LTV policy as a macroprudential tool and its effects on residential mortgage loans in developing countries

The global financial crisis of 2007-2009 underlined the need for central banks to take a macroprudential perspective on financial risk, i.e., to monitor and regulate the buildup of systemic financial risk in the economy as a whole, as opposed to simply monitoring the condition of individual financial institutions (microprudential regulation). Although macroprudential measures generally lost favor in advanced economies, they have been actively used in developing countries.

The loan-to-value (LTV) ratios cap is the most commonly used macroprudential policy. In this paper, we analyze the effectiveness of LTV on the growth of residential mortgage loans (RML) in developing countries. The key empirical question is the extent to which LTV can inhibit the growth of bank lending. Our benchmark hypothesis is that LTV would restrain the growth of RML. Most studies have focused on the impacts of LTV policies on macro-level variables such as total bank credit or housing prices. Ours is one of the few studies that analyzes bank-level responses to LTV policies, and the first one to look directly at the mortgage market.

The dependent variable is RML which is a function of bank and country-specific variables, including LTV (and other macroprudential tools). We will compare the performance of alternative estimators to take into account fixed effects and endogeneity. Ordinary least squares (OLS) estimates are based on a static panel data model of mortgage loans. A straightforward augmentation of the static model to the dynamic panel data model, introduces the first lag of the dependent variable. This provides a clear justification to use lags of the dependent variable, as well as the explanatory variables, as instruments. Therefore we compare our results to the first difference and system generalized method of moments (GMM) estimators, the two most commonly used estimators in applied work. The results suggests the difference between the OLS and GMM estimates are not statistically significant. The presence of outliers is also considered. Robust to outlier estimators, the results suggest as much as 24% of the sample may be contaminated with outliers.

Our results provide evidence LTV may result in significantly lower mortgage loan creation by around 7.4%. Once other macro-prudential tools are introduced, the direct effect of LTV is around 4.5% but the use of other macro-prudential tools act complimentarily to LTV reducing further RML growth. Other macro-prudential tools found relevant are  debt-to-income ratios, leverage ratio and domestic currency loans limits policies.