An empirical analysis was conducted on the behavior of Japanese rice producers from the standpoint of efficiency in production using the panel data of the Rice Production Cost Statistics by the Ministry of Agriculture, Forestry and Fisheries. The stochastic frontier production function was estimated and the inefficiency indices of production were calculated. Based on this information, the efficient and inefficient rice producers were identified, and the factor demand behavior and characteristics of the land use for rice production were compared. We find that the production-efficient certified farmers lowered the land utilization rate of paddy for rice production greatly.
This is a preview of subscription content, access via your institution.
Buy single article
Instant access to the full article PDF.
Price excludes VAT (USA)
Tax calculation will be finalised during checkout.
For example, there are direct subsidies paid to farmers for utilization of paddy fields for producing crops such as grains, soybeans, rice for feed, rice for rice flour, etc.
Note that dry fields (hatachi) and pasture are not included in defining the percent area planting rice.
The purpose of production adjustment policy of rice is to prevent the fall of rice price and secure income for rice farmers when the rice consumption has decreased relative to rice production. The policy measure was initially an allocation of set-aside area to each prefecture, but it was switched in 2004 to an allocation of production target quotas of rice to each prefecture. This production adjustment policy was abolished in 2018.
Appendix Table 11 shows the number of observations of rice producers by prefecture.
Pitt and Lee (1981) is an empirical study of the stochastic frontier production function that assumes a half-normal distribution for the probability distribution of inefficiency. Battese and Coelli (1988) generalized the probability distribution function for inefficiency to a truncated normal distribution and estimated the stochastic frontier production function.
See Schmidt and Sickles (1984) for the fixed-effect model of inefficiency.
When the model with dummy variables corresponding to the individual prefectures was estimated, convergence was not attained. Therefore, we use the regional dummies instead.
Unobservable soil quality or shocks of pest insect might affect the production of rice as well as the quantity of factor inputs, which leads to endogeneity problem in estimating production function or factor demand function. Considering endogeneity in estimating the stochastic frontier production function is an important agenda for the future research.
The obtained elasticity estimates are close to the values in Saito et al. (2010), which estimated the Cobb–Douglas production function with the microdata of the Agriculture and Forestry Census. The estimated elasticities of labor, capital stock, and land obtained by them are 0.0523–0.0678, 0.0214–0.0291, and 0.9571–1.0556, respectively. In their estimation, the amount of materials input has not been controlled as an explanatory variable.
When the null hypothesis of constant returns to scale was tested by a Wald statistics, it was rejected at the 1% level.
The p value of the null hypothesis that the correlation coefficient is zero is 0.00 for all cases.
The analysis in the subsequent sections is almost entirely unaffected even if we use the inefficiency indices that are obtained under the assumption of the Cobb–Douglas production function and truncated normal distribution.
Income is defined as follows: gross agricultural income + family labor cost − agricultural expenditures (all expenses necessary for farming) − interest payment − paid rent for land.
Kawasaki (2010) has shown that a large number of parcels lead to inefficiency in rice production by employing the panel data of rice production cost statistics.
This description about certified farmers is taken from Ministry of Agriculture, Forestry and Fisheries (2018).
One might argue that the calculated inefficiency measure might be regressed on the 17 items of farmer’s attributes as explanatory variables. I do not take this regression approach since we cannot exploit the time-varying information of farmer’s attributes fully in regressing time-invariant inefficiency measure on the farmer’s attributes. In fact we fail to obtain significant coefficient estimates of many explanatory variables.
Some may argue that efficient rice producers have increased production efficiency by producing rice of lower quality. We cannot deny this possibility since the price per kg of the rice harvested by inefficient producers is significantly higher than that by efficient producers. We need more detailed information on the rice cultivar of individual farmers for further discussions on this issue.
This condition is satisfied in the Cobb–Douglas production function.
In calculating the real factor prices, we use the output price in the previous year since the output price of the current year is not available in making decision of the current factor inputs. See the “Data appendix” for the procedure to construct the factor prices.
The ratio of the amount received from agricultural mutual aid to the amount contributed to agricultural mutual aid (AGRIAID) was added as an explanatory variable of factor demand function, but it was statistically insignificant.
The long run wage elasticity is calculated as the short run elasticity divided by 1 − γ5N.
All other coefficient estimates are not shown to save space. Detailed estimation results of wage equation are available from the author upon request.
For example, the imputed wage rates of family labor in Hokkaido, which has the largest observations, are 1719 yen, 1547 yen, 1375 yen and 1203 yen for the age categories below 65, between 65 and 70, between 70 and 75 and over 75, respectively, in 2013. The difference in the wage rate between the age categories below 65 and between 65 and 70 is 172 yen, which is close to the wage difference obtained from the regression analysis.
There is a caveat in interpreting the relationship between productivity of rice production and percent area planting rice for production-efficient producers. Under economies of scope, an increase in productivity of rice production for production-efficient producers might result from growing rice in relatively large-scale plots and at the same time other crops in small plots. If this is the case, then specialization in rice production by efficient rice producers might not necessarily increase productivity of rice production. It is an important research agenda to investigate whether economies of scope are working in rice production in Japan.
Aigner, D., Lovell, K., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier production function models. Journal of Econometrics, 6, 21–37.
Battese, G., & Coelli, T. (1988). Prediction of firm-level technical efficiencies with a generalized frontier production function and panel data. Journal of Econometrics, 38, 387–399.
Battese, G., & Coelli, T. (1992). Frontier production functions, technical efficiency and panel data: With application to paddy farmers in India. Journal of Productivity Analysis, 3, 153–169.
Hayami, Y., & Kawagoe, T. (1989). Farm mechanization, scale economies and polarization: The Japanese experience. Journal of Development Economics, 31, 221–239.
Jondrow, J., Lovell, K., Materov, I., & Schmidt, P. (1982). On the estimation of technical inefficiency in the stochastic frontier production function model. Journal of Econometrics, 19, 233–238.
Kako, T. (1979). Inasaku niokeru Kibo no Keisoku (Estimation of scale economies in rice production). The Economic Studies Quarterly, 30(2), 160–171. (in Japanese).
Kawasaki, K. (2010). The costs and benefits of land fragmentation of rice farms in Japan. The Australian Journal of Agricultural and Resource Economics, 54, 509–526.
Meeusen, W., & van den Broeck, J. (1977). Efficiency estimation from Cobb-Douglas production functions with composed error. International Economic Review, 18, 435–444.
Ministry of Agriculture, Forestry and Fisheries. (2018). FY2017 summary of the annual report on food, agriculture and rural areas in Japan.
Pitt, M., & Lee, L. (1981). The measurement and sources of technical inefficiency in the Indonesian weaving industry. Journal of Development Economics, 9, 43–64.
Saito, K., Ohashi, H., & Nishimura, K. (2010). Syuraku Eino ga Inasaku no Seisan oyobi Hiyo ni Ataeru Eikyo–Daikibo Inasaku Keiei no Simulation Bunseki (The effects of community farming on production and cost of rice cultivation—a simulation analysis of large-scale rice farming), RIETI discussion paper series 10-J-009 (in Japanese).
Saito, K., & Ohashi, H. (2015). Inasaku Seisan Chosei ni kansuru Simulation Bunseki–Tensaku oyobi Solar Sharing nikansuru Seisaku Koka (A simulation analysis on rice production adjustment: Policy effects on switching crops and solar sharing), RIETI discussion paper series 15-J-055 (in Japanese).
Schmidt, P., & Sickles, R. C. (1984). Production frontiers and panel data. Journal of Business and Economic Statistics, 2, 367–374.
This study was conducted as a part of the Project “Study on Corporate Finance and Firm Dynamics” undertaken at Research Institute of Economy, Trade and Industry (RIETI). This study utilizes the micro data of “Rice Production Cost Statistics” which is conducted by the Ministry of Agriculture, Forestry and Fisheries. This paper was presented at Hitotsubashi-RIETI International Workshop on Real Estate Market, Productivity, and Prices held at Research Institute of Economy, Trade and Industry in Tokyo, Japan on October 13–14, 2016, Discussion Paper seminar at RIETI and Economic Development Workshop/CEI Seminar at Hitotsubashi University. The author is grateful for extremely helpful comments and suggestions by two anonymous referees, co-editor of the Japanese Economic Review, Yutaka Arimoto, Ian Coxhead, Takashi Kurosaki, Hiroshi Ohashi, Iichiro Uesugi and the participants at the workshops and seminar. This research was financially supported by KAKENHI Grant-in-Aid for Scientific Research (S) #15H05728, (B) #16H03604 and the program of the Joint Usage/Research Center for “Behavioral Economics” at ISER, Osaka University.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
In this appendix, we describe the sources and the methods of constructing the variables used in this study. The data are mainly from the Rice Production Cost Statistics (Kome Seisanhi Chosa Tokei) by the Ministry of Agriculture, Forestry and Fisheries.
Y: output is the quantity of rice that was produced as the staple product (syu sanbutsu) for sale and home use (unit: kg).
N: labor input is the working time spent on paddy rice production (inasaku futanbun). It includes both family labor and hired labor used for paddy rice production. The working hours of family labor are the sum of those in age categories below 65, between 65 and 70, between 70 and 75 and above 75 (unit: h).
L: land is the planted area of rice (sakuduke menseki) (unit: a).
K: capital stock is calculated by deflating the nominal stock of buildings and structures, land improvement facilities (tochi kairyo setsubi), automobiles, agricultural machinery, and tools in the fixed capital by the corresponding price indices, and by summing up them. The 2015-price deflators corresponding to the respective items are buildings and materials, automobiles and related fees, and agricultural machinery and tools (comprehensive). The data source of price deflators is the Agricultural Price Index reported by the Ministry of Agriculture, Forestry and Fisheries.
M: materials were calculated by dividing the expenditure on five types of materials (seed and seedling costs, fertilizer costs, agricultural chemical costs, light, heat and power costs, and various other materials costs) by the deflators in the Agricultural Price Index corresponding to each of these five items, and summing up them.
p: output price was calculated by dividing the nominal sales of the rice produced as the staple product for sales and home use by the output quantity of rice.
w: wage rate was calculated as follows. First, the family labor cost is calculated by multiplying the working hours of family labor in the four age categories defined above by the respective hourly imputed wage rate. The hourly imputed wage rate is based on wage data for business establishments with 5–29 workers in the construction, manufacturing and transportation/postal industries in the Monthly Labor Survey by the Ministry of Health, Labor and Welfare. The hired labor cost is also calculated by multiplying the working hours of hired labor by the actual hourly wage rate. Then, total labor cost, defined as the sum of the family labor cost and the hired labor cost, is divided by total working hours of family labor and hired labor to obtain the wage rate.
pK: rental price of capital was calculated by summing up the land improvement and water conservancy fees; the rent and fees; the depreciation costs for buildings, automobiles, agricultural machinery, tools, and production management; paid interest; imputed interest on the farmers’ own capital and the self-supplied portion out of the building, automobile, agricultural machinery and tool costs; then, dividing the total sum by the capital stock.
pM: materials price was calculated by dividing the nominal expenditure on the five items of materials calculated above by the corresponding real amount.
Rights and permissions
About this article
Cite this article
Ogawa, K. Inefficiency in rice production and land use: a panel study of Japanese rice farmers. JER 71, 641–669 (2020). https://doi.org/10.1007/s42973-019-00015-w
- Stochastic frontier production function
- Factor demand
- Land use
- Rice production adjustment