capital deepening, land use policy, and self-sufficiency in china's grain sector

China Economic Review 24 (2013) 95–107

Contents lists available at SciVerse ScienceDirect

China Economic Review

Capital deepening, land use policy, and self-sufficiency in China's grain sector

Junichi ITO a,⁎, Jing NI b

a Division of Natural Resource Economics, Graduate School of Agriculture, Kyoto University, Kitashirakawa Oiwake-cho, Sakyo-ku, Kyoto 606-8502, Japanb Basic Research Division, Japan-Cooperative General Research Institute, Japan

a r t i c l e i n f o

⁎ Corresponding author. Tel.: +81 75 753 6202; faxE-mail address: [email protected] (J. Ito).

1 China's AMS (Aggregate Measurement of Support)exceed the de minimis exemption for product-specificinput subsidies are imposed on Chinese agriculture (H

2 China's soybean production has completely lost itsfrom abroad. Production of maize, one of the most im

1043-951X/$ – see front matter © 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.chieco.2012.11.003

a b s t r a c t

Article history:Received 5 June 2012Received in revised form 12 November 2012Accepted 18 November 2012Available online 27 November 2012

The objectives of this paper are twofold. First, we aim to clarify the mechanism by which use offarmmachinery in Chinese agriculture has grown rapidly over the past decades, using a separateCobb–Douglas (SCD) production function. Second, we determine under what condition willChina's grain self-sufficiency be secured in the next decade. Our empirical results reveal that thesupply and factor demand functions based on the SCD form can significantly explain the reality, inparticular, the capital demand function. This finding suggests that the recent capital deepening inChinese agriculture is caused by farmers' behavior in response to the government's directivesaimed at securing arable land. Our simulation analysis reveals that the attainment of a 95%self-sufficiency rate would be quite challenging for China, unless the terms of trade in agricultureimprove substantially in favor of producers. China's policy makers must therefore seriouslyreconsider whether adhering to the policy goal of grain self-sufficiency is worth the effort.

© 2012 Elsevier Inc. All rights reserved.

JEL classification:D24Q18Q24

Keywords:Capital deepeningGrain self-sufficiencyProduction functionChina

1. Introduction

All member countries of the World Trade Organization (WTO), including China, are restricted more or less to employinginterventional and/or protectionist policy measures, such as price support and subsidy payments, in boosting domestic agriculturalproduction (Zhu, 2004).1 Meanwhile, there are many countries in the world that try to sustain or improve food self-sufficiency, inspite of the fact that their comparative advantage in agriculture is on the verge of being eroded or has been lost entirely. The StateCouncil of the People's Republic of China reiterated in 2008 that the government pursues a national goal of maintaining a self-sufficiency rate for grain at the level of 95%, and complete self-sufficiency for food staples, even under the current circumstanceswhere the grain sector is barely able to sustain its international competitiveness.2

When the land-using characteristics of grain production are taken into consideration, one of themost effectivemeasures to secureself-sufficiency, compatiblewith the international treaty, is to regulate the conversion of farmland for other uses, and then to enhanceland productivity. However, in order for the protected farmland not to be left fallow, a certain amount of resources, including farmlabor andmachinery, have to bemobilized forwork on the fields. Owing to considerable increase in the opportunity cost of farm laborin China for the past few decades, rural labor has significantly migrated out of agriculture. This necessitated the replacement ofdecreasing labor with capital-intensive technologies (Van den Berg et al., 2007). In fact, the past 20 years witnessed a rapid capital

: +81 75 753 6191.

commitment is insignificant in theWTO Agreement on Agriculture. Thus, the government subsidies cannotsupport of 8.5% of the total value of farm product. Moreover, other stringent restrictions on investment anduang & Rozelle, 2008).international competitiveness recently, the consequence of which is a substantial increase in importationportant cereals in China, is also on the verge of losing its international competitiveness.

ll rights reserved.

http://dx.doi.org/10.1016/j.chieco.2012.11.003

mailto:[email protected]

http://dx.doi.org/10.1016/j.chieco.2012.11.003

http://www.sciencedirect.com/science/journal/1043951X

96 J. Ito, J. Ni / China Economic Review 24 (2013) 95–107

deepening in Chinese agriculture, and the speed with which farm machinery spread throughout the country during the period hasoutpaced that of other factor inputs.

The objectives of this paper are twofold. First, we aim to clarify themechanism by which use of farmmachinery has grown rapidlyover the past decades, depending on the production function analysis. This paper, on this regard, hypothesizes that the recent capitaldeepening in grain production is caused by not so much a simple factor substitution induced by a rise in the wage–rental ratio, butrather by farmers' behavior in response to the government's directives aimed at securing arable land. To verify the hypothesis, wespecify the production function in a separate Cobb–Douglas (SCD) form. Although it has rarely been used in empirical studies, ourresults show that the SCD outperforms the CD in terms of an explanatory power of the time-seriesmovement of farmmachinery input.

Second, we determine under what condition China's grain self-sufficiency will be secured in the next decade, with a specialemphasis on the terms of trade in agriculture, land use policy, and technological progress. Analystswho tackle this important questionusually employ computable general equilibrium (CGE) models (e.g., Chen & Duncan, 2008; Felloni, Gilbert, Wahl, & Wandschneider,2003; Yang & Tyers, 1989). However, considering the fact that the large variations in projected supply–demand balance comeprimarily from the production side (Fan&Agcaoili-Sombilla, 1997), we shed light only on supply responses to price changes, land use,and technological change, relying on an outside source for the demand analysis. The major goal of this paper is not to predict theself-sufficiency rate accurately, but rather to identify the key factors that influence grain supply capacities. Our simulation analysisreveals that Chinawould find it very difficult tomaintain the grain sufficiency rate at the current level, even if the terms of trade in thenext decade follow a similar track in the past 20 years. Although it is possible that the more stringent regulations on farmlandconversion and accelerated productivity increase can help attain the policy goal, this is not likely to occur without a more seriouscommitment by the government to farmland preservation and intervention in grain market.

The remainder of this paper is organized as follows. In Section 2, we provide relevant background information on grainproduction and land use policy in China. Section 3 contains a brief explanation of the empirical model. In Section 4, we estimatethe two types of production functions, the SCD and CD, and compare their explanatory powers for grain supply and factordemands. This section also shows the simulation results with respect to grain self-sufficiency. Finally, Section 5 concludes with abrief summary of our results and draws some policy implications.

2. Grain production and land use policy

2.1. Grain production

Fig. 1 illustrates how the producer's price of grain, prices of fertilizer and farmmachinery, and the non-farmwage rate in rural areashave changed over the past two decades. We computed the Laspeyres index of grain price using FAOSTAT (Food and AgricultureOrganization Statistical Database) that provides data on domestic producer's prices of wheat, rice, maize, soybean, tubers (starchy roots)and others (barley, oats, and rye), and the corresponding output quantities. Data on factor input prices are obtained from the ChinaStatistical Yearbook (CSY) published by theNational Bureau of Statistics. The non-farmwage rate in rural areas ismeasured by rewards oftownship andvillage enterprise (TVE) employeesdivided by the total number of employees (thedata source is the Statistical Year BookofChina's TVEs). Over the past two decades, the prices of fertilizer and farmmachinerywere relatively stable, while that of grain fluctuatedsignificantly. The non-farm wage rate has consistently increased for the period 1991–2009 with an annual rate of more than 10%.

The year 1991 is a turning point for China's agricultural price policy in the sense that the government introduced the price-supportprogram for the first time in the interest of farmers, putting an end to the scissors-form differential price system (or the below-marketprocurement pricing system) that had been in place. However, the reform, accompanied by a rapid rise in producer's prices and

0

100

200

300

400

500

600

700

800

900

0

50

100

150

200

250

300

350

400

450

1990 1995 2000 2005 2010

Grain Fertilizer Machinery Rural wage (right axis)

Fig. 1. Price indexes.Sources: China Statistical Yearbook (CSY), Statistical Year Book of China's Township and Village Enterprises (TVEs), and Food and Agriculture Organization StatisticalDatabase (FAOSTAT).

60

70

80

90

100

110

50

100

150

200

250

300

1990 1995 2000 2005 2010

Fertilizer Machinery

Small tractors Labor (right axis)

Sown areas (right axis)

Fig. 2. Factor input indexes.Sources: China Statistical Yearbook (CSY).

97J. Ito, J. Ni / China Economic Review 24 (2013) 95–107

overstocked inventory, created a huge financial burden, throughwhich the government reversed the pricing policy toward liberalizationbefore China acceded to the WTO in 2001. Thus, we can say that the wide variation in the grain price during the 1990s reflects thetransformation of the agricultural price policy fromprotectionist tomarket-oriented (Huang, Liu,Martin, & Rozelle, 2009;Martin, 2001).

Fig. 2 shows the index of factor inputs that are used for grain production. Section 4 explains how the data are processed. Farmlabor has decreased over the years, declining more pronouncedly from the 2000s onward. An increase in the non-farm wage rate,shown in Fig. 1, is considered to be a major contributor to the migratory movement out of agriculture. While the consumption offertilizer stagnated from the mid-1990s to the outset of the 2000s, it increased thereafter. Although sown areas for grainproduction remained virtually unchanged during the 1990s, it decreased continuously until 2003, and then turned around in2004. The fluctuation of sown areas is discussed in more detail below in the context of land use policy.

Of particular interest in Fig. 2 is the fact that farmmachinery inputmeasuredbyhorsepower kilowatt and thenumber of small tractorsgrewmore rapidly than fertilizer use from the late 1990s. The number of large tractors, although not shown in Fig. 2, has increasedmorequickly than small tractors since the turn of the century. Indeed, farmmechanization is relatively easier in grain production, compared toproduction of other cash crops, but this is somewhat surprisingwhenwe consider the situation in Chinawheremost farmers face severeliquidity constraints in the presence of credit market imperfection, and therefore cannot afford to purchase farmmachinery (Feder, Lau,Lin, & Luo, 1992). According to a household survey byXi and Chen (2007) conducted in Shanxi Province, 66% of the sampled farms do notown farm machinery, and 68% have no intention to purchase it in 1 or 2 years. This suggests that instead of purchasing machineryequipment on their own, they cultivate their farmland using machinery services provided by some specialized farm households and/ororganizations (Carter, Zhong, & Zhu, 2012; Ji, Yu, & Zhong, 2012). Anecdotal evidence also indicates that this is why credit constraintsassociatedwith imperfect capitalmarket do notmatter for Chinese farmer tomechanize the crop production process. Since the farm sizeof Chinese agriculture is quite smallwith around 0.6 ha per household, this type of farmmechanization is quite rational and economical.3

2.2. Grain self-sufficiency and land use policy

Using the FAOSTAT, we computed the grain self-sufficiency rate, defined as production divided by domestic supply quantity, theresults of which are illustrated in Fig. 3. Grain includes wheat, rice, maize, soybean, tubers, and others. A caveat is needed to interpretthe figure, because it takes no account of the self-sufficiency rate of each crop and the changing composition of grain production andconsumption.We computed another indicator of grain self-sufficiency rate in terms of land use.4 Fig. 3 shows that the self-sufficiencyrate measured by land use is always lower than that in terms of weight, with growing discrepancies between the two indicators. Thisis due to the fact that soybean whose yield is far lower than that of other crops increased its import volume substantially in recentyears. Although the self-sufficiency rate in terms ofweight had never been less than 95% at least since the 1980s until 1999, it droppedsharply in 2000, and dipped slightly below 90% during 2001–2003. Although not shown in Fig. 3, grain production recorded thelowest level in 2003 over the past 20 years at 431 millionmetric tons.5Meanwhile, the ratio of sown areas for grain production to thetotal areaswas also on a downward trend in the 1990s, becomingmore pronounced during 1999–2003. The loss of cultivated land onthe whole has also significantly accelerated since the mid-1990s, especially during the first several years of the 2000s (Chen, 2007).

3 Agricultural machinery stations and/or rural producer organization play an important role in providing farm households with farm machinery services.4 We estimate sown areas that are required for meeting the total consumption of grain with domestic supply, using data on the yield and consumption quantity

of wheat, rice, maize, soybean, tubers, and others.5 Statistics from the CSY follow the rule that the output of tubers (sweet potatoes and potatoes, not including taros and cassava) are converted into that of grain

at the ratio 5:1, that is, 5 kg of fresh tubers are equivalent to 1 kg of grain.

65

70

75

80

85

60

65

70

75

80

85

90

95

100

105

1980 1985 1990 1995 2000 2005 2010

Self-sufficiency rate (weight) Self-sufficiency rate (land use)

Ratio of grain sown areas (right axis)

Fig. 3. Grain self-sufficiency rate and sown area ratio for grain production.Sources: China Statistical Yearbook (CYS), and Food and Agriculture Organization Statistical Database (FAOSTAT).


However, the sown areas for grain production increased by 10.5 million hectares between 2003 and 2010 (or equivalently, a 10.5%increase for the period), while the quantity of grain production reached a record high of 546 million metric tons in 2010. As a result,both the self-sufficiency rate and the sown area ratio rebounded sharply from 2004.

It is not surprising that the self-sufficiency rate fluctuated in tandem with the ratio of sown area, when the land-usingcharacteristics of grain production are taken into consideration; the sown areas are likely to directly affect the quantity of grainproduction, other things being equal. Yet, the reverse causality is also likely in the case of China. Policy makers who are deeplyconcerned about food security associated with a fall of the self-sufficiency rate may commit strongly to the determination ofarable land. In 2004, their anxiety culminated in the wake of five successive years of decreasing grain sown areas and production(Deng, Huang, Rozelle, & Uchida, 2006).

In response to these issues, the State Council of the People's Republic of China published a document in 2008 entitled “TheMedium- and Long-Term Plan for National Food Security.” It refers to the adherence of grain self-sufficiency at the level of 95%and the securing farmland for that purpose. Another document, “The Comprehensive Plan for National Land Use” released in 2006stipulates that the cultivated acreages in 2010 and 2020 shall not be less than 1.818 billion Mu and 1.805 billion Mu, respectively(1 Mu=1/15 ha). It also emphasizes the need to retain basic farmlands for grain production with no less than 1.56 billion Mu. Inaddition to the simple emphasis on retaining such areas, the document refers to the necessity to ensure the quality of such lands(Chen, 2007). The Basic Farmland Protection Regulation promulgated in 1994 and its successive amendment set a strict limitationon the exploitation and conversion of farmland under the comprehensive plan of land use. In order to ensure that county andtownship governments observe these guidelines, thereby stopping the rampant development of farmland associated withurbanization, the upper government must undertake obligations to oversee the operations.

Back in 1986, the government enacted the Land Management Law (LML) that continues to legally predominate all regulationsand directives regarding land use policy in China. The revised version of the LML in 1999 “requires provincial governments toadopt measures to ensure that the total amount of cultivated land within its administration region is not reduced, and if reducedto take responsibility for the reclamation of an equal amount of land within its administration region or in a different location”(Ho & Lin, 2003). Specifically, the new LML imposes a so-called “dynamic balance (no net loss) of cultivated land” policy with aview of keeping the total amount of basic farmland constant. It has been believed that this strategy is the most effective for Chinato stabilize food supply and sustain food security (Lichtenberg & Ding, 2008; Yang & Li, 2000).

However, it turned out that a series of land use regulations alone was not sufficient to produce the initial momentum. Thus, thegovernment decided to introduce the Producer Subsidy Program in 2004. This programwas designed to promote grain production, andat the same time to enhance farm income, by way of direct payments to farmers on the basis of sown areas they cultivate. In someregions, the subsidies are paid to farmers based on total sown areas, instead of those sown to grain crops. The Producer Subsidy Programis composed of five operations: (a) direct payments to grain growers; (b) subsidy for the adoption of excellent varieties; (c) subsidy forfarm machinery purchase; (d) comprehensive support for the purchase of production materials; and (e) agricultural insurance. Thesubsidy payment of (d) accounts for more than half of the total amount at present. The total amount of the payment increased from14.5 billion Yuan in 2004 to 143.8 billion Yuan in 2010, with the result that the ratio of the amount to the total agricultural productionvalue reached 4% in 2010. Meanwhile, the budgetary expenditure defrayed for the program accounts for more than 20% of theagricultural appropriation at the central government level in 2010. Since China has a regional comparative advantage in cash cropproduction, rather than in grain production, due to its factor endowments characterized by scarcity of arable land relative to farm labor,the program must have helped convert more rural resources inclusive of farmlands for grain production than otherwise possible.

In addition to the Producer Subsidy Program, the government implemented the Grain Purchasing Policy at the Minimum Prices forwheat and rice in 2004, and the Temporary Stockpiling Policy for maize and soybean in 2007 and 2008, respectively. The government is


supposed to implement these programs onlywhen themarket prices fall below theminimumprices, and apply them only to the regionsof leading producers (Carter et al., 2012). However, there is no doubt that the bounce back of grain production after 2003 is a reflection ofenhanced motivations for farmers to produce grain more driven by a wide array of subsidy payment and price support schemes.6

3. Estimation model

The Cobb–Douglas or trans-log form has beenmost widely used for the quantitative analyses of agricultural production technology.However, this paper specifies the production function in the following SCD formwith reference to Egaitsu (1979) and Ito (2010):

Substit

From E

6 Thegrain pr

7 Weinput.

8 The9 The

Howeveemerge

Q ¼ exp að ÞVavSas ð1Þ

S ¼ exp bð ÞLaLKaK ð2Þ

Q, V, S, L, and K denote grain output, chemical fertilizer, sown areas, farm labor, and capital input, respectively. Both exp(a) and
whereexp(b) represent the technological levels. The original form of the SCD can be seen in Evenson and Kislev (1975), and Kislev andPeterson (1982). It is expressed as the following two sub-processes in agricultural production: Q=F[fb(V,S), fm(L,K)], where fb(V,S)and fm(L,K) represent biochemical (BC) and machinery (M) technology, respectively (Mundlak, 2005). They argue further that Mtechnology, produced in the first stage, is combined in the second stage with intermediate input (fertilizer) and land to produceoutput. Thus, we have Q=F(V,S). Egaitsu (1979) contends that labor and capital (farmmachinery) inputs are used substitutively forthe cultivation of farming land, from which the first-stage production process can be described as S=G(L,K).
Although the SCD is not immune from the restrictive assumption on input substitutions, it succeeds in capturing agriculturalproduction characteristics, which can be broadly categorized into BC and M technologies. The former mainly contributes to anincrease in land productivity through the intensive use of intermediate inputs, the adoption of new varieties, and the improvement ofsoil and genetic resources, while the latter contributes to the replacement ofmanualworkwith farmmachinery. Eq. (1) indicates thatland productivity, Q/S, exhibits decreasing returns to the intensity of fertilizer under the condition that 0bαVb1 and αV+αS=1.Meanwhile, M technology exhibits increasing returns to scale when αL+αK>1.

We assume that farm households determine factor inputs except sown areas such that short-term profit may be maximizedsubject to the production function.7 Since it is widely believed that the marginal product principle of agricultural labor is violated forChinese agriculture (Cook, 1999; Liu & Wang, 2005; Yang & Zhou, 1999), we assume

p∂Q∂L ¼ θw 0bθb1 ð3Þ

p and w denote the producer price of grain and the non-farm wage rate in rural areas, respectively. Eq. (3) means that the
wheremarginal value product of labor (MVPL) is less than w, which indicates that the grain sector remains overstaffed.8 The institutionalimpediments tomigration, known as the household registration (hukou) system, gives rise to the pervasivewage income differentiationin China, as suggested by several scholarly works (e.g., Chan & Zhang, 1999; Fan, 2008; Hertel & Zhai, 2006; Meng & Zhang, 2001).9 Forfertilizer, we assume p(∂Q/∂V)=u, where u denotes fertilizer price. Further, Model I assumes that capital input is determined such thatthe rate of change in sown areas (d ln S) is kept constant. By differentiating Eq. (2) and letting d ln S bem (constant), we have
d lnK ¼ −dbþ αLd lnL−mαK

ð4Þ

uting Eq. (4) into the labor demand function derived from the short-term profit maximization, we have

d lnL ¼ daαS

þ d lnpθw

� �þ αV

αSd ln

pu

� �þm ð5Þ

qs. (4) and (5), we have

d lnQ ¼ daαS

þ αV

αSd ln

pu

� �þm: ð6Þ

Model II specifies the production function in the Cobb–Douglas (CD) form: Q ¼ exp cð ÞVβV SβS LβL KβK , and assumes that farmhouseholds choose capital input, along with labor and fertilizer inputs, such that short-term profit may be maximized. Thus, the

temporary halt of the “Grain for Green” (or conservation set-aside) program that started in 2007 is also responsible for the recovery of sown areas foroduction in the mid-Western regions.can solve the short-term profit maximization problem even if the production technology exhibits constant returns to scale, because sown areas are fixed

Ministry of Agriculture announced that the surplus rate of agricultural labor in China reached 47% in 2004.hukou system does not directly affect the wage differential between agriculture and TVEs, because it restricts rural–urban migration exclusively.r, due to the institutional impediment to migration, a considerable number of rural people are forced to stay in their home village, which causes thence of the wage differential between farm and non-farm sectors in rural areas.


following equations are derived:

10 Linoutput.11 Thestatisticproductconfirm

d lnQ ¼ 1βS

dcþ∑X;q

βXd lnpq

� �" #þm X ¼ V ; L;K; q ¼ u; θw; rð Þ ð7Þ

d lnX ¼ d lnpq

� �þ 1βS

dcþ∑X;q

βXd lnpq

� �" #þm: ð8Þ

We have d ln(K/L)=d ln(θw/r) from Eq. (8), which suggests that a rise in the wage–rental ratio results in an increase in thecapital–labor ratiowith the elasticity of substitution equal to one.Meanwhile, we see from Eqs. (4) and (5) that a rise in thewage ratein Model I results in a decrease in labor input and an increase in capital input. It follows that both Model I and Model II are able tocapture factor substitution between labor and capital inputs. However, the effects that terms of trade in agriculture have on the factordemands are completely different between the two models. Let us confirm this inference based on the assumption that m=0.

Eqs. (4) and (5) show that labor input decreases when the terms of trade worsen (d ln(p/θw)b0, d ln(p/u)b0), and da/αS is notlarge enough to cancel out the effect, which results in an increase in capital input, as long as db is non-positive. That is, farmershave to replace decreasing labor with capital input for sown areas to be kept constant. On the other hand, Eq. (8) shows thatm=0 is possible, even if both labor and capital inputs decrease simultaneously due to the worsening terms of trade. This, however,appears at odds when we take into account the trait of the M technology inherent to agriculture in which sown areas dependheavily on how much labor and capital inputs are mobilized on the fields for cultivation. The empirical results below will verifywhich model is better able to explain the reality.

4. Empirical results

4.1. Data and the estimation of production functions

This paper uses province-level data from 1991 to 2009 to estimate the agricultural production functions. Although Chongqingwasseparated from Sichuan Province in 1997 as amunicipal administrative area, they have been integrated in this analysis. As a result, 30provincial data are available for each year. The data source is the CSY. Output was measured as the grain production value at 1991constant prices (100 million Yuan). Fertilizer was measured by the weight of its net ingredients (10,000 tons). Other inputs such aspesticide, farm manure and seed for agricultural use should have been included as intermediate inputs. However, data on thesefactors are not available. Labor input was measured as the total number of workers engaged in grain production (10,000 people).Capital inputwasmeasured by farmmachinery power used for grain production (10,000 kW). The CSY provides data on factor inputsused for agricultural production on the whole, not only for grain production, with the exception of the sown areas. As Zhang and Fan(2001) argues, one of the most difficult problems for the estimation of grain production function is that crop-specific input data arenot available. We estimate factor inputs for grain production by prorating the whole amount by the sown area ratio, the methods ofwhich are in line with Yao, Liu, and Zhang (2001).10 The Appendix Table shows the descriptive statistics of the panel data.

Other explanatory variables of the production function include time trend as a proxy for technological change, the irrigation rate,the natural calamity index, the average temperature, its square, the annual amount of precipitation divided by 1000, its square. Tocontrol for the heterogeneity of output and land input, the ratios of sown areas of wheat, rice, maize, soybean, and tubers to the totalsown areas for grain are included (Liu & Zhuang, 2000). Further, in order to identify the effects of two distinct policies, namely thePrice Support Program (1991–1999) and Producer Subsidy Program (2004–present) on grain production, we added two dummyvariables into the regression equations.

The sown areas, S, is an explanatory variable of Eq. (1), and is also a dependent variable of Eq. (2). With the aim of controlling forthe simultaneous bias, this paper estimated the parameters of the SCD production functionwith the three-stage least squaresmethod(3SLS) with the restriction of αV+αS=1. Since panel data are available, we used the fixed effect model for the estimation. For the CDfunction estimation, we chose the fixed effect model based on the Hausman test (an estimated p-value: 0.00). The argument inSection 3 indicates that factor demand is a function of exogenous variables such as relative prices and sown areas. This inevitablyposes an endogeneity problem for the estimation of the production functions. Although the relative factor prices are eligible for thevalid instruments, the data are not available for each province. Thus, this paper, at the expense of an appropriate identificationstrategy, estimates the production function first, and then uses the estimated parameters for the computation of Eqs. (4)–(8).

Table 1 shows the estimation results of the production functions (the coefficients regarding the ratio of sown areas of eachcrop are not shown). We estimated the CD function with the restriction of βV+βS+βL+βK=1, because it was not rejected at the1% level of significance. The significance levels of fertilizer and labor inputs for the CD function are quite low. In addition, theoverall explanatory power of the regression is not high. The sum of the parameters αL and αK is less than unity, and the nullhypothesis αL+αK=1 is rejected, indicating that the M technology of the SCD function exhibits decreasing returns to scale.11

(1992), and Zhang and Carter (1997) employ an alternative approach to estimate factor inputs according to the value share of crop output in total agriculturalWhichever method we choose to employ, some restrictive assumptions concerning production technology must be impose (Zhang & Fan, 2001).re is no containment relationship between the CD and SCD, as in the case where the CD function is nested in translog model. Thus, we cannot testally which function form is appropriate for the analysis. A critical assumption of the SCD is that the labor input intensity (L/S) has no influence on landivity. From the fixed effect estimation of Q=S ¼ exp að Þ V=Sð ÞαV L=Sð ÞαL , we find that the parameter of αL is positive but insignificant (t-value: 1.49), whichs the validity of the assumption.

Table 1Estimation results of the production functions.

SCD (fixed effect, 3SLS) CD (fixed effect)

Estimates z-values Estimates t-values

BC technologyFertilizer 0.109*** 2.59 Fertilizer 0.032 0.85Sown areas 0.891*** 21.23 Sown areas 0.890*** 15.86Time 0.008*** 3.82 Labor 0.011 0.29Price support dummy 0.028*** 2.66 Machinery 0.067** 2.33Producer subsidy dummy −0.001 −0.04 Time 0.007*** 2.62Irrigation rate 0.009 0.23 Price support dummy 0.035*** 3.17Natural calamity index −0.090*** −2.90 Producer subsidy dummy 0.001 0.04Temperature 0.062*** 3.16 Irrigation rate 0.022 0.54Temperature2 −0.002*** −2.96 Natural calamity index −0.075** −2.30Precipitation 0.177*** 3.20 Temperature 0.064*** 3.15Precipitation2 −0.063*** −3.17 Temperature2 −0.002*** −2.94M technology Precipitation 0.166*** 2.90Labor 0.387*** 15.16 Precipitation2 −0.059*** −2.85Machinery 0.361*** 23.65Time −0.018*** −11.25BC R2 0.994 Overall R2 0.474M R2 0.996Sample size 570 Sample size 570

Note: The symbols *, ** and *** refer to the 1%, 5% and 1% significance levels, respectively.


Egaitsu (1979), using data on Japanese rice production, estimatesαL+αKmore than unity. Likewise, Hayami and Kawagoe (1991),depending on the same data source, estimates economies of scale by way of the cost function approach. Although the agriculturalproduction structures in China and Japan have much similarity in terms of the predominance of small-scale farms and farmlandfragmentation and as well as a rapid rise in the farmwage for the past decades, the way scale economies emerge differs significantlybetween the two countries. This is due to the fact that most farmers in Japan have purchased machinery equipment on their own,while Chinese counterparts have outsourced many field works such as land preparation and harvesting to specialized serviceproviders, as discussed above. This means that indivisibility of farmmachinery combined with immobility of land transaction causesincreasing returns to scale in Japanese agriculture.12 Meanwhile, farm mechanization in China is not incompatible with small-scaleoperation because of the absence of scale economies associated with divisibility of machinery services (Liu & Zhuang, 2000).

The SCD estimates suggest that the annual growth rate of BC technological progress is equal to 0.8%, while that of M technologicalprogress is equal to−1.8% (both coefficients are significant at the 1% level).13 The CD function estimates suggest that the growth rateof neutral technological progress is equal to 0.7%. The coefficient of irrigation rate is positive, but not significant. The coefficient of thenatural calamity index is negative and statistically significant, consistent with our expectation. The average temperature and annualprecipitation bear an inverse “U-shape” relationship with production. The two policy dummy variables indicate that productiontechnology progressed more rapidly during the period when the Price Support Program was implemented than in the period whenthe Produce Subsidy Program was implemented.

4.2. Estimation fitness and causes of capital deepening

Using Eqs. (6) and (7), we computed d ln Q. The parameters and values of da/αS and dc/βS are obtained from the productionfunction estimates. The price data in Fig. 1 are available for themeasurement of d ln p and d ln q (q=u,θw,r) for each year. The value ofθ is computed fromEq. (3) as the ratio ofMVPL tow, fromwhich d ln θ is obtainable. The value of d ln S for the nationalmeanwas usedfor the measurement of m. Using these data, we computed d ln Q, and then converted it into the index with 1991 as the base year.Fig. 4 depicts the results, showing that both Model I and Model II have high explanatory powers.

The labor and capital demand indexes were also computed by similar methods, and are depicted in Figs. 5 and 6, respectively.Fig. 5 shows that both Model I and Model II underestimate the labor demand index in the late 2000s, but the discrepancy is notsignificant. Meanwhile, Fig. 6 shows that for Model II there is a large discrepancy between the estimated and actual values of

12 Indivisibility of farm machinery provides farmers with incentives and opportunities to enlarge their operational size. However, imperfect land market andprotectionist policies for small-scale farms in Japan have been preventing such an agrarian structural change from occurring. Scale economies are thus considereddisequilibria in the sense that small-scale farms with high average cost survive in spite of their production inefficiency (Hayami & Kawagoe, 1991). The absence ofscale economies in China does not imply that the concentration of landholdings is not necessary.13 At first glance, a negative growth rate of M technology appears strange given the rapid modernization of M technology. However, this is not surprisingbecause the structure of China's grain sector in terms of per-worker sown areas has not improved considerably during the period concerned. As shown in Ito(2010), M technology stagnated during 1991–2004 in most provinces, but it grew faster in Zhejiang and Jiangsu Provinces where farm labor has become scarcerelative to arable land. Another conceivable reason for a negative growth rate of M technology lies in the fact that this paper measures labor input as the totalnumber of workers engaged in grain production. It is easy to imagine that the degree to which working hours of grain production decreased over the past decadesis more substantial than the number of labor force. The production function estimate using such data could capture the labor saving effect of M technology, whichmay result in an increase in its growth rate. However, data on working hours are not available.

90

95

100

105

110

115

120

125

1990 1995 2000 2005 2010

Actual value Model I Model II

Fig. 4. Grain supply index.

50

60

70

80

90

100

110

1990 1995 2000 2005 2010


Fig. 5. Labor demand index.


capital demand. In contrast, the performance of Model I is quite satisfactory although there exists some discrepancy in the late2000s. Model I outperforms Model II because compared to the CD function, the SCD is better able to capture the factorsubstitution between labor and capital inputs in response to the change in the terms of trade. In other words, the poorperformance of Model II stems arguably from its failure to describe the M technology specific to agriculture. If this is the case, arecent capital deepening results not so much from a simple replacement of decreasing labor with farm machinery induced by arise in the wage-rental ratio, but rather farmers' response to the government's directives aimed at securing arable land.

4.3. Simulation 1: factor demands, grain supply, and farm income

Our focus in this subsection is on simulating the effect of output price change on factor demands, grain supply, and per capitafarm income. The methods for the analysis are in principle the same as those employed in Section 4.2. We compute d ln Q and d lnX (X=V, L, K) by substituting exogenous variables into Eqs. (4)–(8). Per capita farm income is given by y=(pQ−uV−rK)/L. Weassume that an annual rate of change in grain price (d ln p) runs from 1.8% through 7.8%. For comparison, d ln p is equal to 4.8%during the period of 1991–2009. Annual rates of change in factor prices (d ln q) are fixed at the average values for the periodconcerned, with fertilizer, farm machinery, and non-farm wage rates equal to 3.8%, 1.4% and 10.6%, respectively.14

According to the OECD-FAO (2011), agricultural commodity prices in the world market at the constant prices are likely toremain on a higher plateau during the next 10 years, compared to the previous decade. Rosegrant et al. (2008) also show that theinternational food prices are anticipated to continue on an upward trend in the medium- to long-term. On the other hand, thegeneral price of production materials in Chinese agriculture has increased annually by 4.3% for the past 20 years, according to the

14 This paper computes the average rates of change in prices and quantities by regressing the log variables on the time trend.

50

100

150

200

250

300

350

1990 1995 2000 2005 2010


Fig. 6. Capital demand index.


CSY. If the factor prices in the future follow the same track as in the past, and the terms of trade in Chinese agriculture move alongwith those in the global market, then d ln p>4.3% is likely in the near future, as a rule of thumb.15

Table 2 shows the estimation results. We assumed in this simulation that d ln S=−0.4% (the sown areas decrease at the samerate as in the past 20 years). The table shows that Model I outperforms Model II in terms of the prediction power. Mostnoteworthy is the fact that d ln K is negatively related to d ln p for Model I, but is positively related to d ln p for Model II, indicatingthat d ln K responds to a change in d ln p in an opposite direction in the two models. The results of Model I predict that theworsening terms of trade bring about the movement of labor out of agriculture, compelling farmers to use machinery input moreintensively in order to keep the rate of change in sown areas constant, consistent with Eqs. (4) and (5). The results of Model II arealso consistent with the theoretical prediction of Eq. (8).

The growth rate of per capita farm income in Model II is given by d ln y=d ln(pQ/L)=d ln(MVPL/αL) when the equilibriumconditions with respect to fertilizer and capital inputs are met.16 From Eq. (3), this can be rewritten as d ln y=d ln(θw/αL). To theextent that both θ andw are not influenced by d ln p, d ln y is independent of d ln p. As shown in Table 2, per capita farm income growthin Model II remains constant at 8.2%, irrespective of d ln p. This appears at odds, but at the same time consistent with the theoreticalprediction. In Model I, d ln y depends heavily on d ln p, and they are positively associated, which is fairly consistent with our intuitiveunderstanding. The major reason for the decrease in d ln y in response to a decline in d ln p is the fact that farmers incur increasedcapital costs.17 Indeed, incremental costs of farm machinery adversely affect farm income, but the replacement of farm labor withmachinery may help farmers to gain access to non-farm earnings opportunity, which is likely to enhance their economic welfare.18

4.4. Simulation 2: grain self-sufficiency

Using the results in Section 4.3, we analyze how Model I and Model II predict the grain self-sufficiency during the next 10 years(from 2011 to 2020).19 We obtained data on the predicted quantity of demand for grain in China from OECD-FAO (2011). With theaim of considering the effect of grain price change on the demand quantity, we used the own-price elasticity of−0.3 (Zhang, Mount,& Boisvert, 2001; Zhuang & Abbott, 2007). This treatment, however, does not exert a significant influence on the results. The growthrate of d lnQ shown in Table 2was used to compute the quantity of grain supply during the forecast period. The self-sufficiency rate in2010 is equal to 91.4%.

Fig. 7 depicts the computation results. Model I with d ln p=4.8% predicts that the self-sufficiency ratewill decline to 89.8% in 2020in a linear fashion, which is roughly in accordancewith the forecast of OECD-FAO (2011). This strongly suggests that themaintenanceof the self-sufficiency rate at the current level is impossible for the next decade, even if the terms of trade follow the same trend as inthepast 20 years.Model IIwith d ln p=4.8% predicts that the self-sufficiency ratewill go downmore rapidly than in the case ofModel I,

15 Yang, Qiu, Huang, and Rozelle (2008) claim that at least during the past decade, China's open borders have ensured that its domestic food prices mostly followthose of international prices, except for the period of the world food crisis in 2007–2008.16 The equilibrium conditions with respect to fertilizer and capital inputs in Model II are given by d ln X=d ln(p/q)+d lnQ (X=V, K; q=u, r). Substituting theseequations into the definitional identity of per capita farm income growth rate, we have d ln y=d ln(pQ/L).17 We assume that the ratio of Y/pQ is equal to 0.55 for the computation of d ln y. The data source is the Document on the Costs and Revenues of Farm Products(China Statistics Press). Model I estimates d ln y as equal to 4.8%, when d ln p is 4.8%. According to the CSY, the actual growth rate of per capita farm income during1991–2009 is 6.5%. This discrepancy is considered to stem from the fact that farm households sampled in the CSY include those that are engaged not only in grainproduction but also in other farm activities, including those in the horticulture, livestock, and aquaculture industries.18 According to the CSY, the per capita net income of rural households is 5,919 Yuan on average in 2010, with wage income accounting for more than 40% of thetotal income.19 Our simulation analysis ignores the impact of increasing degradation of environmental stress represented by high degrees of soil salinity and water shortageon grain supply (Chen, 2007; Huang, Rozelle, & Rosegrant, 1999; You, Spoor, Ulimwengu, & Zhang, 2011).

Table 2Simulation analyses (%).

Grain price Fertilizer Sown areas Labor Machinery Grain supply Per capita farm income

d ln p d ln V d ln S d ln L d ln K d ln Q d ln y

Actual 4.8 2.6 −0.4 −2.2 5.5 0.70 n.a.Model I(SCD)

1.8 −1.6 −0.4 −6.0 10.3 0.32 0.52.8 −0.5 −0.4 −4.9 9.1 0.44 1.93.8 0.7 −0.4 −3.8 7.9 0.56 3.44.8 1.8 −0.4 −2.7 6.7 0.69 4.85.8 2.9 −0.4 −1.5 5.5 0.81 6.36.8 4.0 −0.4 −0.4 4.3 0.93 7.77.8 5.1 −0.4 0.7 3.1 1.05 9.1

Model II(CD)

1.8 −1.8 −0.4 −6.3 −2.4 0.07 8.22.8 −0.7 −0.4 −5.2 −1.2 0.20 8.23.8 0.4 −0.4 −4.0 −0.1 0.32 8.24.8 1.5 −0.4 −2.9 1.0 0.44 8.25.8 2.7 −0.4 −1.8 2.1 0.57 8.26.8 3.8 −0.4 −0.7 3.3 0.69 8.27.8 4.9 −0.4 0.5 4.4 0.81 8.2


suggesting that Model II provides more pessimistic prospects on self-sufficiency. Model I with d ln p=1.8% gives a more devastatingforecast. In contrast, the grain self-sufficiency rate shows an upward tendency when d ln p=7.8%.

Finally, we conduct a sensitivity analysis, looking upon Model I with d ln p=4.8% as a benchmark. This paper presents the followingscenarios. Scenario 1 is the case where the sown areas are held constant at the current level for the next 10 years. Specifically, d ln S=0%,instead of d ln S=−0.4%, is assumed. Scenario 2 is the casewhere BC technology progresses twice as fast aswhatwas actually estimated.20

Underlying this scenario is the fact that the yield of coarse grains is quite low in China by international standards, and therefore still hasmuch room for growth (Felloni et al., 2003). Scenario 3 is the casewhere BC technology progresses at the rate of 1.6%under d ln S=−0.8%.Fig. 8 shows that in Scenario 1 the grain self-sufficiency rate would show an increasing trend, and reach 93.3% in 2020. In Scenario 2, theaccelerated BC technological progress would have a more significant and positive effect on grain supply, the consequence of which is therestoration of grain self-sufficiency to the level of 98.5% in 2020. Scenario 3 predicts that the supply–demand balance would improve,suggesting that a productivity increase compensates for a decrease in sown areas (Deng et al., 2006; Lichtenberg & Ding, 2008).

Needless to say, Scenarios 1 to 3 would not be realized without additional costs. Due to consumers' dietary shift associated withrapid income growth and urbanization, Chinese farmers today are strongly motivated to convert their farm products from staplecereals into cash crops. Nevertheless, if the government clings to the idea of grain self-sufficiency, it not only has to tighten thecontrols on the farmland conversion more than ever, but also has to provide farmers with an economic incentive to continue grainproduction. In this regard, we should be reminded of the fact that the ratio of sown areas for grain production did not rebound untilthe government intervention in the form of subsidy payment and price support was made in grain market.

In order for Scenarios 2 and 3 (accelerated technological progress) to be realized, the government has to invest in research anddevelopment (R&D) and irrigation more than before (Huang et al., 1999; Jin et al., 2010). It can be argued that Scenarios 2 and 3 arepromising, partly because public investments in Chinese agriculture are considered to yield a significant return (Alston, Pardey, &Smith, 1999; Fan, Qian, & Zhang, 2006; Huang & Rozelle, 1996), and partly because growth strategies based on public investments donot contradict China's WTO accession commitments (Zhu, 2004). However, Felloni et al. (2003) conclude that even the overalladoption of genetically-modified (GM) crops is unlikely to be sufficient in order for China to attain self-sufficiency. Jiang (2008) takesa similar stance, claiming that China should almost double the level of agricultural R&D in order to maintain food self-sufficiency.

5. Conclusions

Apart from the price support program, the most effective way for the government to secure food self-sufficiency is to commitstrongly to the determination of farmland areas, and on that basis, to enhance land productivity. However, in order for the protectedland not to be left fallow, a certain amount of resources, including farm labor and machinery, have to be mobilized for work on thefields. This paper specifies the production function in a separate Cobb–Douglas (SCD) form, in lieu of an orthodox CD form, by takinginto account the fact that the former is better able to capture agricultural production characteristics that can be broadly categorizedinto biochemical and machinery technologies.

Our empirical results reveal that the grain supply and factor demand functions based on the SCD form are better able toexplain the reality. In particular, the performance of a capital demand function derived from profit maximization subject to theSCD production function is quite satisfactory, whereas there is a large discrepancy between the estimated and actual values ofcapital input when the production function is specified in the CD form. These findings provide evidence in support of thehypothesis presented in this paper that a recent capital deepening in Chinese agriculture is caused by not so much a simple

20 Tian and Wan (2000) estimated that the rate of technical progress for China's grain production amounts to an annual rate of 0.84%, which is in accordancewith the estimate of this study. Using the cost data of China's major crops, Jin, Ma, Huang, Hu, and Rozelle (2010) estimated the cost function, and then computedthe rates of technological progress for the period 1985–2004. They find that the annual rate of progress in the rice sector is in excess of 2.0%, while that of wheat isequal to 0.1%. The rates of technological progress in the maize and soybean sectors are approximately equal to 1.0%.

85

86

87

88

89

90

91

92

93

94

95

2010 2012 2014 2016 2018 2020

OECD-FAO Model I (dlnp=1.8)

Model I (dlnp=4.8) Model I (dlnp=7.8)

Model II (dlnp=4.8)

Fig. 7. Forecast of grain self-sufficiency.


factor substitution, but rather by farmers' behavior in response to the government's directives aimed at securing arable land. It isimportant to note that such a technical choice exerts a significant impact on income. In particular, when the terms of trade inagriculture worsen, farmers might replace the decreasing labor with farm machinery so as to maintain sown areas, for whichfarmers then incur incremental capital costs, thereby adversely affecting farm income.

This study also determines to what degree China's policy goal of a 95% grain self-sufficiency rate will be achievable. Althoughcommodity and sectoral differences are ignored, and the demand-side prediction relies on an outside source, our simulation analysisreveals that reaching the national targetwould be quite challenging unless the terms of trade in agriculture improve substantially in favorof producers. The simulation in this paper demonstrates that implementingmore stringent controls on sown area conversionwould helpend the downward trend in the self-sufficiency rate observed for the past several years. This policy enhancement, however, would becarried out at the expense of famers who are planning to diversify agricultural production from staple cereals to high-value products. Amore accelerated technological progress would help change the deteriorating supply–demand balance for the better and ensure foodsecurity. This, however, definitely requires the government tomakemassive investments in R&D and irrigation. Viewed in this light, it istime for China's policy makers to seriously reconsider whether adhering to the policy goal of grain self-sufficiency is worth their effort.

Acknowledgements

The author thanks Xiaobo Zhang (Editor), and anonymous referees for their insightful comments on early drafts. Thanks arealso due to many participants who commented on this paper in an annual conference of Theoretical Economics and Agriculture,and workshops held in Kyoto University. Funding from the Japan Society for the Promotion of Science is acknowledged gratefully.

89

90

91

92

93

94

95

96

97

98

99

2010 2012 2014 2016 2018 2020

benchmark Scenario 1 Scenario 2 Scenario 3

Fig. 8. Scenario analysis. Note. Benchmark: d ln S=−0.4%, BC technological change is 0.8%. Scenario 1: d ln S=0.0%, BC technological change is 0.8%. Scenario 2: dln S=−0.4%, BC technological change is 1.6%. Scenario 3: d ln S=−0.8%, BC technological change is 1.6%.

Appendix TableDescriptive statistics of the panel data.

Mean SD

Grain output (100 million Yuan) Overall 91.5 71.0Between 70.3Within 15.9

Fertilizer (10,000 tons) Overall 95.7 80.5Between 78.9Within 21.1

Sown areas (1000 ha) Overall 3607.1 2672.9Between 2686.5Within 394.3

Labor (10,000 people) Overall 418.2 365.9Between 354.8Within 109.3

Farm machinery (10,000 kW) Overall 1235.2 1322.3Between 1211.4Within 572.2

Irrigation rate Overall 0.587 0.359Between 0.339Within 0.132

Natural calamity index Overall 0.513 0.141Between 0.054Within 0.131


References

Alston, J. M., Pardey, P. G., & Smith, V. H. (1999). Paying for agricultural productivity. Baltimore and London: Johns Hopkins University Press.Carter, C. A., Zhong, F., & Zhu, J. (2012). Advances in Chinese agriculture and its global implications. Applied Economic Perspectives and Policy, 34, 1–36.Chan, K. W., & Zhang, L. (1999). The hukou system and rural–urban migration in China: Processes and changes. China Quarterly, 160, 818–855.Chen, J. (2007). Rapid urbanization in China: A real challenge to soil protection and food security. Catena, 69, 1–15.Chen, C., & Duncan, R. (2008). Agriculture and food security in China: What effect WTO accession and regional trade arrangements? Canberra: Asia Pacific Press.Cook, S. (1999). Surplus labour and productivity in Chinese agriculture: Evidence from household survey data. Journal of Development Studies, 35, 16–44.Deng, X., Huang, J., Rozelle, S., & Uchida, E. (2006). Cultivated land conversion and potential agricultural productivity in China. Land Use Policy, 23, 372–384.Egaitsu, F. (1979). Fall in growth rate of the Japanese economy and consequential charges in the costs of farm production. Journal of Rural Economics, 51, 52–60 (in

Japanese).Evenson, R., & Kislev, Y. (1975). Agricultural research and productivity. New Haven and London: Yale University Press.Fan, C. C. (2008). China on the move: migration, the state, and the household. London and New York: Routledge.Fan, S., & Agcaoili-Sombilla, M. (1997). Why projections on China's future food supply and demand differ. The Australian Journal of Agricultural and Resource

Economics, 41, 169–190.Fan, S., Qian, K., & Zhang, X. (2006). China: An unfinished reform agenda. In P. G. Pardey, J. M. Alston, & R. R. Piggott (Eds.), Agricultural R&D in the Developing

World: Too Little, Too Late?. Washington, D.C.: International Food Policy Research Institute.FAOSTAT (2012). Food and Agriculture Organization.Feder, G., Lau, L. J., Lin, J. Y., & Luo, X. (1992). The determinants of farm investment and residential construction in post-reform China. Economic Development and

Cultural Change, 41, 1–26.Felloni, F., Gilbert, J., Wahl, T. I., & Wandschneider, P. (2003). Trade policy, biotechnology and grain self-sufficiency in China. Agricultural Economics, 28, 173–186.Hayami, Y., & Kawagoe, T. (1991). Farm mechanization, scale economies, and structural change. In Y. Hayami, & S. Yamada (Eds.), The Agricultural Development

of Japan. Tokyo: University of Tokyo Press.Hertel, T., & Zhai, F. (2006). Labor market distortions, rural–urban inequality and the opening of China's economy. Economic Modelling, 23, 76–109.Ho, S. P. S., & Lin, G. C. S. (2003). Emerging land markets in rural and urban China: Policies and practices. China Quarterly, 175, 681–707.Huang, J., Liu, Y., Martin, W., & Rozelle, S. (2009). Changes in trade and domestic distortions affecting China's agriculture. Food Policy, 34, 407–416.Huang, J., & Rozelle, S. (1996). Technological change: Rediscovering the engine of productivity growth in China's rural economy. Journal of Development Economics,

49, 337–369.Huang, J., & Rozelle, S. (2008). Agricultural development and policy before and after China's WTO accession. In C. Chen, & R. Duncan (Eds.), Agriculture and Food

Security in China: What Effect WTO Accession and Regional Trade Arrangements?. Canberra: Asia Pacific Press.Huang, J., Rozelle, S., & Rosegrant, M. W. (1999). China's food economy to the twenty-first century: supply, demand, and trade. Economic Development and Cultural

Change, 47, 737–766.Ito, J. (2010). Inter-regional difference of agricultural productivity in China: Distinction between biochemical and machinery technology. China Economic Review,

21, 394–410.Ji, Y., Yu, X., & Zhong, F. (2012). Machinery investment decision and off-farm employment in rural China. China Economic Review, 23, 71–80.Jiang, T. (2008). WTO accession and food security in China. In C. Chen, & R. Duncan (Eds.), Agriculture and Food Security in China: What Effect WTO Accession and

Regional Trade Arrangements?. Canberra: Asia Pacific Press.Jin, S., Ma, H., Huang, J., Hu, R., & Rozelle, S. (2010). Productivity, efficiency and technical change: measuring the performance of China's transforming agriculture.

Journal of Productivity Analysis, 33, 191–207.Kislev, Y., & Peterson, W. (1982). Prices, technology, and farm size. Journal of Political Economy, 90, 578–595.Lichtenberg, E., & Ding, C. (2008). Assessing farmland protection policy in China. Land Use Policy, 25, 59–68.Lin, J. Y. (1992). Rural reforms and agricultural growth in China. American Economic Review, 82, 34–51.Liu, Y., & Wang, X. (2005). Technological progress and Chinese agricultural growth in the 1990s. China Economic Review, 16, 419–440.Liu, Z., & Zhuang, J. (2000). Determinants of technical efficiency in post-collective Chinese agriculture: Evidence from farm-level data. Journal of Comparative

Economics, 28, 545–564.Martin, W. (2001). Implications of reform and WTO accession for China's agricultural policies. The Economics of Transition, 9, 717–742.


Meng, X., & Zhang, J. (2001). The two-tier labor market in urban China: occupational segregation and wage differentials between urban residents and ruralmigrants in Shanghai. Journal of Comparative Economics, 29, 485–504.

Mundlak, Y. (2005). Economic growth: lessons from two centuries of American agriculture. Journal of Economic Literature, 43, 989–1024.OECD-FAO (2011). OECD-FAO Agricultural Outlook 2011–2020. : Food and Agriculture Organization.Rosegrant, M., Ringler, C., Msangi, S., Sulser, T. B., Zhu, T., & Cline, S. A. (2008). International Model for Policy Analysis of Agricultural Commodities and Trade

(IMPACT). Washington, D.C.: International Food Policy Research Institute.Tian, W., & Wan, G. (2000). Technical efficiency and its determinants in China's grain production. Journal of Productivity Analysis, 13, 159–174.Van den Berg, M. M., Hengsdijk, H., Wolf, J., Van Ittersum, M. K., Guanghuo, W., & Roetter, R. P. (2007). The impact of increasing farm size and mechanization on

rural income and rice production in Zhejiang province, China. Agricultural System, 94, 841–850.Xi, A., & Chen, B. (2007). Farm machinery service survey in rural Shanxi province and agricultural mechanization. (mimeo).Yang, H., & Li, X. (2000). Cultivated land and food supply in China. Land Use Policy, 17, 73–88.Yang, J., Qiu, H., Huang, J., & Rozelle, S. (2008). Fighting global food price rises in the developing world: The response of China and its effect on domestic and world

markets. Agricultural Economics, 39, 453–464.Yang, Y., & Tyers, R. (1989). The economic costs of food self-sufficiency in China. World Development, 17, 237–253.Yang, D. T., & Zhou, H. (1999). Rural–urban disparity and sectoral labour allocation in China. Journal of Development Studies, 35, 105–133.Yao, S., Liu, Z., & Zhang, Z. (2001). Spatial differences of grain production efficiency in China, 1987–1992. Economics of Planning, 34, 139–157.You, L., Spoor, M., Ulimwengu, J., & Zhang, S. (2011). Land use change and environmental stress of wheat, rice and corn production in China. China Economic

Review, 22, 461–473.Zhang, B., & Carter, C. A. (1997). Reforms, the weather, and productivity growth in China's grain sector. American Journal of Agricultural Economics, 79, 1266–1277.Zhang, X., & Fan, S. (2001). Estimating crop-specific production technologies in Chinese agriculture: A generalized maximum entropy approach. American Journal

of Agricultural Economics, 83, 378–388.Zhang, X., Mount, T. D., & Boisvert, R. N. (2001). The demand for food grain in China: New insights into a controversy. Agricultural and Resource Economics Review,

30, 1–9.Zhu, J. (2004). Public investment and China's long-term food security under WTO. Food Policy, 29, 99–111.Zhuang, R., & Abbott, P. (2007). Price elasticities of key agricultural commodities in China. China Economic Review, 18, 155–169.

capital deepening, land use policy, and self-sufficiency in china's grain sector

Documents