\input{anform1.tex} \begin{document} \bc {\large \bf Resource Diversion, Nutrition, and the Chinese Agricultural Collapse of 1959-61}\\[9mm] {\bf Mark Yuying An}, {\bf Wei Li} and {\bf Dennis Tao Yang}\\[3mm] Duke University\\[15mm] A handout prepared for presentation in the 1999 American\\ Economists Association meetings in New York\\[15mm] October 1999\\[45mm] We are grateful to Anne Zhao and seminar participants at University of Toronto for helpful comments. \ec \newpage %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \bc {\large \bf Resource Diversion, Nutrition, and the Chinese Agricultural Collapse of 1959-61} \ec \section{Introduction} China's agricultural crisis of 1959-61 has received considerable attention among economists because of the enormous losses in output and in human lives that were associated with the catastrophe. \footnote{Note the two symposium issues in Journal of Comparative Economics and China Economic Review} According to statistical information that became available only recently, the national grain output fell precipitously by 15 and 16 percent for two consecutive years and then stayed at the same level for another year. These sharp declines in grain availability contributed directly to the worst famine in the world history of an estimated 23 to 33 millions of excess deaths (Ashton et al. 1984 and Peng 1987). \footnote{Other factors also contributed to the severity of the famine, including excessive procurement, urban-biased policies in the allocation of grain consumption, the continued export of grain while people starved, and waste of food in the communal dining halls. See, for instance, Lin and Yang (1998, 1999) for analyses on the famine causes and the relationship between grain shortfalls and the consequent starvation.} Finding the causes of the output collapse are important because it would provide insights into the operations of the centrally planned systems at the peak of its euphoria and would offer implications for preventing future disasters. Investigations into the causes of the output collapse have been largely inspired by the seminar work of Lin (1990) who argued the removal of exit rights from the collectives at the inception of the Great Leap Forward in 1958 was the primary cause for the collapse. He illustrated using primarily the nationwide total factor productivity estimates for the Chinese agriculture that his hypothesis is consistent with empirical evidence. The empirical evidence refutes other competing hypotheses, including three consecutive years of bad weather, bad policies and bad management during the Great Leap period, and low incentives in the large-size communes. Lin's provocative thesis received immediate criticisms. Subsequent research has focused on two sets of issues: (1) did the removal of exit rights actually occur in the fall of 1958? and (2) at the theoretical level, would the removal of exit right result in low productivity? Both Lin's and subsequent research have pointed to possible causes of the output shortfall, but they have not accessed the quantitative importance of the proposed factors. An important reason is that data needed to measure the key variables are either problematic or absent in the publicly available data sources. In this paper, we conduct the first econometric study on the causes of the collapse. Our main contributions are: (1) We have collected a data set that contains historical conditions of agricultural production at the Chinese provinces; (2) We advance and examine systematically two new hypotheses on the causes of agricultural collapse. We argue that (a) resource diversion (input reductions in agriculture, and quantity and quality diversion of inputs away from grain production), and (b) nutrition deficiency of agricultural workers due to sharp reduction in food availability during the famine are important causes of the grain output collapse; (3) We formulate an econometric framework that is amenable to assess the relative importance of the proposed causes and other existing explanations. Decomposition shed light on their quantitative importance. More features: studies the determination of total grain output rather than total factor productivity as seen in the literature. Findings: (1) Based on provincial level data, our empirical results indicate that nutrition deficiency, the diversion of inputs from grain production, bad weather, and the elimination of exit rights are the most important causes of China's agricultural collapse; (2) the grain output figure released by China's State Statistical Bureau is an outlier in our statistical analysis. Our results indicate that the actual output is likely to be about 10 percent lower than the public released figure, a result consistent with historical records that over reporting of grain output was likely in 1958. General implications?? Organization of the paper. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{China's Great Leap Forward and Historical Data} See section II of Lin (1990) for detailed descriptions of agricultural collectivization process. See Riskin (1987) for descriptions of the Great Leap Forward. See CER (1998, #2): Johnson, Riskin on data issues. One or two paragraphs on the great Leap Forward: overall strategy; emphasis on rapid industrialization; production of steel. Rural resources to support industrialization in cities: rural labor movements to cities (Liang 1996; Roskin 1987); increased demand for agricultural products as raw materials - backyard furnace in rural regions; land devoted to cash crops, such as cotton and oil seeds. One or two paragraphs on agricultural collectivization movement in rural areas: from mutual- aid teams, to advanced cooperatives, and to communes in 1958. Rationale for collectivization: scale economy, mobility of labor for large projects: irrigation projects, dams, roads, etc. The sudden output collapse. Descriptions on grain output declines and the consequent famine. Major episode in China development. Turn to discuss data sources. The establishment of statistical system in China. Disturbances during the Great Leap Forward and the consequent Cultural Revolution. Despite interruptions of historical records, there are records on important agricultural variables. Some data are missing. Our data sources: "A Compilation of China's Historical Data at the Provincial Level." %% See Wen (1993) and other studies quoted in his paper on the performance of Chinese agricultural sector, as well as their data sources. Descriptions of our survey: purpose is to collect important information that influence agricultural performance but are missing in published data sources. Methods: rural survey teams of SSB, to each province, a team of at least three experts consist of agricultural experts and knowledgeable personnel from provincial statistical office, examination of historical records. Variables: exit rights in China's collectives; scale of a typical production unit; local weather conditions based on records of rain fall, temperature, information on hail and frost, and other special weather conditions. This survey contain information on both institutional and natural environments for agricultural production (see Data Appendix for details). Pooling data from published sources and from our survey together offers a unique set of data to examine China agricultural performance over time. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Causes of Grain Output Collapse} The following provides a quick sketch of the main explanations for the production collapse. Let's discusses these alternative hypotheses in light of available data. Table 1 presents both national and provincial level data on the main variables to illustrate our discussions. Analysis goes beyond stating the potential causes. We relate our discussions to the previous literature: background for each argument; various measurement issues; interactions among variables; propose specific proxies for the key variables. The econometric analysis to be presented in the following sections assesses the quantitative impact and the relative importance of the factors in causing the sharp output decline. \subsection{Bad Weather} This was the government's official explanation for the catastrophe. According to official weather records, the average percentage of sown areas hit by natural calamities during the period was 15.27 percent, which compares with an average of 7.63 percent disaster areas in the three years prior to the crisis (Disaster areas are defined as the sown acreage that are hit by flood, drought, frost, and hail, and have 30 percent or more reduction in yield compared to normal yield; MOA, 1984). Although bad weather may severely damage agricultural production, scholars have been cautious about the magnitude of the negative impact. For instance, the average percentage of disaster hit areas in 1962 and 1963 was 13.1 percent, similar to the crisis period, but the negative impact on grain supply was limited. While researchers generally agree that the natural calamities may be responsible for a fraction of the decline in grain production, the government emphasized that natural calamities were the main cause for the catastrophe. Weather variable as defined by China's statistical sources would be endogenous. Would introduce problems in econometric analysis. How our survey measure that variable, etc. \subsection{Size of Organization} Incentive problems due to the unwieldy size of the communes could result in lower production efficiency (e.g. Perkins and Yusuf 1984), and the lack of managerial abilities in operating large size organizations may also result in poor performance in communes and brigades that existed as basic production units during the Great Leap crisis. Other implications of larger scale: monitoring costs; residual claimants; scale economy, etc. Historical data and our survey on scale, measured by number of households per production unit. Discuss the changes. Shall we discuss the following point? Moreover, the initiation of local food self-sufficiency policies in 1958 might have caused the loss of regional comparative advantage resulting in a decline in aggregate food supply (Lardy 1983). While these institutional changes and new policies could have negatively affected food supply, their importance needs to be assessed empirically. \subsection{Exit Rights} In a seminal paper, Lin (1990) proposed a game theory hypothesis that the main cause of the agricultural collapse was the deprivation of the peasants' right to withdraw from the collectives with the communization starting in the fall of 1958. This switch in the form of organization changed the incentive structure for the peasants who chose to shirk within the communes because the mechanism of self-discipline breaks down with compulsory participation. Empirical studies by other researchers (Tang 1984 and Wen 1989) suggested a sudden drop in total factor productivity in 1959 and the index stayed at a low level for the entire collectivization period 1958-78, which give support to Lin's exit right hypothesis to explain low efficiency during the commune period. However, these productivity indices do not give direct assessments on the role of exit rights in causing aggregate output decline in the three years of the crisis. Lin's explanation for the abrupt collapse of the Chinese agriculture provoked a heated debate over the nature of incentives within agricultural collectives. The articles that appeared in the 1993 symposium issue of the {\it Journal of Comparative Economics} were, in effect, criticisms and comments on Lin's paper. The debate focuses on two critical issues. The first is a theoretical postulation that the right to exit is necessary for high effort-supply among cooperative members. This theoretical possibility suggested by Lin contrasts with a theory proposed by MacLeod (1993) in which the imposition of exit costs on individuals is considered as necessary to avoid the case that some members may shirk and then quit the cooperative. The second issue is on the voluntary principle of participation practiced during the collectivization movement before the establishment of communes. While the disputes were certainly not resolved by the symposium, the issues raised by Lin and his critics were clarified. There was convergence on the key issue that elimination of exit right caused a decline in the productivity of collective farms was both logically and historically defensible (Putterman and Skillman,1993). The relevant references: Lin 1990. What is the historical evidence? Kung (1993), Kung and Putterman (1997), the JCE issue. Key question: when did the collectives start compulsory participation? Theoretical implications of this action? Evidence from documents versus our survey? We stress provincial variations in policies and implementations. To our knowledge, hypotheses (1) - (3) have been discussed extensively at the theoretical level. But researchers have not yet assessed the quantitative importance of the factors in causing aggregate grain output decline during the crisis of 1959-61. \subsection{Resource Diversions} We argue that two sets of factors could have contributed to the shortage of grain production during the Great Leap crisis. First, the diversion of productive inputs away from agriculture may have undermined the capacity to produce food, and secondly, nutrition deficiency due to limited calarie intake during the famine may have dramatically aggravated the situation. We systematically investigate the role of these two factors in causing the production collapse. \subsubsection{Reduction in Sown Areas for Grain} Key variables are: total rural labor force, total steel output, food availability. Radical government polices resulted in severe reductions in the grain-sown areas. In 1958, there were signs of a bumper harvest, and the government started to accept outrageously high estimates of grain production, a part of the "wind of exaggeration." With the delusion that China had solved its grain problem, Mao personally initiated a "three-three system" of agricultural land utilization, in which grain would only occupy one-third of the sown area. Another official policy was "sow less, harvest more." The implementation of these policies in 1959 caused a sharp reduction of nearly 10 percent in sown areas for grain production, which most likely contributed to the output decline. \subsubsection{Reduction in Agricultural Labor} There were also massive outflow of labor away from agriculture during the crisis period. For the Great Leap industrialization, about 41 million workers exited agriculture between 1957 and 1958, which represented a 21 percent decline (Riskin 1987). Among these workers, approximately 17 million worked in the iron, steel and other heavy industrial undertakings in the countryside, while close to 16 million migrated into cities working in state industrial enterprises. In response to the production decline and the famine in 1959 and 1960, the government sent 10 million workers back to their rural homes in 1961 to release the pressure of urban food demand and to increase labor inputs for agricultural production. However, the massive exodus of labor must have undermined the capacity of grain production during the whole crisis period. \subsubsection{Resource Diversion within Agriculture} To support industrialization, the production of industrial raw materiels (cotton was a prominent example) increased dramatically during the Great Leap, which drew resources away from grain production and consequently may cause output reduction. A set of bad policies to support industrialization were also likely to reduce the capacity of labor both in quality (high capacity labor to cities) and quantity (backyard furnace within rural areas) for grain production. However, the official statistics collected and released by China's State Statistical Bureau, the only data source for studying long term changes in agricultural production, only record the utilization of labor and capital for the agriculture as a whole. Therefore, these data do not reflect quality and quantity diversion of productive inputs. To study output collapse, these data need adjustments. Land area for cotton production above trend could be used as a proxy for resource diversion away from grain production within agriculture. \subsection{Nutrition Deficiency} Empirical studies in development economics have found that nutrition and calory intake of agricultural workers significantly impact labor productivity (Strauss). During the crisis of 1959-61, food availability to rural population was sharply reduced. Reports of weak agricultural workers who could merely walked in the fields. The lack of physical strength due to nutrition deficiency may significantly reduce the work capacity of grain producers. We shall use a food availability measure and death rates of local regions to approximate the degree of nutrition deficiency. \subsectioin{Measurement of Output} The initial grain estimate by the State Statistical Bureau was 375 million tons, and it was reduced to 360 million tons in December 1958 and, then, to 250 million tons in August 1959. The most current source indicates the grain output to be 200 million tons. Output number is very likely incorrect: political incentives to fulfill the production forecast (quota), etc. We shall perform test on outliers to see whether the 1958 output is reasonably reliable. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Empirical Framework} \subsection{Production function} Grain output is determined not only by factor inputs, but also by the institutional environment. Given effective labor, $L^*_{it}$, land sow to grain, $A_{it}$, a vector of other inputs, $X_{it}$, and a vector of institutional environment, $R_{it}$, grain output in province $i$ in year $t$ can be written as \begin{equation} Q_{it} = F_i(L^*_{it}, A_{it}, X_{it}, R_{it})\xi_{it} \end{equation} where $F_i$ is a province-specific production function, and $\xi_{it}$ is a random productivity shock. The effective labor, $L^*_{it}$, is measured in efficiency units, therefore incorporates the supply of effort. This formulation allows us to estimate the impact of institutional changes on labor productivity in addition to total factor productivity. Following the convention in productivity estimation, other factor inputs are measured in physical units that are available for grain production. % Less active or efficient use of the other factors would thus result in % a reduction in total factor productivity. In order to derive an estimable specification, we adopt the commonly used Cobb-Douglas functional form and rewrite the production function as \begin{equation} \ln Q_{it} = \alpha_{L} \ln L_{it}^* + \alpha_A \ln A_{it} + \alpha_X' \ln X_{it} + \beta_R' R_{it} + \gamma W_{it} + u_i + v_t + \epsilon_{it} \label{eq:CD.1} \end{equation} where $Q_{it}$ is Province $i$'s grain output in in year $t$, $L^*_{it}$ is effective labor used in grain production, $A_{it}$ is effective land area sow to grain, $X_{it}$ is a vector of other inputs ({\it e.g}, fertilizers, machine power, and irrigation), $R_{it}$ is a vector of institutional variables that affect productivity, $W_{it}$ is the proportion of cultivated land affected by bad weather. $u_i$ measures province specific shocks $v_t$ measures time specific shocks $\epsilon_{it}$ measures idiosyncratic shocks. As it stands, (\ref{eq:CD.1}) is not estimable since effective labor, $L*_{it}$, is not observed. What is observed is the aggregate number of agricultural workers, $L_{it}$. To make it estimable, one must specify a model for estimating the effective labor using available data. In this paper, we use a linear specification so that the effective labor used in grain production (in logarithm) is a proportion of the aggregate agricultural labor force (in logarithm), or \begin{equation} \ln L^*_{it} = (1 + \delta'_a a_{it} + \delta'_s s_{it} + \delta'_n n_{it} + \delta'_e e_{nt} ) \ln L_{it} \label{eq:diversion} \end{equation} where $a_{it}$ is a column vector of the proportions of the cultivated area used for other crops. {\it E.g.}, proportion of land sown to cotton. This vector measures the substitution of labor from grain to other crops. We expect the coefficients $\delta_a$ to be negative so an increase land allocated to other crops reduces both the quantity and quality of labor allocated to grain production. $s_{it}$ is a column vector that measures the diversion to non-agricultural activities of the labor that would have been allocated to grain production. Examples of such diversions are numerous. There was the infamous ``backyard iron smelting campaign'' during the ``Great Leap Forward'' era, when the strong and skilled agricultural laborers were turned into backyard iron workers. It is also known that many agricultural (as well as industrial and office) workers were ``volunteered'' to white-elephant projects. Statistics of such massive diversion of manpower, while available on national level, are not available on provincial level. In the estimation, we use a period dummy, GLF, which equals 1 between 1959--1961 and 0 in other years. $n_{it}$ is a column vector of variables that provide some measures of nutrition received by agricultural workers. In this study we use food availability and above-trend death rate as proxies for nutrition deficiency that affects the quality (or productivity) of labor. $e_{it}$ is a column vector of variables that provide some measures of the institutional changes that affect the supply of labor effort. These variables are intended to capture the impact the institutional changes on labor productivity. In principle, other factor inputs can be treated similarly to capture the effects of resource diversion and institutional changes on both their availability and their effective usage. Substituting (\ref{eq:diversion}) into (\ref{eq:CD.1}), we obtain our estimation equation: \begin{eqnarray} \ln Q_{it} & = & (\alpha_{L} + \alpha'_{La} a_{it} + \alpha'_{Ls} s_{it} + \alpha'_{Ln} n_{it} + \delta'_e e_{nt}) \ln L_{it} \nonumber \\ & \ \ & + \alpha_A \ln A_{it} + \alpha_X' \ln X_{it} + \beta_R' R_{it} + \gamma W_{it} + u_i + v_t + \epsilon_{it} \label{eq:CD.2} \end{eqnarray} where $\alpha'_{La} = \alpha_{L} \delta'_a$, $\alpha'_{Ls} = \alpha_{L} \delta'_s$, $\alpha'_{Ln} = \alpha_{L} \delta'_n$, are row vectors of coefficients. The production function is estimated using fixed-effects technique. Equation~(\ref{eq:CD.2}) is estimated using fixed-effects panel data technique. \subsection{Hypothesis Testing} \subsection{Decomposing the Causes of Output Collapse} Let $y_{it}=f(X_{it};\beta)$ be our production function (period t and province i), where $X_{it}=(x^1_{it}, ^E, x^J_{it})$ is the J-vector of function arguments Let $y_{it}=f(X_{it};\beta)$ be our production function (period t and province i), where $X_{it}=(x^1_{it}, ^E, x^J_{it})$ is the J-vector of function arguments (inputs as well as variables affecting effective inputs), $\beta$ is the parameter vector in the model. Let b be the estimates the model parameters $\beta$. If we want decompose the reduction in y from year s to year t, (s could be $t-1$ a la Dennis, or a fixed base year (1957) a la Wei) here is what we do: Let $\hat{y}_{is} = f(X_{is}; b)$, the predicted base year value. Let \begin{equation} \label{yj} \tilde{y}^j_i = f(x^1_{is}), x^2_{is}, ..., x^{j-1}_{is}, x^j_{it}, x^{j+1}_{is}...x^J_{is}; b), \end{equation} be the hypothetical value of $y$ which is different from $\hat{y}_{is}$ only in the j-th argument. Then the reduction in y caused by changes in $x^j$ will be simply $\tilde{y}^j_i - \hat{y}_{is}$. The total effect attributable to $x^j$ is the sum over all the provinces in the sample. This approach can be modified to do that of sequential \be \[\bar{y}^j_i = f(x^1_{it}), x^2_{it}, ..., x^{j-1}_{it}, x^j_{it}, x^{j+1}_{is}...x^J_{is}; b) \] \ee and the contribution due to change in $x^j$ is simply $\bar{y}^j_I -\bar{y}^{j-1}_I$. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Data and Estimation Results} \subsection{Data} Data used for empirical analysis are provincial level data for Chinese agricultural production and other economic variables covering the period 1952-91. Some of the data are from published sources, such as {\it Statistical Yearbook of China}, and others are from a joint data collection with China's State Statistical Bureau. We are still gathering additional variables to complement the existing data at hand. \subsection{Estimation Results} Our empirical analysis is still largely incomplete. However, Table 3 contains some key results of our fixed-effect estimation of the grain production. This table indicates: (a) Sown area and irrigation contribute positively to grain output; (b) Labor is a largest share of contribution in the Cobb-Douglas specification; (c) Estimates for the two proxies of resource diversion (cotton and GLF) indicate that quantity and quality reductions in the labor force for grain production had significant and negative impacts on output; (d) Nutrition factors are also important determinants of grain output; (e) The exit right has a significantly impact on output through its effects on labor effort during the commune period; and (f) Natural calamities significantly reduce grain output. \subsection{Decomposing the Contributions to Output Decline} Table 4 presents the decomposition results based on the methods outlined in section 3.3. Our findings paints the following picture. Among the factors that contributed to the total grain output decline of about 68 million tons between 1958-60: (a) Nutrition factors account for about 2/5 of the total; (b) Resource diversions account for about 3/10; (c) Exit right accounts for about 1/5; (d) Inputs reductions account for 1/10 ; (e) Bad weather accounts for less than 1 percent of the total declines. We plan to add additional data, improve the econometric strategy to deal with a number of complexities, and conduct extensive sensitivity analyses. By then, we will report more robust empirical findings %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Conclusions} Summary of the major aspects of our project. The dynamics of the crisis: problems of collectivization in 1956 and 1957 due to the removal of exit right, but output continued to grow because of increases in modern inputs and improved nutrition. Output increases brewed optimism, combined with the initiation of the Great Leap Forward, several things took place in 1957-58: higher procurements - undermines nutrition for 1958, various forms of resources diversions in sown area, labor and steel production. These factors should have reduced grain output in 1958, but weather was great, combined with increases in modern inputs, they largely offset other negative impacts, including the heavy procurement in 1957. The good harvest (or the exaggerated grain output in 1959) and the increased urban demand for food further pushed up the target for 1959 procurement. 1959 was the first year of catastrophe: dramatic turn around of the climate, nutrition deficiency, continued resource diversions were the main causes. 1960 the second year: nutrition was a lingering factor, bad weather, and steel production. The large negative effects of nutrition deficiency continued the dynamic effects until 1961. All negative policies stopped in 1961, leading to the first yeaer of recovery. Largely man-made mistakes. Results devastating. General implications. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \clearpage \section*{References (incomplete)} \bd \item Ashton, B., Hill, K., Piazza, A., & Zeitz, R. 1984. Famine in China, 1958- 61. {\em Population and Development Review}, 10: 613-45. \item Dong, X. Y. & Dow, G. K. 1993. Does Free Exit Reduce Shirking in Production Teams? {\em Journal of Comparative Economics}, 17: 472-484. \item Johnson, D. G. 1990. {\em The People's Republic China: 1978-1990.} San Francisco, California: ICS Press. \item Kung, J. K. 1993. Transaction Costs and Peasants' Choice of Institutions: Did the Right to Exit Really Solve the Free Rider Problem in Chinese Collective Agriculture? {\em Journal of Comparative Economics}, 17: 485-503. \item Liang, Zai and White, Michael. 1996. Internal Migration in China, 1950-1988. {\em Demography}, 33(2): 375-384. \item Lin, J. Y. 1990. Collectivization and China's Agricultural Crisis in 1959-61. {\em Journal of Political Economy}, 98(6):1228-52. \item Lin, J. Y. and Yang, D. T. 1998. On the Causes of China's Agricultural Crisis and the Great Leap Famine. {\em China Economic Review}, 9(2): 125-140. \item Lin, J. Y. & Yang, D. T. 1999. Food Availability, Entitlements, and the Chinese Famine of 1959-61. {\em Economic Journal}, October. \item Liu, M. Q. 1993. Exit Right, Retaliatory Shirking, and the Agricultural Crisis in China. {\em Journal of Comparative Economics}, 17: 540-559. \item MacLeod, W. B. 1993. The Role of Exit Costs in the Theory of Cooperative Teams: A Theoretical Perspective. {\em Journal of Comparative Economics}, 17: 540-559. \item Peng, X. Z. 1987. Demographic Consequences of the Great Leap Forward in China's Provinces. {\em Population and Development Review}, 13(4): 639-670. \item Perkins, D. H. 1966. {\em Market Control and Planning in Communist China}. Cambridge, MA: Harvard University Press. \item Putterman, L. & Skillman, G. L. 1993. Collectivization and China's Agricultural Crisis. {\em Journal of Comparative Economics}, 17: 530-539. \item Riskin, C. 1987. {\em China's Political Economy: the Quest for Development Since 1949}. Oxford, U.K.: Oxford University Press. \item SSB, State Statistical Bureau. 1993. {\em A Compilation of Historical Statistics for China's Provinces and Municipalities, 1949-89}. ZhengZhou: China Statistical Publishing House. \item Wen, Guanzhong James. 1993. "Total Factor Productivity Change in China's Farming Sector: 1952-1989." {\em Economic Development and Cultural Change}. P1-41. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \end{document}