A 'differential' differential demand system: an application to US meat demand

I present a 'differential' (threshold) Rotterdam demand model. I argue that a standard model, which holds parametric relationships constant across all sizes of shocks to prices and expenditures, may produce misleading elasticity estimates. This arises because consumers may react less or even not at all to small shocks to prices and expenditures. However, when shocks are large enough to exceed any adjustment or transactions costs, consumers may react more strongly, thereby producing more elastic responses. I estimate the demand for meats in the US over the 1980-2019 period. I consider four meat products - beef, pork, chicken, and turkey - and assume that meats are weakly separable from other demand decisions. I confirm that reactions to larger shocks will result in more elastic price and expenditure adjustments. A test of the difference in the regimes delineated by the size of shocks to prices and expenditures is statistically significant.