Economies of scale and mechanization in Chinese corn and wheat production

ABSTRACT

As the growth in Chinese farm productivity slowed down between 2000 and 2010, modernizing agriculture has become a priority of the Chinese government. Given the important role of mechanization and land reform policies in that context, this study investigates farm production in China with a specific emphasis on the potential role of mechanization as well as land and farm consolidation. A production function is estimated using farm household data on corn and wheat production in the Shandong and Hebei provinces. The results allowed us to explore the potential economies of scale across a range of farm size, the impact of land fragmentation, and assess the impact of machinery usage. Our findings suggest that, taken in isolation, the prospect for efficiency gains from mechanization and land reforms appears limited.

KEYWORDS:

• Chinese agriculture
• economies of scale
• agricultural mechanization
• production function

JEL CLASSIFICATION:

• Q12

Supplementary Material

Supplemented data of this article can be accessed here.

Notes

¹ The exchange rate between yuan and U.S. dollars averaged about 7.73 during the years between 2005 and 2007.

² We do not pursue this approach for lack of a proper state variable that could reflect unobserved heterogeneity. In China, land is not bought and sold on open market making it dubious whether landholding has any relationship with the unobserved productivity of the farm household. Moreover, most machinery or capital is rented such that information on past capital investment is not available.

³ Given the panel nature of the data, the independence assumption of error terms becomes more difficult to uphold in the pooled model. The potential correlation across observations from the same household provides a further reason, in addition to input variable endogeneity, to consider this model as being inferior to other specifications. This limitation is reflected in the interpretation of results and the conclusions are based as much as possible on the results of more appropriate specifications. The model is nevertheless estimated for sake of comparison.

⁴ The basic unit of land area in China is mu. 1 acre ≈ 6 mu.

⁵ We have estimated these models assuming the technical efficiency terms are time- or crop-invariant but results do not change qualitatively.

⁶ Following O’Donnell and Coelli (2005), we used results from previous literature to guide our choice. In particular, we used the results of Brümmer, Glauben, and Lu (2006) and Yu et al. (2013) which reported technical efficiencies near 0.9.

⁷ ${\rho }_{\mathrm{j},\mathrm{i},\mathrm{t}}=\left\{\begin{array}{c}1{\mathrm{i}\mathrm{f}\mathrm{x}}_{\mathrm{j},\mathrm{i},\mathrm{t}}>0\\ 0\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{s}\end{array}\right.$

⁸ Jins are the basic unit of crop outputs in China. 1 jin = 0.5 kg.

⁹ Market rental prices of the three types of machinery at the time of the survey were used as the basis to calculate machinery expenditures. The respondents were asked how much time each type of machinery was used for each crop. Expense on machinery was then calculated as rental prices times usage time for each machinery category and crop.

¹⁰ Training sessions include the skill training of farm management, such as fertilizer and pesticide use, machinery operation.

¹¹ The Bayes factor is defined as $\frac{P\left(y|{\theta }_0,h_0\right)}{P\left(y|{\theta }_1,h_1\right)}$ where $P\left(y|\theta ,h\right)\propto h^{NT/2}{e}^{-\frac{h}{2}{\left[y-W\theta \right]}^{\prime }\left[y-W\theta \right]}$is the likelihood function (see O’Donnell and Coelli 2005), $W=\left\{X,Z,D\right\}$, and $\theta =\left\{\beta ,\delta ,\mathrm{\partial }\right\}$ is the set of parameters obtained under two different.

¹² MCMC is defined as Markov chain Monte Carlo.

¹³ The production gain is computed as the percentage change in total production due to land consolidation $\mathrm{d}\mathrm{Y}=100\cdot \frac{{\mathrm{y}}_1^{\mathrm{P}}-{\mathrm{y}}_0^{\mathrm{P}}}{{\mathrm{y}}_0^{\mathrm{P}}}$, where ${\text{y}}_0^{\text{P}}={\displaystyle {\sum }_{\text{i}}^{\text{N}}{\text{y}}_{0,\text{i}}^{\text{P}}}$ is total production obtained under current land fragmentation conditions and ${\text{y}}_1^{\text{P}}={\displaystyle {\sum }_{\text{i}}^{\text{N}}{\text{y}}_{1,\text{i}}^{\text{P}}}$ assumes full consolidation. The posterior predictive density for this measure is defined as ${\mathrm{f}}^{\mathrm{P}}\left(\mathrm{d}\mathrm{Y}\right)={\int }^{\mathrm{f}}\left(\mathrm{d}\mathrm{Y}|\theta \right)\mathrm{d}\theta$, where $\theta =\beta ,\delta ,\mu$.

¹⁴ The 10th and 90th percentile of the posterior are 1.26% and 3.52%.

¹⁵ The 10th and 90th percentile of the posterior are 1.70% and 11.49%.

¹⁶ For corn (${\mathrm{y}}_{\mathrm{c}}$) the average marginal value product of the input k is, $\mathrm{M}\mathrm{V}{\mathrm{P}}_{\mathrm{c},\mathrm{k}}=\frac{1}{\mathrm{N}}\frac{1}{\mathrm{T}}\sum _{\mathrm{i}}^{\mathrm{N}}\sum _{\mathrm{t}}^{\mathrm{T}}\frac{\mathrm{\partial }\mathrm{l}\mathrm{n}\left({\mathrm{y}}_{\mathrm{c},\mathrm{i},\mathrm{t}}\right)}{\mathrm{\partial }\mathrm{l}\mathrm{n}\left({\mathrm{x}}_{\mathrm{k},\mathrm{i},\mathrm{t}}\right)}\cdot \frac{{\mathrm{y}}_{\mathrm{i},\mathrm{t}}{\mathrm{P}}_{\mathrm{y},\mathrm{t}}}{{\mathrm{x}}_{\mathrm{k},\mathrm{i},\mathrm{t}}}$, and similarly for wheat. The posterior can be defined as $f^P\left(MV{P}_{c,k}\right)={\int }^f\left(MV{P}_{c,k}|\theta \right)d\theta$, where $\theta =\beta ,\delta ,D$ for the fixed effects models.

¹⁷ Results presented in Table 5 are from the ’joint’ model. Results from frontier models and fixed effects models were similar in nature and are available upon request from the first author.

Additional information

Funding

This work was supported by the China Agricultural University’s Central University Basic Research Special Funds [2010JS021].“The Effect of Rural Labor Price Change on Demand to Large Machinery for Rural Households”.