation in off-farm employment and the hiring of labor. The results are presented in Table 2. The estimated labor participation model is statistically significant at a level of 1% or better as indicated by a Wald χ2 test statistic.
More precisely, because off-farm work in our sample includes wage employment as well as self-employment, it would be better to use the term ‘average household earnings from any off-farm occupation’. To simplify matters, the term ‘off-farm wage’ encompasses all sorts of off-farm income from working activities. 14 It is probable that the coefficients in the bivariate probit model might be overestimated since it is not controlled for unobserved household characteristics. Although there are attempts to estimate univariate probit models using panel data (Ooms & Hall, 2005; Wooldridge, 2002), to our knowledge there is no solution at the moment for bivariate probit models. Because results show that the estimated cross-correlation equation is statistically significant, we decided to use the bivariate probit model.
13

298

X. Wang et al. / China Economic Review 18 (2007) 287–308

Table 2 Results of households' labor participation from bivariate probit model Dependent variables Hired labor (Ih = 1 or 0) Coefficients Household characteristics ELEMENTS SECONDS HIGHS SKILLS MLABOR FLABOR DEPEND TRANSFER CADRE PMEMBER Farm characteristics ASSET LANDPC LIVESTOCK VEGETABLE TOT Village characteristics LRENT UNEMP ANIPC POPDEN Time trend TIME Constant Rho Log likelihood Wald χ2 (df) Observations 0.2270 0.4848??? 0.5085?? 0.3601?? 0.1788??? ? 0.0196 0.1271??? ? 0.0164 0.2494?? 0.3215??? 0.0640??? 0.0346??? 0.0019 0.1739 0.0001 0.4910? ? 0.5177? 0.1054??? ? 0.1754??? ? 0.0333?? ? 2.2998??? 0.1439??? ? 2574.9863 473.16 (40) 3304 Z-values 1.52 3.20 2.67 2.48 3.62 0.41 3.80 1.38 2.31 3.62 Off-farm employment (Jo = 1 or 0) Coefficients 0.3347??? 0.5001??? 0.6387??? 0.5165??? 0.1897??? 0.1339??? 0.1367??? ? 0.0119 ? 0.2766??? ? 0.2598??? ? 0.0694??? 0.0035 0.0002 0.0960 ? 0.0010??? 0.5443? 0.0895 0.0916??? ? 0.0589 Z-values 3.12 4.28 3.76 3.25 3.99 3.03 4.45 1.10 2.71 3.10

3.86 7.97 0.26 1.17 0.53

3.72 0.70 0.03 0.64 3.69

1.73 1.93 5.74 4.33

1.80 0.37 4.88 1.55

2.19 12.88 3.11

0.0042 ? 0.3978???

0.32 2.84

Note: ???, ??, and ? indicate statistical significance at 1%, 5% and 10% levels, respectively. Rho indicates the crossequation correlation and it is distributed following a χ2-distribution.

The estimated cross-correlation equation between the decision to hire labor and to work off the farm is positive and significantly different from zero. This estimate indicates that the decision to hire labor increases the likelihood that the household participates in off-farm employment, and vice versa. More specifically, omitted factors which explain the probability of hiring labor or participating in off-farm occupation are positively correlated (Greene, 2000). Estimated parameters of the explanatory variables are in line with previous research on labor market participation using Chinese data (Chen et al., 2004; de Brauw et al., 2002; Glauben et al., 2005, 2006; Hare, 1994; Rozelle et al., 1999; Tuan et al., 2000; Zhang et al., 2001). As expected, education (SECONDS, HIGHS) has a higher impact on the probability of working off the farm than hiring labor. Social networks, as captured by the variables CADRE and PMEMBER, raise the probability of hiring labor but reduce the probability of working off the farm. The same applies to the variable ASSET. Increasing relative agricultural prices (TOT) reduces the probability of participating in off-farm work.

X. Wang et al. / China Economic Review 18 (2007) 287–308 Table 3 Estimation of production function and off-farm income function Production function Dependent variable Log(OUTPUT) Fixed-effects estimation Explanatory variables Log(LAND) Log(FAMILYL) Log(HIREDL) Log(CAPITAL) Log(FERTILIZER) Log(INTERMED) ELEMENTS SECONDS HIGHS SKILLS CADRE PMEMBER Constant F-test(12, 1584) Within R2 Between R2 Overall R2 Observations Coefficient 0.1609??? 0.3146??? 0.0433??? ? 0.0063 0.0262 0.4397??? ? 0.2610?? ? 0.1686 ? 0.4426??? ? 0.0267 0.1993??? ? 0.3152??? 3.6649??? 234.75 0.6401 0.8469 0.7893 1975 t-values 5.16 11.73 2.67 0.39 1.11 29.01 2.34 1.35 2.66 0.21 2.58 2.83 15.34 Explanatory variables ELEMENTS SECONDS HIGHS SKILLS MLABOR FLABOR CADRE PMEMBER UNEMP ANIPC POPDEN TIME Constant F-test(12, 2242) Within R2 Between R2 Overall R2 Observations Off-farm wage function Dependent variable

299

Log(WAGE-off-farm worker) Fixed-effects estimation Coefficient 0.1808 0.2980? 0.2076 ?0.0377 0.0740 ?0.1475??? 0.0521 0.3856??? ?0.2770 0.0658??? ?0.8529??? 0.0361??? 3.9990??? 10.87 0.0550 0.0235 0.0112 2717 t-values 1.13 1.67 0.92 0.25 1.45 3.83 0.45 2.57 1.01 2.88 3.74 3.59 13.13

Note: ???, ??, and ? indicate statistical significance at 1%, 5% and 10% levels, respectively.

Regarding village characteristics, the probability of hiring labor as well as participating in offfarm employment is higher in wealthier villages (ANIPC). The probability of hiring labor is significantly lower in villages with a higher unemployment rate (UNEMP) and in more densely populated villages (POPDEN). The impact of the land rental market's development at the village level is noted as the substantial hypothesis in the econometric analysis. It is hypothesized that a higher integration on the land market relaxes restrictions on labor time allocation, and therefore enhances the integration of labor markets. The estimated coefficients for the share of households which lease land (LRENT) in local villages are statistically significant and positive for both decisions, while the magnitude of the coefficient is slightly higher in the case of off-farm participation. This indicates that land market integration significantly enhances households' participation in the labor market and has a more important influence on off-farm employment than in the hiring labor decision. Table 3 presents the estimated results of household-level agricultural production and average wages of off-farm workers. First, the wage of a hired laborer is derived from the estimated elasticity of hired labor from the agricultural production function. A Cobb–Douglas production function with fixed-effects specification yields an elasticity of family labor (FAMILYL) of 0.31 and hired labor (HIREDL) of 0.04.15 Both estimates are statistically significant at the level of 1%.
We use a Hausman test to compare between random-effects and fixed-effects specifications. The resulting Chi-square a Hausman test to compare between random effect and fixed-effect specifications. The statistic of 32.69 with 12 degrees of freedom strongly rejects the random-effects model at the 1% significance level, of freedom strongly rejects the random effect suggesting that the unobserved factors are correlated with the explanatory variables in the estimation of households' agricultural production.
15

300

X. Wang et al. / China Economic Review 18 (2007) 287–308

These results indicate that, on average, farms in Zhejiang do not experience hidden unemployment.16 The estimated elasticity is quite close to the results of Bruemmer, Glauben, and Lu (2006) as well as Liu and Wang (2005). Based on the elasticity, the marginal product of hired labor is calculated by multiplying it with the average income of a hired laborer from agricultural production, which could be obtained by dividing the total value of agricultural production by the total working days of hired laborers. Thus, the marginal product of hired labor is distributed around a mean of 26.29 Yuan per day. The temporal development shows no clear trend (see Table A1 in the Appendix). The off-farm wage function is directly estimated with these data. The models are estimated in loglinear form, and a consistent test of sample selection bias is applied (Heckman, 1974, 1979; Sumner, 1982). Selection bias is rejected for these data. Thus, the average income of off-farm workers is estimated at the household level by dividing the total income from non-agricultural activities by the total off-farm working days with the fixed-effects model.17 The resulting variable is regressed on the educational attainment and skill characteristics of the household, gender distribution of the family's work force, controls for social networks, and local labor market features. The average income per capita at the village level as a demand side variable acts as an exogenous instrument. Table 3 presents the results, which support the general hypothesis that the educational attainment is positively related to daily earnings, although only the coefficient of secondary education is statistically significant (Huffman & Lange, 1989; Sumner, 1982; Yang, 1997a). Increasing the share of family members who finished secondary school (SECONDS) by 10 percentage points will, ceteris paribus, raise households' average off-farm income by 3.3%. The positive estimate of MLABOR and the strongly negative estimate of FLABOR indicate that earning differences by gender still exist. Interestingly, party membership (PMEMBER) has a statistically significant positive impact on off-farm wages. The coefficients of village characteristics imply that the wage of the off-farm worker is significantly higher in relatively wealthier villages but lower in more densely populated areas. The statistically significant coefficient of TIME in Table 3 indicates an increase of off-farm wages between 1995 and 2002. The expected average non-agricultural wage of a household obtained from the wage function is 52.37 Yuan per day, which is twice the marginal product of on-farm labor (see Table A1). This is in line with findings by Cook (1999) as well as Fleisher and Yang (2003). As mentioned earlier, sample selection bias is rejected for these data. Thus, the traditional econometric techniques are applied to the following estimations of the hired labor demand and off-farm labor supply functions. In the econometric analysis, working days of hired laborers and family members' off-farm labor supply are regressed on the predicted wages for hired labor and predicted wages from off-farm work, and household, farm, and village characteristics. Table 4 presents the results of these estimations. 4.1. Demand of hired labor Hired labor demand functions are estimated separately for households that only hire on-farm labor in column 1 of Table 4 (regime h) and for households that hire and supply labor
16 Estimates of agricultural surplus labor at the national level vary from 30 % to 40% of the agricultural labor force. One main advantage of this micro-level database is the avoidance of possible mismeasurement of rural employment in official statistics (Bhattacharyya & Parker, 1999). 17 Specification tests are also performed for random-effects and fixed-effects models. The Hausman test statistic strongly rejects the random-effects model in favor of the alternative fixed-effects specification.

X. Wang et al. / China Economic Review 18 (2007) 287–308 Table 4 Estimated results of labor demand and supply functions Labor demand functions Dependent variable Households' labor participation Explanatory variables Wage Log(WAGE-hired labor) Log(WAGE-off-farm worker) Log(hiring-in days) Regime h OLS Regime oh Fixed-effects estimation ? 0.3657??? (11.24) 1.3369 (0.83) Labor supply functions Log(off-farm working days) Regime o Fixed-effects estimation Regime oh

301

Random-effects estimation ?0.0692??? (5.22) 1.3355??? (3.74)

? 0.2900?? (2.68)

?0.2134 (0.62)

Household characteristics ELEMENTS SECONDS HIGHS SKILLS MLABOR FLABOR DEPEND TRANSFER CADRE PMEMBER

0.3832 (0.40) 2.2846?? (2.24) 0.3236 (0.22) 2.1706 (1.33) 0.1490 (0.22) ? 0.3026 (0.57) ? 0.3477 (1.14) ? 0.1121 (0.81) 0.1296 (0.14) 1.5568??? (1.78) 1.4115? (1.95) ? 0.0106 (0.29) ? 0.0068 (0.11) 2.3473? (1.79) ? 0.0004 (1.09)

? 0.7493 (0.78) ? 0.5787 (0.54) ? 0.0451 (0.04) ? 0.1125 (0.15) 0.1091 (0.35) 0.1898 (0.60) ? 0.0679 (0.40) 0.0054 (0.14) ? 0.0318 (0.07) ? 0.7013 (0.89)

?0.1888 (1.20) 0.0672 (0.35) 0.0131 (0.06) 0.1048 (0.74) 0.2939??? (5.51) 0.2083??? (3.38) 0.0405 (1.39) ?0.0153?? (2.43) 0.0270 (0.24) 0.0742 (0.36)

?0.2603 (1.14) ?0.4356? (1.81) 0.0112 (0.04) 0.3631? (1.91) 0.1390? (1.86) 0.5383??? (6.64) 0.0622 (1.34) ?0.0055 (0.33) ?0.2727? (1.93) ?0.4734??? (2.82)

Farm characteristics ASSET LANDPC LIVESTOCK VEGETABLE TOT

0.0723 (0.91) ? 0.0212 (0.50) 0.0684?? (2.47) 0.0190 (0.06) 0.0055? (1.93)

?0.0175 (0.66) ?0.0117?? (1.96) ?0.0017 (0.30) 0.0747 (0.70) ?0.0009?? (2.18)

?0.0233 (0.85) ?0.0088? (1.87) ?0.0223?? (2.29) 0.0588 (0.39) ?0.0016 (1.31)

Village characteristics LRENT UNEMP

? 2.6531 (0.79) 0.7503 (0.31)

? 0.2901 (0.51) 0.7657 (0.62)

0.1764 (1.17) ?0.7173?? (2.29)

?0.2257 (0.83) ?0.1059 (0.33) (continued on next page)

302 Table 4 (continued)

X. Wang et al. / China Economic Review 18 (2007) 287–308

Labor demand functions Dependent variable Households' labor participation Explanatory variables Village characteristics POPDEN Log(hiring-in days) Regime h OLS Regime oh Fixed-effects estimation 2.6143 (1.47)

Labor supply functions Log(off-farm working days) Regime o Fixed-effects estimation 0.0003 (0.00) Regime oh Random-effects estimation 1.0418??? (3.77) ?0.0655??? (2.59) ?0.3803 (0.25) 441.33 (20)

0.4301 (1.32)

Time trend TIME Constant F-test (df ) Wald χ2 (df ) Adjusted-R2 Within-R2 Between-R2 Overall-R2 Observations

0.1419 (1.15) 1.2258 (1.00) 4.74 (20, 24) 0.6293

? 0.0218 (0.25) ? 3.0245 (0.46) 11.63 (21, 226)

0.0280 (1.39) 5.9096??? (4.15) 7.02 (20, 1847)

45

0.5194 0.2277 0.2839 397

0.0706 0.2565 0.2009 2320

0.0592 0.4070 0.2608 397

Note: t-values in parentheses, ???, ??, and ? indicate statistical significance at 1%, 5% and 10% levels, respectively.

simultaneously in column 2 (regime oh). From the OLS estimation, the adjusted R2 -value amounts to 0.63 for households which only hire labor (regime h). This indicates that the included explanatory variables explain more than half the variation of hired farm labor demand. Fitting this hired labor demand function with random- or fixed-effects methods, the Hausman specification test strongly rejects the random-effects model, suggesting that the fixed-effects estimation is appropriate for further interpretation. Comparing the within-R 2 and the between-R 2 for households participating in both markets (regime oh) points to a better explanation of the variation in hired labor use within households over time than between households. An increase in hired labor wages (WAGE-hired labor) leads to a decreasing demand for hired labor days and a possible substitution between hired and a household's own labor. An increase of 1% in the wages for hired work force is followed by a decrease in the hired labor demand by 0.29 and 0.37% for households hiring labor and simultaneously participating in both markets, respectively. As expected, households that participate in both markets (regime oh) are more flexible and show higher elasticity as suggested by the absolute magnitude of the estimated coefficients. This confirms the theory that agents will be more flexible if the number of choices in the labor market increases. However, the wage of off-farm workers (WAGE-off-farm worker) appears to have no statistically significant effect on labor demand. All of the farm characteristics included in the econometric analysis which contribute significantly to the explanation of labor demand are positive. In particular, productive assets (ASSET) and vegetables' share on sown area (VEGETABLE) increase labor demand significantly

X. Wang et al. / China Economic Review 18 (2007) 287–308

303

for households that exclusively hire farm labor (regime h). An increase in livestock production (LIVESTOCK) of 1% (regime oh) will raise hired labor demand by 0.07%. Increasing relative agricultural prices (TOT) also has a positive effect on the use of hired labor. Although we are not able to narrow down the exact use of hired labor if a household participates in both markets, these results indicate that specialization in agricultural production has a significant effect on labor demand. 4.2. Supply of off-farm family labor Turning to the estimation of households' labor supply, the results in Table 4 show a significant impact of the wage of off-farm workers (WAGE-off-farm worker) on labor supply only for households that participate in both markets.18 The estimated wage elasticity is larger than unity and points to an elastic reaction of households to changes in off-farm earnings. Increasing income differentials between non-agricultural and agricultural work will further motivate migration out of agriculture or a reduction of leisure. However, results indicate that households which exclusively supply off-farm labor do not respond significantly to off-farm income changes. This is in line with findings by Lass and Gempeshaw (1992) as well as Findeis and Lass (1994), who obtained statistically insignificant uncompensated own wage elasticities analyzing US farm household data.19 Increasing the wages of off-farm workers (WAGE-off-farm worker) by 1% will increase the average labor supply by 1.34%for households participating in both markets (regime oh). For these households, the hired labor wage is negatively and significantly related to the labor supply of family members in off-farm work. Specifically, increasing the hired-labor wage (WAGE-hired labor) by 1% will decrease off-farm labor supply by 0.07%. Again, this confirms the interdependence between family and hired labor. Increasing the number of working family members affects off-farm labor supply significantly. Interestingly, the coefficient of the female labor force (FLABOR) is significant for both regimes and much larger for households in regime oh. Two explanations seem possible: first, additional female family members substitute for male family workers in household work and farm production; secondly, female family members employed off the farm react in a more elastic way than off-farm employed men. Attachment to social networks approximated by CADRE and PMEMBER reduces labor supply. This effect probably is caused by significantly higher incomes of households with party members (see Table 3). A larger cultivated area (LANDPC) or more labor-intensive production such as husbandry (LIVESTOCK) will reduce off-farm labor supply. However, coefficients are only significantly different from zero for households which hire and supply labor simultaneously. In both regimes, labor supply is quite similarly negatively affected by increasing relative agricultural prices (TOT). Regarding the interdependency between land and labor markets, the estimated coefficients for LRENT are not significantly different from zero in all four specifications. Therefore, integration

18 For off-farm labor supply functions, we conduct Hausman specification tests of the null hypothesis of a randomeffects model in comparison to the alternative hypothesis of a fixed-effects model. Finally, a fixed-effects model is appropriate for households in regime o, while a random-effects model accurately characterizes the relationship between the working days of off-farm workers and the explanatory variables for households in regime oh. 19 However, results in previous studies show no clear picture. Jacoby (1993), for example, finds significant own wage elasticities for the labor supply of Peruvian farm households. On the contrary, Rosenzweig (1980) obtains a negative elasticity estimating Indian male farmers' labor supply.

304

X. Wang et al. / China Economic Review 18 (2007) 287–308

on the land market seems to facilitate the participation in labor markets, on the demand side as well as on the supply side, but has no further impact on the time allocation once households participate in labor markets. 5. Conclusion The adjustment of rural labor markets to economic reform is undoubtedly an important indicator of the progress of transition, although there are also other benefits of emerging labor markets. Despite the emergence of hiring in rural labor markets and despite the fact that offfarm employment appears to be general, there seem to be few studies that assess both dimensions of the labor market in rural China in a completely theoretical framework. Our study fills the gap by exploring the determinants of households' labor participation decisions and identifying which factors enable or constrain the households' ability to hire labor or join the off-farm labor market. In particular, we test whether households' participation decision regarding the two labor markets is a joint or completely separate behavior by applying a bivariate probit model. We contribute to the on-going debate on the rural labor markets by quantitatively assessing the response of households' demand for hired laborers and supply of off-farm workers to the endogenous measure of time value of rural labor, and other exogenous household and village characteristics, whereas the shift in households' production structure occurs inseparably. The empirical estimations use the fixed-point survey data over the period 1995–2002 for 10 villages of Zhejiang province, which has the most advanced rural economy in China. Evidence is found to support the behavioral assumption that hiring and supplying labor is a joint decision within the household. The outcome of this decision is positively correlated; the likelihood of participating in any off-farm occupation increases with hired labor, and vice versa. The major results indicate statistically significant effects of the wages of off-farm workers or those of hired labor on the corresponding hired labor demand and off-farm labor supply functions. It should be noted that households which both hire in and hire out labor respond much more sharply to the change of wages on both markets, compared to households which only participate in one of the two markets. This result confirms that Chinese rural households' time allocation reacts to market price signals as well as household, farm, and local characteristics and indicates functioning labor markets in rural China. The improving labor market is also proven by evidence that the households with larger shares of educated or skilled members are more likely to participate in off-farm employment or the hired labor market. However, that market imperfections may still exist is indicated by a significantly higher participation in labor markets in villages with more activity involving the leasing of land. However, land market integration appears to have no impact on labor time allocation. Appendix A. Definition of variables 1) Hired labor: paid part-time and full-time employees during the year in the household 2) Off-farm labor: reported hours and income from self-employment and wage employment a) Self-employment: labor input in and income from non-agricultural household business such as manufacturing, construction, transportation, retailing, restaurants and other services, and other activities b) Wage work: labor input beyond household-run businesses and resulting incomes

X. Wang et al. / China Economic Review 18 (2007) 287–308

305

Fig. 1

Fig. A1. The location of the ten survey spots in Zhejiang province.

Fig. 2

Fig. A2. The columns report the demand of working days of hired labors at the household level; the line presents the households' participation rate in the hired labor market. Source: RCRE survey for Zhejiang, 1995–2002.

306

X. Wang et al. / China Economic Review 18 (2007) 287–308

Fig. 3

Fig. A3. The columns report the supply of working days of off-farm workers at the household level; the line presents the households' participation rate in the off-farm labor market. Source: RCRE survey for Zhejiang, 1995–2002.

Table A1 Estimated return to labor Year Marginal product of hired labor a (Yuan/day) N 1995 1996 1997 1998 1999 2000 2001 2002 1995–2002 51 46 46 50 68 69 51 61 442 Mean 33.98 27.15 31.86 13.91 38.08 25.15 19.14 19.28 26.29 SD (49.59) (56.76) (52.46) (27.89) (110.91) (53.58) (49.66) (39.62) (61.98) Average wage from off-farm occupation b (Yuan/day) N 258 352 349 348 367 366 327 350 2717 Mean 40.72 44.81 46.21 48.06 51.70 55.37 65.81 63.98 52.37 SD (29.43) (26.90) (30.00) (30.16) (33.08) (35.04) (37.47) (41.16) (34.26)

a The marginal product of hired labor in agriculture is calculated by multiplying the elasticity of hired labor estimated from the production function (Table 3) with the average income of hired labor as MPHired-l = βHired-l × (value of output) / (working days of hired labor). b The average income from off-farm occupation for households that supply labor off-farm is the predicted wage from the off-farm income function (Table 3).

References
Agra Europe Germany. (2005). Labor Excess in Chinese Agriculture (pp. L14). 42/ 05. Becker, G. S. (1965). A theory of the allocation of time. Economic Journal, 75, 493?517. Benjamin, D. (1992). Household composition, labor markets, and labor demand: Testing for separation in agricultural household models. Econometrica, 60, 287?322. Benjamin, D., & Brandt, L. (2002). Property rights, labour markets, and efficiency in a transition economy: The case of rural China. Canadian Journal of Economics, 35, 689?716. Benjamin, D., Brandt, L., & Giles, J. (2001). The evolution of income inequality in rural China. Working Paper 04/01 University of Toronto: Department of Economics and Institute for Policy Analysis. Bhattacharyya, A., & Parker, E. (1999). Labor productivity and migration in Chinese agriculture: A stochastic frontier approach. China Economic Review, 10, 59?74. Bowlus, A. J., & Sicular, T. (2003). Moving toward markets? Labor allocation in rural China. Journal of Development Economics, 71, 561?583.

X. Wang et al. / China Economic Review 18 (2007) 287–308

307

Bruemmer, B., Glauben, T., & Lu, W. C. (2006). The impact of policy reform on productivity and efficiency in Chinese agriculture: A distance function approach. Journal of Development Economics, 81, 61?79. Carter, R. M., & Yao, Y. (2002). Local versus global separability in agricultural household models: The factor price equalization effect of land transfer rights. American Journal of Agricultural Economics, 84, 702?715. Chen, A. Z., Huffman, W. E., & Rozelle, S. (2004). Migration and local off-farm working in rural China. Contributed paper to 2004 AAEA Annual Meeting, Denver, August 1–3, 2004. Cook, S. (1999). Surplus labour and productivity in Chinese agriculture: Evidence from household survey data. Journal of Development Studies, 35, 16?44. de Brauw, A., Huang, J., Rozelle, S., Zhang, L. X., & Zhang, Y. G. (2002). The evolution of China's rural labor markets during the reforms. Journal of Comparative Economics, 30, 329?353. Deolalikar, A. B., & Vijverberg, W. B. (1987). A test of heterogeneity of family and hired labour in Asian agriculture. Oxford Bulletin of Economics and Statistics, 49, 291?303. Dong, X. Y., & Putterman, L. G. (2000). Prereform industry and state monopsony in China. Journal of Comparative Economics, 28, 32?60. Fang, C., Wailes, E., & Cramer, G. (1998). China's rural and urban household survey data: Collection, availability, and problems. Contributed paper to 1998 AAEA Annual Meeting, Salt Lake City, August 5, 1998. Findeis, J. L., & Lass, D. A. (1994). Labor decisions by agricultural households: Interrelationships between off-farm labor supply and hired labor demand. PRI working paper 94-08 Pennsylvania: University Park. Fleisher, B. M, & Yang, D. T. (2003). China's labor market. Conference on China's Market Reforms, Stanford Center for International Development, September 19–20, 2003. Glauben, T. (2000). Besteuerung im Agrarsektor Eine Theoretische und Empirische Analyse auf der Grundlage von Unternehmens-Haushalts-Modellen. Kiel: Wissenschaftsverl. Vauk. Glauben, T., Herzfeld, T., & Wang, X. (2005). Labor market participation of Chinese agricultural households. Contributed paper to XIth conference of the EAAE, Copenhagen, Denmark, August 24–27, 2005. Glauben, T., Herzfeld, T., & Wang, X. (2006). The dynamics of rural households' participation in labor markets. Contributed paper to 26th IAAE conference, Brisbane, Australia, August 12–18, 2006. Goetz, S. J. (1992). A selectivity model of household food marketing behavior in sub-Sahara Africa. American Journal of Agricultural Economics, 74, 444?452. Greene, W. H. (2000). Econometric analysis, 4th edition. Upper Saddle River: Prentice Hall. Hare, D. (1994). Rural nonagricultural activities and their impact on the distribution of income: Evidence from farm households in Southern China. China Economic Review, 4, 59?82. Hausman, J. A. (1980). The effect of wages, taxes, and fixed costs on woman’s labor force participation. Journal of Public Economics, 14, 161?194. Heckman, J. J. (1974). Shadow prices, market wages, and labor supply. Econometrica, 42, 679?693. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153?161. Hertel, T., & Zhai, F. (2006). Labor market distortions, rural–urban inequality and the opening of China's economy. Economic Modelling, 23, 76?109. Huffman, W. E. (1980). Farm and off-farm work decisions: The role of human capital. Review of Economics and Statistics, 62, 14?23. Huffman, W. E., & Lange, M. D. (1989). Off-farm work decisions of husbands and wives: Joint decision-making. Review of Economics and Statistics, 71, 471?480. Jacoby, H. G. (1993). Shadow wages and peasant family labor supply: An econometric application to the Peruvian Sierra. Review of Economic Studies, 60, 903?921. Jacoby, H. G., Li, G., & Rozelle, S. (2002). Hazards of expropriation: Tenure insecurity and investment in rural China. American Economic Review, 92, 1420?1447. Key, N., Sadoulet, E., & de Janvry, A. (2000). Transaction costs and agricultural household supply response. American Journal of Agricultural Economics, 82, 245?259. Knight, J. B., & Yueh, L. (2002). The role of social capital in the labour market in China. Discussion Paper, vol. 121. Department of Economics, University of Oxford. Kung, J. K. (2002). Off-farm labor markets and the emergence of land rental markets in rural China. Journal of Comparative Economics, 30, 395?414. Lass, A. D., & Gempesaw II, M. C. (1992). The supply of off-farm labor: A random coefficients approach. American Journal of Agricultural Economics, 74, 400?411. Liu, Y. H., & Wang, X. (2005). Technological progress and Chinese agricultural growth in the 1990s. China Economic Review, 16, 419?440.

308

X. Wang et al. / China Economic Review 18 (2007) 287–308

Lopez, R. E. (1994). Estimating labor supply and production decisions of self-employed farm producers. European Economic Review, 24, 61?82. Low, A. (1986). Agricultural development in Southern Africa: farm-household economics and the food crisis. London: James Curry. Meng, X. (2000). Labour market reform in China. Cambridge: Cambridge University Press. Ooms, D. L., & Hall, A. R. (2005). On-and off-farm labour supply of Dutch dairy farmers: Estimation and policy simulations. Contributed paper to XIth EAAE Conference, Copenhagen, August 24–27, 2005. Rosenzweig, M. R. (1980). Neoclassical theory and the optimizing peasant: An econometric analysis of market family labor supply in a developing country. Quarterly Journal of Economics, 95, 31?55. Rozelle, S. (1994). Decision-making in China's rural economy: The linkage between village leaders and farm households. The China Quarterly, 137, 99?124. Rozelle, S., Brandt, L., Li, G., & Huang, J. (2002). Land rights in China: Review of the facts, fictions, and issues. China Journal, 47, 67?97. Rozelle, S., Guo, L., Shen, M., Hughart, A., & Giles, J. (1999). Leaving China's farms: Survey results of new paths and remaining hurdles to rural migration. The China Quarterly, 143, 697?730. Sadoulet, E., de Janvry, A., & Benjamin, C. (1998). Household behavior with imperfect labor markets. Industrial Relations, 37, 85?108. Singh, I., Squire, L., & Strauss, J. (1986). A survey of agricultural household models: Recent findings and policy implications. World Bank Economic Review, 1, 149?179. Skoufias, E. (1994). Using shadow wages to estimate labor supply of agricultural households. American Journal of Agricultural Economics, 76, 215?227. State Statistical Bureau (SSB). (2004). Zhejiang Statistical Yearbook. Beijing. State Statistical Bureau (SSB). Statistical Yearbook of China. Various issues, Beijing. Sumner, A. D. (1982). The off-farm labor supply of farmers. American Journal of Agricultural Economics, 64, 499?509. Tokle, G. J., & Huffman, W. E. (1991). Local economic conditions and wage labor decisions of farm and rural nonfarm couples. American Journal of Agricultural Economics, 73, 652?670. Tuan, F., Somwaru, A., & Diao, X. (2000). Rural labor migration, characteristics, and employment patterns: A study based on China's agricultural census. TMD Discussion Paper, vol. 63. Washington D.C.: International Food Policy Research Institute. Waldman, C. (2004). The labor market in post-reform China: History, evidence, and implications. Business Economics, 39, 50?62. Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge: MIT Press. Yang, D. T. (1997). Education and off-farm work. Economic Development and Cultural Change, 45, 613?632. Yang, D. T. (1997). China's land arrangements and rural labor mobility. China Economic Review, 8, 101?116. Yang, D. T. (2004). Education and allocative efficiency: Household income growth during rural reforms in China. Journal of Development Economics, 74, 137?162. Zhai, F., Hertel, T., & Wang, Z. (2003). Labor market distortions, rural–urban inequality and the opening of China's economy. GTAP Working Paper, vol. 27. West Lafayette. Zhang, L. X., Huang, J., & Rozelle, S. (2002). Employment, emerging labor markets, and the role of education in rural China. China Economic Review, 13, 313?328. Zhang, L. X., Rozelle, S., & Huang, J. (2001). Off-farm jobs and on-farm work in periods of boom and bust in rural China. Journal of Comparative Economics, 29, 505?526. Zhao, Y. (2002). Causes and consequences of return migration: recent evidence from China. Journal of Comparative Economics, 30, 376?394.


相关文章:
更多相关标签: