Research articles for the 2019-08-10
Forecasting Realized Volatility of Agricultural Commodity Futures with Infinite Hidden Markov HAR Models
SSRN
We construct a set of HAR models with three types of infinite Hidden Markov regime switching structures. Particularly, jumps, leverage effects, and speculation effects are taken into account in realized volatility modeling. We forecast five agricultural commodity futures (Corn, Cotton, Indica Rice, Palm oil and Soybean) based on high frequency data from Chinese futures markets and evaluate the forecast performances with both statistical and economic evaluation measures. The statistical evaluation results suggest that HAR models with infinite Hidden Markov regime switching structures have better precision compared the benchmark HAR models based on the MZ-R², MAFE, and MCS results. The economic evaluation results suggest that portfolios constructed with infinite Hidden Markov regime switching HARs achieve higher portfolio returns for risk averse investors compared to benchmark HAR model for short-term volatility forecasts.
SSRN
We construct a set of HAR models with three types of infinite Hidden Markov regime switching structures. Particularly, jumps, leverage effects, and speculation effects are taken into account in realized volatility modeling. We forecast five agricultural commodity futures (Corn, Cotton, Indica Rice, Palm oil and Soybean) based on high frequency data from Chinese futures markets and evaluate the forecast performances with both statistical and economic evaluation measures. The statistical evaluation results suggest that HAR models with infinite Hidden Markov regime switching structures have better precision compared the benchmark HAR models based on the MZ-R², MAFE, and MCS results. The economic evaluation results suggest that portfolios constructed with infinite Hidden Markov regime switching HARs achieve higher portfolio returns for risk averse investors compared to benchmark HAR model for short-term volatility forecasts.
Independent Boards and Bondholder Agency Risk
SSRN
Do independent boards affect bondholder agency risk? Using mandatory board reforms, I show that firms that transition to independent boards experience economically significant reductions in payout, financing, and event risk covenants in their bond contracts. This effect is not offset by a higher cost of debt. My findings also reveal that the contracting consequences of the reform are particularly pronounced in Delaware, where state law limits directors’ incentives to favor equity over debt, when firms are more vulnerable to takeovers, and when firms have fewer distracted directors. These results suggest that independent boards can contribute to ameliorating bondholder agency risk.
SSRN
Do independent boards affect bondholder agency risk? Using mandatory board reforms, I show that firms that transition to independent boards experience economically significant reductions in payout, financing, and event risk covenants in their bond contracts. This effect is not offset by a higher cost of debt. My findings also reveal that the contracting consequences of the reform are particularly pronounced in Delaware, where state law limits directors’ incentives to favor equity over debt, when firms are more vulnerable to takeovers, and when firms have fewer distracted directors. These results suggest that independent boards can contribute to ameliorating bondholder agency risk.
Information Choice, Uncertainty, and Expected Returns
SSRN
We investigate how information choices impact equity returns and risk. Building upon the theory of Van Nieuwerburgh and Veldkamp (2010), we estimate a learning index that reflects the expected benefits of learning about an asset. High learning index stocks have 6.2% lower returns per year and an order of magnitude lower abnormal volatilities compared to low learning index stocks. Long run patterns in returns and volatilities, other measures of information flow, and the information environment surrounding earnings announcements confirm our interpretation of the learning index. Our findings support the model’s predictions and illustrate a novel empirical measure of investor learning.
SSRN
We investigate how information choices impact equity returns and risk. Building upon the theory of Van Nieuwerburgh and Veldkamp (2010), we estimate a learning index that reflects the expected benefits of learning about an asset. High learning index stocks have 6.2% lower returns per year and an order of magnitude lower abnormal volatilities compared to low learning index stocks. Long run patterns in returns and volatilities, other measures of information flow, and the information environment surrounding earnings announcements confirm our interpretation of the learning index. Our findings support the model’s predictions and illustrate a novel empirical measure of investor learning.
Machine Learning Explainability in Finance: An Application to Default Risk Analysis
SSRN
We propose a framework for addressing the ‘black box’ problem present in some Machine Learning (ML) applications. We implement our approach by using the Quantitative Input Influence (QII) method of Datta et al (2016) in a realâ€'world example: a ML model to predict mortgage defaults. This method investigates the inputs and outputs of the model, but not its inner workings. It measures feature influences by intervening on inputs and estimating their Shapley values, representing the features’ average marginal contributions over all possible feature combinations. This method estimates key drivers of mortgage defaults such as the loanâ€'toâ€'value ratio and current interest rate, which are in line with the findings of the economics and finance literature. However, given the nonâ€'linearity of ML model, explanations vary significantly for different groups of loans. We use clustering methods to arrive at groups of explanations for different areas of the input space. Finally, we conduct simulations on data that the model has not been trained or tested on. Our main contribution is to develop a systematic analytical framework that could be used for approaching explainability questions in real world financial applications. We conclude though that notable model uncertainties do remain which stakeholders ought to be aware of.
SSRN
We propose a framework for addressing the ‘black box’ problem present in some Machine Learning (ML) applications. We implement our approach by using the Quantitative Input Influence (QII) method of Datta et al (2016) in a realâ€'world example: a ML model to predict mortgage defaults. This method investigates the inputs and outputs of the model, but not its inner workings. It measures feature influences by intervening on inputs and estimating their Shapley values, representing the features’ average marginal contributions over all possible feature combinations. This method estimates key drivers of mortgage defaults such as the loanâ€'toâ€'value ratio and current interest rate, which are in line with the findings of the economics and finance literature. However, given the nonâ€'linearity of ML model, explanations vary significantly for different groups of loans. We use clustering methods to arrive at groups of explanations for different areas of the input space. Finally, we conduct simulations on data that the model has not been trained or tested on. Our main contribution is to develop a systematic analytical framework that could be used for approaching explainability questions in real world financial applications. We conclude though that notable model uncertainties do remain which stakeholders ought to be aware of.
Optimally Solving Banks' Legacy Problems
SSRN
We characterize policy interventions directed to minimize the cost to the deposit guarantee scheme and the taxpayers of banks with legacy problems. Non-performing loans (NPLs) with low and risky returns create a debt overhang that induces bank owners to forego profitable lending opportunities. NPL disposal requirements can restore the incentives to undertake new lending but, as they force bank owners to absorb losses, can also make them prefer the bank being resolved. For severe legacy problems, combining NPL disposal requirements with positive transfers is optimal and involves no conflict between minimizing the cost to the authority and maximizing overall surplus.
SSRN
We characterize policy interventions directed to minimize the cost to the deposit guarantee scheme and the taxpayers of banks with legacy problems. Non-performing loans (NPLs) with low and risky returns create a debt overhang that induces bank owners to forego profitable lending opportunities. NPL disposal requirements can restore the incentives to undertake new lending but, as they force bank owners to absorb losses, can also make them prefer the bank being resolved. For severe legacy problems, combining NPL disposal requirements with positive transfers is optimal and involves no conflict between minimizing the cost to the authority and maximizing overall surplus.
Risky Bank Guarantees
SSRN
Applying standard portfolio-sort techniques to bank asset returns for 15 countries from 2004 to 2018, we uncover a risk premium associated with implicit government guarantees. This risk premium is intimately tied to sovereign risk, suggesting that guaranteed banks, defined as those of particular importance to the national economy, inherit the risk of the guarantor. Indeed, this premium does not exist in safe-haven countries. We rationalize these findings with a model in which implicit government guarantees are risky in the sense that they provide protection that depends on the aggregate state of the economy.
SSRN
Applying standard portfolio-sort techniques to bank asset returns for 15 countries from 2004 to 2018, we uncover a risk premium associated with implicit government guarantees. This risk premium is intimately tied to sovereign risk, suggesting that guaranteed banks, defined as those of particular importance to the national economy, inherit the risk of the guarantor. Indeed, this premium does not exist in safe-haven countries. We rationalize these findings with a model in which implicit government guarantees are risky in the sense that they provide protection that depends on the aggregate state of the economy.
Should the CCyB Be Enhanced with a Sectoral Dimension?
SSRN
The paper investigates whether there is sufficient empirical support in Italy for the introduction of a sectoral countercyclical capital buffer (CCyB) in the macroprudential framework. We study the sectoral decomposition of the credit-to-GDP gap over the period 1990Q1-2017Q2. Overall, our results suggests that a sectoral CCyB could be a useful addition to the macroprudential framework as both the timing for activation and the size of the capital buffer can differ when accounting for the sectoral dimension of the credit-to-GDP gap. We find that the synchronicity of sectoral credit cycles decreases as we move from a two-sector to a six-sector decomposition. Moreover, the contribution of sectoral cycles to systemic stress, as measured by the system-wide new bad debt rate, as well as the prudential requirements associated with their risk exposure differ quite significantly. While exuberance in the non-real-estate related segment of corporate lending is usually followed by a surge in systemic stress, exuberance in the real-estate related segment of business lending does not.
SSRN
The paper investigates whether there is sufficient empirical support in Italy for the introduction of a sectoral countercyclical capital buffer (CCyB) in the macroprudential framework. We study the sectoral decomposition of the credit-to-GDP gap over the period 1990Q1-2017Q2. Overall, our results suggests that a sectoral CCyB could be a useful addition to the macroprudential framework as both the timing for activation and the size of the capital buffer can differ when accounting for the sectoral dimension of the credit-to-GDP gap. We find that the synchronicity of sectoral credit cycles decreases as we move from a two-sector to a six-sector decomposition. Moreover, the contribution of sectoral cycles to systemic stress, as measured by the system-wide new bad debt rate, as well as the prudential requirements associated with their risk exposure differ quite significantly. While exuberance in the non-real-estate related segment of corporate lending is usually followed by a surge in systemic stress, exuberance in the real-estate related segment of business lending does not.
The Expansion of Consumer Credit in Italy and in the Euro Area: What Are the Drivers and the Risks?
SSRN
Since 2015 consumer loans have been rising fast in France, Germany, Italy, and Spain. Credit demand, specifically for consumer durables, has played a crucial role; the easing of supply conditions has been relevant only in Italy and Spain, which experienced stronger credit tightening during the past crises. Risks stemming from the growth of consumer credit are mitigated by its lower incidence, compared with mortgages, on households’ total debt and income; exposure to interest rate risk is also decreasing due to the high share of fixed-rate contracts. There is wide risk heterogeneity across countries, with Italy and Spain having the highest share of delinquent households (even for less than 90 days). In Italy, however, debt is increasingly concentrated among more affluent households, which are better able to withstand negative economic shocks; this trend is sustaining the drop in the ratio of new non-performing consumer loans.
SSRN
Since 2015 consumer loans have been rising fast in France, Germany, Italy, and Spain. Credit demand, specifically for consumer durables, has played a crucial role; the easing of supply conditions has been relevant only in Italy and Spain, which experienced stronger credit tightening during the past crises. Risks stemming from the growth of consumer credit are mitigated by its lower incidence, compared with mortgages, on households’ total debt and income; exposure to interest rate risk is also decreasing due to the high share of fixed-rate contracts. There is wide risk heterogeneity across countries, with Italy and Spain having the highest share of delinquent households (even for less than 90 days). In Italy, however, debt is increasingly concentrated among more affluent households, which are better able to withstand negative economic shocks; this trend is sustaining the drop in the ratio of new non-performing consumer loans.
Urban Agglomerations and Firm Access to Credit
SSRN
The paper investigates whether firms have better access to bank credit in areas with a larger degree of urbanization. It uses bank-firm data drawn from the Credit Register maintained at the Bank of Italy to devise an indicator of ease of access to credit. The paper proposes an instrumental variable strategy that uses as instruments past population density and urbanization driven by considerations of political economy. The results show that urbanization affects access to credit positively for construction firms, whose collateral greatly benefits from thicker real estate markets. No impact is found for service and manufacturing firms.
SSRN
The paper investigates whether firms have better access to bank credit in areas with a larger degree of urbanization. It uses bank-firm data drawn from the Credit Register maintained at the Bank of Italy to devise an indicator of ease of access to credit. The paper proposes an instrumental variable strategy that uses as instruments past population density and urbanization driven by considerations of political economy. The results show that urbanization affects access to credit positively for construction firms, whose collateral greatly benefits from thicker real estate markets. No impact is found for service and manufacturing firms.
Using Credit Variables to Date Business Cycle and to Estimate the Probabilities of Recession in Real Time
SSRN
Following the debate on the relationship between business and financial cycle rekindled in the last decade since the global financial crisis, we assess the ability of some financial indicators to track the Italian business cycle. We mostly use credit variables to detect the turning points and to estimate the probability of recession in real time. A dynamic factor model with Markov-switching regimes is used to handle a large dataset and to cope with the nonlinear evolution of the business cycle. The in-sample results strongly support the capacity of credit variables to estimate the probability of recessions and the implied coincident indicator proves their ability to fit the business cycle. Also in real time the contribution of credit is not negligible compared to that of the industrial production, currently used for the conjunctural analysis.
SSRN
Following the debate on the relationship between business and financial cycle rekindled in the last decade since the global financial crisis, we assess the ability of some financial indicators to track the Italian business cycle. We mostly use credit variables to detect the turning points and to estimate the probability of recession in real time. A dynamic factor model with Markov-switching regimes is used to handle a large dataset and to cope with the nonlinear evolution of the business cycle. The in-sample results strongly support the capacity of credit variables to estimate the probability of recessions and the implied coincident indicator proves their ability to fit the business cycle. Also in real time the contribution of credit is not negligible compared to that of the industrial production, currently used for the conjunctural analysis.
Why US Firms Use More Long-Term Debt Post Activist Interventions?
SSRN
We find that US firms increase the use of longer-term debts post hedge fund activism. The target firms' median proportion of debt maturing in more than 3 years increases by 19% in three years around activists’ interventions. Our results suggest that this debt maturity change may be influenced by both bankers’ reluctance to provide capital (supply constraints) and targets’ increasing reliance on longer-term public debts (demand-side factors). Hedge fund activism increases the propensity to raise longer-term public debts in target firms. This indicates that new longer-term debtholders believe in ‘shared benefits’ hypothesis by extending longer-term debts to target firms. The overall increase in debt maturity is more pronounced in target firms associated with governance reforms. Collectively, our findings suggest possible governance substitution from shorter-term debtholders to the activist hedge funds.
SSRN
We find that US firms increase the use of longer-term debts post hedge fund activism. The target firms' median proportion of debt maturing in more than 3 years increases by 19% in three years around activists’ interventions. Our results suggest that this debt maturity change may be influenced by both bankers’ reluctance to provide capital (supply constraints) and targets’ increasing reliance on longer-term public debts (demand-side factors). Hedge fund activism increases the propensity to raise longer-term public debts in target firms. This indicates that new longer-term debtholders believe in ‘shared benefits’ hypothesis by extending longer-term debts to target firms. The overall increase in debt maturity is more pronounced in target firms associated with governance reforms. Collectively, our findings suggest possible governance substitution from shorter-term debtholders to the activist hedge funds.
