Cornell University Graduate School >
Theses and Dissertations (OPEN) >
Please use this identifier to cite or link to this item:
|Title: ||Efficiency of New York Dairy Farms: Exploring the Role of Managerial Ability|
|Authors: ||Byma, Justin|
|Issue Date: ||4-Aug-2006|
|Abstract: ||This paper explores the role of management ability in explaining efficiency on New York dairy farms using both Data Envelopment Analysis (DEA) and stochastic frontier estimation. First, we test whether computed technical, cost, and revenue efficiencies under DEA are due to a missing input, which we argue may be the management input. Using an unbalanced panel of individual farm data from the Cornell University?s Dairy Farm Business Summary (DFBS) from 1993 - 2004 we define 6 inputs, including operator labor, hired labor, purchased feed, livestock, capital, and crop inputs, and two outputs, including milk output and all other outputs. We define the management input in two ways. First, the DFBS asks farmers to estimate their own values of labor and management. Second, the panel nature of the data set allows us to use the previous year?s net farm income as a measure of farmer management ability. Using the lagged data prevents any contemporaneous bias in efficiency measurement and is consistent with Stigler?s conjecture that differences in management ability should be captured in profits.
To test for the effects of the missing management input we first calculate DEA efficiency scores using the original six inputs and two outputs. These efficiencies are recalculated first using operators? values of labor and management, and then using lagged net farm income, in place of the operator labor input. The resulting efficiency scores are compared. We find weak evidence of the missing management input using our two measures, and that change in computed efficiencies resulting from including the management input depends on whether one uses an input or output orientation. The change in efficiencies using operators? values of labor and management are small, often less than 1 percent. Using lagged net farm income as the management input increases computed input-oriented technical efficiency by an average of 1 percent and cost efficiency by 1.2 percent. Output oriented technical efficiency increases by 1.7 percent and revenue efficiency increases by nearly 2 percent. The impact of this measure of the management input on the allocative components of cost and revenue efficiencies was negligible, indicating that this measure of management ability serves more to explain differences in technology choice than allocative abilities.
We also estimate input- and output-oriented technical efficiencies, cost efficiencies and revenue efficiencies using stochastic frontier functions. The technical efficiencies are estimated using distance function methodologies. We transform our management input variables to a per cow basis and include them as efficiency effects variables along with operator age, education, farm size, and years of participation in the DFBS. This allows us to measure the impacts of management ability on farm efficiency while controlling for other factors that may also affect efficiency. We estimate conditional mean and heteroscedastic efficiency term specifications for each frontier model. We again find that using lagged net farm income per cow may be a preferred measure of management ability than farmers? own estimates of the value of their labor and management per cow. We find that, at the margin, this measure of management ability increases input-oriented technical efficiency by 1.4 ? 1.5 percent and cost efficiency by between 1.7 ? 2.9 percent, depending on specification. Output oriented technical efficiency and revenue efficiency increase at the margin by 1.8 ? 3.0 percent and by 2.4 ? 4.2 percent respectively. We also find increasing efficiency with operator education, farm size, and extended DFBS participation and decreasing efficiency with operator age.
Finally, we present a discussion of whether an input or output orientation is more appropriate for the farms in our sample and compare the dairy farm efficiencies predicted by DEA and the stochastic frontiers. We find that, for our data set, the distributions of farm efficiencies are very similar under DEA and stochastic frontier estimations, but individual farm rankings are quite different between the two.|
|Appears in Collections:||Theses and Dissertations (OPEN)|
Items in eCommons are protected by copyright, with all rights reserved, unless otherwise indicated.