Events

Professor Andrea Gamba
Professor of Finance
Warwick Business School
The University of Warwick

We analyze interest rate setting by an econometrician adaptively estimating probit-logit coefficients. Absent covariate inflation, rates are fair and the lender breaks even. If borrowers can inflate the covariate (facing heterogeneous penalties), but rates are set using clean-data estimates: credit performance declines, especially at bottom of spectrum; subsidies are positive and decreasing in the covariate; log-likelihood declines; and the estimated slope increases if penalties are sufficiently homogeneous. At econometric fixed-points, vertical inequity can occur: Bottom and top of the covariate spectrum receive subsidies, with the middle paying above-fair rates. Horizontal inequity is pronounced in the left tail. We identify three sources of loan portfolio losses and market abandonment: non-convergence of coefficient estimates if covariate mean is low or tails fat; loss incurred prior to convergence; and losses at fixed points if left tails are fat. To remedy the latter, we propose/analyse zero-profit constrained fixed-point estimation.