Sanjiv Ranjan Das

Yahoo! for Amazon: Sentiment Extraction from Small Talk on the Web

Extracting sentiment from text is a hard semantic problem. We develop a methodology for extracting small investor sentiment from stock message boards. The algorithm comprises different classifier algorithms coupled together by a voting scheme. Accuracy levels are similar to widely used Bayes classifiers, but false positives are lower and sentiment accuracy higher. Time series and cross-sectional aggregation of message information improves the quality of the resultant sentiment index, particularly in the presence of slang and ambiguity. Empirical applications evidence a relationship with stock values---tech-sector postings are related to stock index levels, and to volumes and volatility. The algorithms may be used to assess the impact on investor opinion of management announcements, press releases, third-party news, and regulatory changes.

Of Smiles and Smirks: A Term-Structure Perspective

An extensive empirical literature in finance has documented not only the presence of anomalies in the Black-Scholes model, but also the term structures of these anomalies (for instance, the behavior of the volatility smile or of unconditional returns at different maturities). Theoretical efforts in the literature at addressing these anomalies have largely focused on two extensions of the Black-Scholes model: introducing jumps into the return process, and allowing volatility to be stochastic. We employ commonly used versions of these two classes of models to examine the extent to which the models are theoretically capable of resolving the observed anomalies. We find that each model exhibits some term structure patterns that are fundamentally inconsistent with those observed in the data. As a consequence, neither class of models constitutes an adequate explanation of the empirical evidence, although stochastic volatility models fare somewhat better than jumps.

The surprise element: jumps in interest rates

That information surprises result in discontinuous interest rates is no surprise to participants in the bond markets. We develop a class of Poisson-Gaussian models of the Fed Funds rate to capture surprise effects, and show that these models offer a good statistical description of short rate behavior, and are useful in understanding many empirical phenomena. Jump (Poisson) processes capture empirical features of the data which would not be captured by Gaussian models, and there is strong evidence that existing Gaussian models would be well-enhanced by jump and ARCH-type processes. The analytical and empirical methods in the paper support many applications, such as testing for Fed intervention effects, which are shown to be an important source of surprise jumps in interest rates. The jump model is shown to mitigate the non-linearity of interest rate drifts, so prevalent in pure-diffusion models. Day-of-week effects are modelled explicitly, and the jump model provides evidence of bond market overreaction, rejecting the martingale hypothesis for interest rates. Jump models mixed with Markov switching processes predicate that conditioning on regime is important in determining short rate behavior.

Credit Risk Derivatives

This article models the pricing of derivatives on credit risk, instruments proposed in 1992 by the International Swap Dealers Association that have started attracting market attention. The exact structure of the instruments continues to evolve today. We develop a framework to understand the key features of this class of products. It is shown that the price of the credit risk option is the expected forward value of a put option on a risky bond with a credit level-adjusted exercise price. Stripping of credit risk from the total risk of the bond is enabled by employing a stochastic strike price for the credit risk option. The article provides a framework for trading and hedging credit risk.

A Simple Approach to Three-Factor Affine Term Structure Models

RANGARAJAN SUNDARAM is assistant professor at New York University. his article develops a simple estimation approach for three-factor models of the term structure of interest rates, exploiting the exponential-affine structure of these models. The three factors in the model are the short-term rate of interest, the long-run mean of the short-term rate, and the volatility of the short-term rate. The objective of term structure modeling from a practitioner’s viewpoint is to develop a parsimonious representation of the yield curve matching the time series and cross-sectional variation of bond yields. The more factors, the richer the time series and cross-sectional properties of bond returns the model can accommodate. Of course, complexity is often traded off against parsimony for practical reasons of implementation. So far, practicality has kept researchers in the field from going beyond three-factor models. The simple estimation method in this article should make the implementation of three-factor models easier. Implementation of the model on data over a thirty-year period indicates that it captures known theoretical features and aspects of the term structure well. Duffle and Kan [1996] show that a wide range of choices of stochastic processes for interest rate factors yield bond pricing solutions of a form now widely called exponential-a

Portfolio Optimization with Mental Accounts

ne models. It is the affine form that we use in this article to develop our implementation methodology. Even though there are three factors, the affine class of solutions enables the use of a single equation maximum-likelihood model to approximate the multivariate estimation of a three-factor model. The model fits the data well, and also enables the filtering of the

Exact solutions for bond and option prices with systematic jump risk

We integrate appealing features of Markowitz’s mean-variance portfolio theory (MVT) and Shefrin and Statman’s behavioral portfolio theory (BPT) into a new mental accounting (MA) framework. Features of the MA framework include an MA structure of portfolios, a definition of risk as the probability of failing to reach the threshold level in each mental account, and attitudes toward risk that vary by account. We demonstrate a mathematical equivalence between MVT, MA, and risk management using value at risk (VaR). The aggregate allocation across MA subportfolios is mean-variance efficient with short selling. Short-selling constraints on mental accounts impose very minor reductions in certainty equivalents, only if binding for the aggregate portfolio, offsetting utility losses from errors in specifying risk-aversion coefficients in MVT applications. These generalizations of MVT and BPT via a unified MA framework result in a fruitful connection between investor consumption goals and portfolio production.

Correlated Default Risk

A variety of realistic economic considerations make jump-diffusion models of interest rate dynamics an appealing modeling choice to price interest-rate contingent claims. However, exact closed-form solutions for bond prices when interest rates follow a mixed jump-diffusion process have proved very hard to derive. This paper puts forward two new models of interest-rate dynamics that combine infrequent, discrete changes in the interest-rate level, modeled as a jump process, with short-lived, mean reverting shocks, modeled as a diffusion process. The two models differ in the way jumps affect the central tendency of interest rates; in one case shocks are temporary, in the other shocks are permanent. We derive exact closed-form solutions for the price of a discount bond and computationally tractable schemes to price bond options.

Extracting, Linking and Integrating Data from Public Sources: A Financial Case Study

Using a comprehensive and unique data set of default probabilities from Moody’s, we examine correlations between default risk for over 7,000 US public firms. This is the first paper to empirically document the correlation structure both in the time-series and in the cross-section across almost all US non-financial firms. We find that default probabilities of issuers vary substantially over time, and are positively correlated. Moreover, the correlations across firms also varies over time systematically, in a manner that is related to an economy-wide level of default risk. Joint default risk increases as the default risk in the economy increases. Our results also suggest that the magnitude of joint default depends on the quality of issuers; highest quality issuers have higher default correlations than medium grade firms. The results in this paper are informative for managers of credit portfolios, and for structured products such as collateralized debt obligations. CORRELATED DEFAULT RISK 3

A simple approach for pricing equity options with Markov switching state variables

We present Midas, a system that uses complex data processing to extract and aggregate facts from a large collection of structured and unstructured documents into a set of unified, clean entities and relationships. Midas focuses on data for financial companies and is based on periodic filings with the U.S. Securities and Exchange Commission (SEC) and Federal Deposit Insurance Corporation (FDIC). We show that, by using data aggregated by Midas, we can provide valuable insights about financial institutions either at the whole system level or at the individual company level. The key technology components that we implemented in Midas and that enable the various financial applications are: information extraction, entity resolution, mapping and fusion, all on top of a scalable infrastructure based on Hadoop. We describe our experience in building the Midas system and also outline the key research questions that remain to be addressed towards building a generic, high-level infrastructure for large-scale data integration from public sources.