Substitution Elasticities between GHG Polluting and Non-polluting Inputs in Agricultural Production: A Meta-Regression

This paper reports meta-regressions of substitution elasticities between greenhouse-gas (GHG) polluting and nonpolluting inputs in agricultural production. We treat energy, fertilizer, and manure collectively as the “polluting input” and labor, land, and capital as nonpolluting inputs. We estimate meta-regressions for samples of Morishima substitution elasticities for labor, land, and capital vs. the polluting input. Much of the heterogeneity of Morishima elasticities can be explained by type of primal or dual function, functional form, type and observational level of data, input categories, the number of outputs, type of output, time period, and country categories. Each estimated long-run elasticity for the reference case, which is most relevant for assessing GHG emissions through life-cycle analysis, is greater than 1.0 and significantly different from zero. Most predicted elasticities remain significantly different from zero at the data means in the long run. These findings imply that life-cycle analysis based on fixed proportions production functions could provide grossly inaccurate measures of GHG of biofuel.