This paper focuses on the dynamics of cross-correlations and conditional risk premia.
Return correlations between assets with different amounts of cash-flow risk exhibit substantial variation over time. This paper argues that this variation in correlations is likely to manifest in the assets’ risk premia, however, the sources of risk premia and time-variation in correlations in financial markets are not yet well understood. This paper attempts to discover the economic mechanisms that drive risk premia on different asset classes.
The author specify two technologies with different amounts of capital risk and adjustment costs to investment. The two technologies can be considered as low-risk companies and hi-risk companies.
When one technology has a good shock, investors want to rebalance towards the other technology. However, they face adjustment costs, which drives up the price of the technology with no shock. This mechanism is called rebalancing, which produces a positive correlation between returns on the two technologies, even when they have different cash flows (similar to Cochrane et al 2008’s “two trees”).
The model also features another mechanism: flight-to-safety effect as a result of the time-varying risk-aversion that investors move from the riskier to the less risky technology, which produces a negative correlation between returns on the two technologies.
The authors argues that when risk aversion is high, the flight-to-safety mechanism dominates, driving the correlation between bond and stock returns and real bond risk premium negative; when risk aversion is low, the rebalancing mechanism dominants and results in a positive correlation between bond and stock return and a positive term premium.
Rebalancing relies on slow physical capital reallocation and thus drives low frequency dynamics. Flight-to-safety operates at higher frequencies and drives most of the variation in financial variables in the model.
Therefore rebalancing and flight-to-safety result in time-varying correlation between returns on assets with different amounts of cash-flow risk, that potentially changes sign, even when their cash flows are independent.
This paper also shows empirical and model-implied proxies for risk aversion, which is measured through credit spreads, VIX, investors fear index by Bollerslev and Todorow (2011).
Model: General Equilibrium Model
Abstract: I use a general equilibrium model to jointly explain the time-variation in real bond and stock risk premia along with time-variation in the comovement of realized returns. The model features multiple investment technologies and produces stock and real bond return correlation which changes both in magnitude and sign. The model also delivers a time-varying real term premium that changes sign. I find that changes in investors’ appetite for risk are an important source of variation in asset prices. In response to this shock, the term premium and the stock risk premium move in opposite directions. Real default-free bonds are good hedges against an increase in stock discount rates. When the level of stock discount rates is high, the bond risk premium and bond-stock correlations are negative (and vice versa). An empirical ICAPM with shocks to discount rates as a factor jointly prices the cross-section of bonds (by maturity) and stock portfolios (value, size, momentum).
The author: Serhiy Kozak, Booth School of Business and Department of Economics, University of Chicago http://home.uchicago.edu/skozak/