Min Qi

Independent Researcher · Credit Risk Analysis

Using a unique and comprehensive dataset of loan-level home equity lines of credit serviced by large US national banks, we confirm that default risk of home equity lines of credit increases at end of draw. More importantly, we quantify the increase in default risk with the size of positive payment shock at end of draw. Furthermore, we find the effe...

We examine the relevance and effectiveness of stock return correlations among financial institutions as an indicator of systemic risk. By analyzing the trends and fluctuations of daily stock return correlations and default correlations among the 22 largest bank holding companies and investment banks from 1988 to 2008, we find that daily stock retur...

We compare six modeling methods for Loss Given Default (LGD). We find that non-parametric methods (regression tree and neural network) perform better than parametric methods both in and out of sample when over-fitting is properly controlled. Among the parametric methods, fractional response regression has a slight edge over OLS regression. Performa...

This paper examines the determinants of the outcomes of the default recovery process. We find that a new variable that incorporates not only the percentage of debt more senior to the debt instrument, but also debt at the same rank, is the most important factor driving the recovery rate. It is responsible for more than half of the total variation in...

We examine the impact of the unobservable systematic risk factor on default prediction model performance. We find that including the unobservable systematic risk factor might help improve predictive accuracy, but it might not help improve rank ordering of firms by default risk. Rank ordering is mainly driven by firm-level variables, while predictiv...

Retail exposure at default (EAD) is one of the weakest areas of risk measurement and modeling in industry practices and in academic literature. The U.S. Basel II Final Rule is not specific about the approach to EAD. In this study, we use borrower and account information from a large national sample of unsecured credit card defaults to capture borro...

This study examines the distribution of extreme values in daily currency changes for nine Asian countries. Using an improved estimator, extreme changes in Asian currencies can generally be represented by Frechet distributions. Our results are robust to the choice of the numeraire currency, the Asian crises and the 1985 Plaza Agreement. These result...

This paper studies loss given default using a large set of historical loan-level default and recovery data of high loan-to-value residential mortgages from several private mortgage insurance companies. We show that loss given default can largely be explained by various characteristics associated with the loan, the underlying property, and the defau...

Despite its great importance, there has been no general consensus on how to model the trends in time-series data. Compared to traditional approaches, neural networks (NNs) have shown some promise in time-series forecasting. This paper investigates how to best model trend time series using NNs. Four different strategies (raw data, raw data with time...

In this paper, we study consumer credit card adoption behavior when individuals are overly optimistic about their future usage of the card. We hypothesized that the more prone consumers are to unrealistic optimism, regarding their future borrowing behavior, the more likely they are to prefer credit cards with features that are sub-optimal in light...

We report evidence on the profitability and statistical significance among 2,127 technical trading rules. The best rules are found to be significantly profitable based on standard tests. We then employ White's (2000) Reality Check to evaluate these rules and find that data-snooping biases do not change the basic conclusions for the full sample. A s...

Neural networks have been widely used as a promising method for time series forecasting. However, limited empirical studies on seasonal time series forecasting with neural networks yield mixed results. While some find that neural networks are able to model seasonality directly and prior deseasonalization is not necessary, others conclude just the o...

This paper employs a neural network (NN) to study the nonlinear predictability of exchange rates for four currencies at the 1-, 6- and 12-month forecast horizons. We find that our neural network model with market fundamentals cannot beat the random walk (RW) in out-of-sample forecast accuracy, although it occasionally shows a limited market-timing...

Despite its great importance, there has been no general consensus on how to model the trends in time series data. Compared to traditional approaches, neural networks have shown some promise in time series forecasting. This paper investigates how to best model trend time series using neural networks. Four strategies (raw data, raw data with time ind...

How to accurately predict customers’ adoption behavior is becoming more important and challenging to many credit card marketers as competition increases. This calls for more knowledge about the consumer utility function and the corresponding decision behavior. In this study, we challenge the commonly used logit model which implies linear utility fu...

The article empirically investigates the stochastic properties of a widely used indicator of country risk: Institutional Investor's creditworthiness ratings. It tests whether Institutional Investor's ratings of Middle Eastern countries follow a random walk by checking for unit root. It is important to test for unit root because estimated relationsh...

The author reports how he generated a set of real time neural network forecasts to win the second place in the Financial Series Competition at the International Joint Conference on Neural Networks in 1999 (IJCNN'99). The accuracy and profitability of his forecasts are evaluated, and the finance implication is discussed. He provides comprehensive ev...

Forecasting future retail sales is one of the most important activities that form the basis for all strategic and planning decisions in effective operations of retail businesses as well as retail supply chains. This chapter illustrates how to best model and forecast retail sales time series that contain both trend and seasonal variations. The effec...

We study the effectiveness of cross validation, Bayesian regularization, early stopping, and bagging to mitigate overfitting and improving generalization for pricing and hedging derivative securities with daily S&P 500 index daily call options from January 1988 to December 1993. Our results indicate that Bayesian regularization can generate signifi...

This paper examines the relevance of various financial and economic indicators in predicting US recessions via neural network models. We share the view that business cycles are asymmetric and cannot be adequately accommodated by linear constant-parameter single-index models. We employ a novel neural network (NN) to recursively model the relationshi...

Like many other economic time series, US aggregate retail sales have strong trend and seasonal patterns. How to best model and forecast these patterns has been a long-standing issue in time-series analysis. This article compares artificial neural networks and traditional methods including Winters exponential smoothing, Box–Jenkins ARIMA model, and...

Artificial neural networks (ANNs) have received more and more attention in time series forecasting in recent years. One major disadvantage of neural networks is that there is no formal systematic model building approach. In this paper, we expose problems of the commonly used information-based in-sample model selection criteria in selecting neural n...

Forecasting future retail sales is one of the most important activities that form the basis for all strategic and planning decisions in effective operations of retail businesses as well as retail supply chains. This chapter illustrates how to best model and forecast retail sales time series that contain both trend and seasonal variations. The effec...

Inspired by the linear predictability and nonlinearity found in the finance literature, this article examines the nonlinear predictability of the excess returns. The relationship between the excess returns and the predicting variables is recursively modeled by a neural-network model, which is capable of performing flexible nonlinear functional appr...

It has been widely accepted that many financial and economic variables are non-linear, and neural networks can model flexible linear or non-linear relationships among variables. The present paper deals with an important issue: Can the many studies in the finance literature evidencing predictability of stock returns by means of linear regression be...

Data-driven modeling approaches, such as Artificial Neural Networks (ANN), are becoming more and more popular in financial applications. ANNs are nonlinear nonparametric models. ANNs allow one to fully utilize the data and let the data determine the structure and parameters of a model without any restrictive parametric modeling assumptions. They ar...

Due to some unrealistic assumptions, the traditional Black-Scholes formula systematically misprices options. This paper applies an alternative multilayer feed-forward neural network to price S&P 500 index calls. Both the in- and out-of-sample accuracy are far better for the ANN than for the Black-Scholes formula. In addition, analysis of the estima...