Document Type

Working Paper

Publication Date

8-31-2017

JEL Codes

Q18, C01, C24, C34, C55, D10, D12

Subject

C01 - Econometrics, C24 - Single Equation Models; Single Variables: Truncated and Censored Models; Switching Regression Models, C34 - Multiple or Simultaneous Equation Models: Truncated and Censored Models; Switching Regression Models, D10 - Household Behavior: General, D12 - Consumer Economics: Empirical Analysis, Q18 - Agricultural Policy; Food Policy

Abstract

We estimate US household monthly elasticities of demand for some of the more popular organic fruits. To our knowledge, this is the first US-wide, multi-year analysis of price and income elasticities for various organic fruits. We calculate elasticities of demand for low-income, middle class, and rich income bracket households using three estimation techniques: two econometric methods and one machine learning method (least absolute shrinkage and selection operator (LASSO)). Demand estimates are based on Nielsen scanner data from approximately 60,000 households collected from 2011 to 2013. Generally, we find that own-price conditional and unconditional elasticities of demand for organic fruits are negative. Unconditional elasticity magnitudes tend to be largest in the representative middle-class household. Income elasticities of demand measurements are inconsistent and often statistically insignificant. This finding is consistent with the survey literature finding that many consumers buy organic food for mostly moral or ethical reasons. We run two policy experiments: a 10% subsidy of organic fruits, and a 10% tax on conventional fruits. Our hypothetical policies engender a stronger reaction among the general public than habitual buyers of organic fruit; unconditional purchase and expenditure elasticities are generally larger than conditional purchase and expenditure elasticities. Finally, we find that elasticities measured with the LASSO technique are not radically different than those measured with econometric methods. The most noticeable difference between the two analytical techniques is that LASSO is more likely to find price and income elasticities of demand that indistinguishable from zero, both substantively and statistically.

Share

COinS