Essays on financial stability

Abstract

This thesis investigates several aspects of economic activity that are strongly linked to the stability of the financial system. First, I examine the possible build-up of systemic risk in the shadow banking system as an unintended consequence of the implementation of macroprudential policies. Second, I analyze the short-run and long-run interactions between house prices and household debt in Australia and study relevant policy implications. Finally, I propose the use of automatic model selection, namely Least Absolute Shrinkage and Selection Operator (LASSO) with cross-validation, to improve the forecasting performance of the conventional early warning system for systemic banking crises. The findings provide insights for policymakers about several aspects of potential vulnerabilities stemming from different segments of the financial sector, and improves the early warning system so as to help policymakers better identify and monitor systemic risk in the financial system.

In Chapter 2, I investigate the impact of macroprudential policies on shadow banking activities using a panel data analysis with fixed effects containing 24 countries from 2002 to 2013. I find that the effectiveness of macroprudential policies targeting the demand for credit, namely loan-to-value and debt-to-income ratio caps, is partially undermined by regulatory arbitrage through the shadow banking system. Such a strategy is generally not found for supply-side policies with two exceptions. Sector-specific capital requirements are found to be associated with negative shadow banking growth, while loan-loss provisionings could lead to counter-productive outcomes.

In Chapter 3, I empirically investigate the short-run and long-run interactions between house prices and their drivers, with a particular focus on household debt, in a Structural Error Correction Model (SVECM) framework for Australia over the sample period 1990-2016. The cointegration analysis suggests one equilibrium relationship exists between the real house prices and the long-run determinants: household debt, housing stock, household disposable income, and the real interest rate. Household debt and housing supply are the main drivers for the equilibrium house prices. Household disposable income and the real interest rate affect the equilibrium house prices through the credit channel. In the short run, house prices and household debt are found to be mutually reinforcing. The dynamic impacts of the structural shocks suggest that macroprudential policies, if they act in a manner similar to an exogenous tightening of credit conditions, may be preferable to monetary policy leaning against the wind, if and when policies are needed to reinforce financial stability.

In Chapter 4, I propose using LASSO with cross-validation approach to automate the variable selection process of the conventional multivariate logit econometric framework, the purpose being to improve the prediction of systemic banking crises. Using a dataset covering 23 OECD countries with quarterly data from 1970Q1 to 2018Q3, the model performance is evaluated in a recursive out-of-sample forecasting exercise, taking policymakers’ preference of missed crises and false alarms into account. The results suggest that the automatic variable selection process can enhance the predictive performance of the early warning system. They also highlight the importance of extracting information from variable interactions and lags that may not be easily identified and accessed by typical subjective variable pre-selection. This simple approach is easy to interpret and is transparent, which are important aspects for effective policy communication. Five variables, namely credit growth, domestic and global credit gaps, real house price growth and the real effective exchange rate, are identified as the most important key indicators of systemic banking crises.