Share

Policy research working paper 5472, agriculture and rural development team, development research group, the World Bank ,Washington, DC,
Butzer, Rita, Yair Mundlak, and Donald F. Larsen. 2010

A new dataset for capital stock which distinguishes between livestock, treestock, and fixed capital, highlights the changing composition of capital stock and finds that fixed capital accounts for most of the change in agricultural productivity.

In empirical studies of agricultural production, accounting for capital stock allows researchers to distinguish growth from applying more inputs, such as labour or seeds, from growth due to changes in input productivity. Growth in the agricultural sector and, subsequently, overall economic growth can then be better understood. However, measures of capital stock have been lacking. Proxies are been used in the literature, such as the Food and Agriculture Organization’s (FAO) use of tractors as an alternative to the value of capital stock. The authors argue that proxies are not truly indicative, which skews analysis of agricultural production.

A new data set is constructed for the value of capital stock across 30 developing and developed countries, over the period of 1970 to 2000. Capital stock is defined by three components. First is fixed capital, which uses data from national income accounts and is modelled in an inventory turnover framework. Livestock, defined as capital stock when used for “breeding and drafting power,” is measured using FAO data on livestock quantities and trade valuation. The third component is treestock, which is considered as land improvement and typically goes unaccounted for. The value is constructed by estimating the present value of expected future income, net of production costs which are estimated to be 80% of income.

The evolution of capital stock in agriculture can be studied using the new data set. Agricultural capital stock grew over the period 1970 to 2000, but the share of livestock in the total fell over the period. The changing composition of capital stock is particularly evident in middle-income and lower-income countries, characterized by rapid growth in treestock and relatively slow growth in fixed capital. With similar compositions of capital at the outset, by 2000 high-income countries count 54% of their stock as fixed capital, compared to 37% in lower income countries. In contrast, livestock accounted for 9% and 17% in higher and lower income countries, respectively.

Agricultural production is modelled as a function of inputs and state variables, and estimated per country over a 30-year time period. To highlight the distinction between measures of capital stock, the results are derived using both the new dataset and the FAO tractor measure. Estimating agricultural production within each country first, the new dataset finds that capital stock is positively associated with production, and that fixed capital accounts for most of the impact on productivity. In contrast, no impact of capital stock on production is found when using the FAO measure. Interestingly, when estimating the across-country production, capital stock is positively associated with production regardless of the measure used, and the magnitude of the impact of tractors (FAO data) is almost identical to that of the fixed capital component (new dataset). The authors warn that “one may erroneously conclude that tractors are a valid proxy for agricultural fixed capital.”

The remainder of the results from the agricultural production model are consistent with prior studies. From a macroeconomic perspective, level of development and education are found to be positively associated with agricultural production. Aridity and inflation are both found to be negatively affected with agricultural production.

Quality and availability of data are integral for accurate economic analyses, and are generally a constraint to conducting research in developing countries. In the absence of comprehensive data on agricultural capital stock, the quantity of tractors has served as an adequate proxy. However, the quantity does not capture the quality or technology of the tractors, nor does it capture other machinery or capital inputs to agriculture production. The findings in this study, using an updated and more comprehensive dataset, provide a more precise economic analysis.