Spatially Targeted LTV Policies and Collateral Values

Abstract

Many governments regulate household leverage at a national level, even when credit and housing market conditions vary substantially across locations. We explore the efficacy of loan-to-value (LTV) limits targeted towards specific neighborhoods as a macroprudential policy designed to curb local house price growth. We combine administrative data from Taiwan covering the universe of all mortgage loans, personal income tax returns, a public database of geocoded housing transactions, and bank branch balance sheets. Applying a series of matched difference-in-differences and border difference-in-discontinuity designs, we find that leverage limits are effective at reducing local house prices compared to alternative policy instruments such as transfer taxes, with no effects on delinquency rates. In response to a statutory tightening of the maximum LTV ratio to 60% from the standard 80% for mortgages on second homes, house prices decline by 6% in policy catchment areas relative to nearby neighborhoods not subject to LTV restrictions. However, we uncover two kinds of efficiency costs associated with place-based mortgage restrictions: (i) real commuting costs driven by homeowners sorting into neighborhoods where credit is easier to obtain, and (ii) mispricing, or “noise” costs, as banks and prospective homebuyers face incentives to obtain inflated appraisals to avoid the limits.

Keywords: loan-to-value ratio, place-based mortgage restrictions, macroprudential policy, intermediation, collateral misreporting, house prices, border discontinuity, location sorting

JEL Classification: E61, G21, G28, R21, R31, R38