Empirical validation of CPR-IE restrictions

Determine whether the multiplicative scale response and exact log-linearity restrictions underlying CPR-IE are stable enough to improve out-of-sample decisions compared with alternative aggregation representations.

Background

The paper compares CPR-IE with additive utility, constant-elasticity-of-substitution, translog, minimum, and Pareto-based representations. Each alternative encodes different assumptions about substitution, interaction, and compensation among compression benefit, prediction benefit, and resource burden.

CPR-IE is distinguished by multiplicative scale response and exact log-linearity, but the paper does not establish empirically that these restrictions are sufficiently stable or decision-useful across deployment settings. The unresolved issue is whether adopting CPR-IE actually improves out-of-sample decisions relative to less restrictive or differently structured representations.