Hwagyun Kim

Do Individuals Have Preferences Used in Macro-Finance Models? An Experimental Investigation

Recent financial studies often assume that agents have Epstein--Zin preferences---preferences that require agents to care about when uncertainty is resolved. Under this “recursive-preference” framework, the preference for uncertainty resolution is entirely determined by an agents preferences for risk and intertemporal substitution. To test the implications of this model, this paper presents an experiment designed to elicit subject preferences on risk, time, intertemporal substitution, and uncertainty resolution. Results reveal that most subjects prefer early resolution of uncertainty and have relative risk aversion greater than the reciprocal of the elasticity of intertemporal substitution, consistent with the predictions by recursive preferences. Subjects are classified in a finite mixture model by their risk, time, and intertemporal-substitution parameters. Regression results show that types predicted by the Epstein--Zin model to prefer early resolution choose early resolution with 20%--50% higher probability.Data, as supplemental material, are available at http://dx.doi.org/10.1287/mnsc.2013.1794 This paper was accepted by Brad Barber, finance.

Evaluating factor pricing models using high‐frequency panels

This paper develops a new framework and statistical tools to analyze stock returns using high frequency data. We consider a continuous-time multi-factor model via a continuous-time multivariate regression model incorporating realistic empirical features, such as persistent stochastic volatilities with leverage effects. We find that conventional regression approach often leads to misleading and inconsistent test results. We overcome this by using samples collected at random intervals, which are set by the clock running inversely proportional to the market volatility. We find that the size factor has difficulty in explaining the size-based portfolios, while the book-to-market factor is a valid pricing factor.

Velocity of money and inflation dynamics

There have been large changes in the velocity of money which could be a potential source of inflation variability. This article investigates how the velocity of money affects inflation dynamics by estimating the Phillips curve derived from a New Keynesian model in which money is introduced via transactions technology. The resultant Phillips curve becomes a function of velocity as well as an output gap and a forward looking inflation terms, a feature for which we provide empirical support. Specifically, we adopt the GMM methodology to estimate the velocity-augmented forward looking Phillips curve using the US data between 1951 and 2005. We observe that historical inflation dynamics is consistent with the view.

Ambiguous Information about Interest Rates and Bond Uncertainty Premiums

This paper studies the impact of ambiguous information regarding future interest rates on bond prices. A simple bond-pricing model with ambiguity aversion shows that positive bond uncertainty premiums exist, and the interest rate ambiguity affects the term structure of interest rates and yield volatilities. Consistent with the theory, empirical measures of interest rate ambiguity based on the Survey of Professional Forecasters data significantly predict U.S. Treasury bond returns, explain variation in term spreads and yield volatility, and bond yields asymmetrically respond to good and bad news from the Federal Reserve. Results are robust to alternative empirical specifications and out-of-sample forecasts.

Transactions Cost and Interest Rate Rules

This paper evaluates quantitatively the effect of real money balances in a New Keynesian framework. Money in our model facilitates transactions and is introduced through a transactions cost technology. This technology acts like a distortionary consumption tax which varies endogenously with the nominal interest rate. In this setup the resultant Phillips curve becomes a function of the nominal interest rate. Our analysis has important policy implications. First, we find, unlike Woodford (2003), accounting for realbalance effects does not result in the policy makers loss function having an interest rate smoothing term. Second, we show that in the case of a temporary shock to productivity the optimal policy response under discretion is to allow for a trade-off between inflation and the output gap. This trade-off arises endogenously in our model. The quantitative effects on the macroeconomic variables are found to be significant.

Ambiguity, Macro Factors, and Stock Return Volatility

Recent studies find stock returns are negatively related to idiosyncratic volatility (IVOL). We find that aggregate variables known to explain stock market volatility affect the IVOL and portfolio returns sorted by IVOL. Macroeconomic volatilities, yield spreads, dividend yield, trading volume and common factors of earnings forecast dispersions are important drivers of IVOL. Macro factors produce the negative pattern, consistent with theories of intertemporal hedging demand. Teasing out the common IVOL part, the residual IVOL is positively and significantly related to stock returns and the idiosyncratic portions of earnings forecast dispersions. This is consistent with ambiguity aversion and incomplete market hypotheses.

Yield Forecasts and Stochastic Volatility in Affine Models with Macro Factors

Affine term structure models (ATSMs) are known to have a trade-off in predicting future Treasury yields and fitting the time-varying volatility of interest rates. This paper empirically studies the role of macroeconomic variables in simultaneously achieving these two goals under affine models. To this end, we incorporate a liquidity demand theory via a measure of the velocity of money into affine models. We find that this considerably reduces the statistical tension between matching the first and second moments of interest rates. In terms of forecasting yields, the models with the velocity of money outperform among the ATSMs examined, including those with inflation and real activity. Our result is robust across maturities, forecasting horizons, risk price specifications, and the number of latent factors.