Session 7: Asset Pricing Theory and Computation
August 18-20, 2019 | Landau Economics Bldg, 579 Serra Mall, Rm 134, Stanford
This session focuses on recent advances in the theory of asset pricing and the use of computational techniques such as machine learning. Possible topics include but are not limited to the following: learning and ambiguity in asset pricing models, investor heterogeneity, new preference structures for pricing models, or using machine learning to understand the cross-section of returns. As the analysis of such models often requires the use of computational methods, we encourage submissions that develop and make use of new numerical techniques.
Organizers: Kenneth Judd (Hoover Institution, Stanford), Walter Pohl (Norwegian School of Economics, Bergen), Karl Schmedders (IMD Lausanne and University of Zurich) and Ole Wilms (Tilburg University)
Aug 18 | 8:30 am to 9:00 am
Check-in | Breakfast
Aug 18 | 9:00 am to 9:45 am
What’s vol REALLY got to do with it
Presented by: Friedrich Lorenz (University of Munster)
Aug 18 | 9:45 am to 10:00 am
Aug 18 | 10:00 am to 10:45 am
Estimating the Anomaly Baserate
Presented by: Alex Chinco (University of Illinois)
The academic literature contains literally hundreds of variables that seem to predict the cross-section of expected returns. This so-called ‘anomaly zoo’ has caused many to question whether researchers are using the right tests for statistical significance. But, here’s the thing: even if a researcher is using the right tests, he will still be drawing the wrong conclusions from his analysis if he is starting out with the wrong priors—i.e., if he is starting out with incorrect beliefs about the ex ante probability of discovering a tradable anomaly prior to seeing any test results.
So, what are the right priors to start out with? What is the correct anomaly baserate?
We propose a new statistical approach to answer this question. The key insight is that, under certain conditions, there’s a one-to-one mapping between the ex ante probability of discovering a tradable anomaly and the best-fit tuning parameter in a penalized regression. When we apply our new statistical approach to the cross-section of monthly returns, we find that the anomaly baserate has fluctuated substantially since the start of our sample in May 1973. The ex ante probability of discovering a tradable anomaly was much higher in 2003 than in 1990. As a proof of concept, we construct a
trading strategy that invests in previously discovered predictors and show that adjusting this strategy to account for the prevailing anomaly baserate boosts its performance.
Aug 18 | 10:45 am to 11:00 am
Aug 18 | 11:00 am to 11:45 am
Valuing Private Equity Investments Strip by Strip
Presented by: Arpit Gupta (NYU Stern)
We propose a new valuation method for private equity investments. First, we construct a cash-flow replicating portfolio for the private investment, using cash-flows on various listed equity and fixed income instruments. The second step values the replicating portfolio using a flexible asset pricing model that accurately prices the systematic risk in listed equity and fixed income instruments of different horizons. The method delivers a measure of the risk-adjusted profit earned on a PE investment, a time series for the expected return on PE fund categories, and a time series for the residual net asset value in a fund. We apply the method to real estate, infrastructure, buyout, and venture capital funds, and find modestly positive average risk-adjusted profits with substantial cross-sectional variation, and declining expected returns in the later part of the sample.
Aug 18 | 11:45 am to 1:00 pm
Aug 18 | 1:00 pm to 1:45 pm
Variance Risk Premium Components and International Stock Return Predictability
Presented by: Nancy Xu (Boston College)
In this paper, we document and explain the distinct behaviors of U.S. downside and upside variance risk premiums (DVP and UVP, respectively) and their international stock return predictability patterns. DVP, the compensation for bearing downside variance risk, is positive, highly correlated with the total variance premium, and countercyclical, whereas
UVP is, on average, borderline positive and procyclical with large negative spikes around episodes of market turmoil. We then provide robust evidence that decomposing VP into its downside and upside components signicantly improves domestic and international stock return predictability. DVP is a robust predictor at four to six months and exhibits a hump-shaped pattern, whereas UVP performs the best at very short horizons. These stylized facts
highlight the importance of acknowledging asymmetry in equity risk premiums. Hence, in the second part of the paper, we rationalize the economic sources of DVP and UVP in an international dynamic asset pricing model featuring asymmetric and time-varying risk aversion and economic uncertainty in a partially integrated world economy. We show that DVP is mostly driven by the upside movements of risk aversion, whereas UVP loads significantly and negatively on downside economic uncertainty. Moreover, we nd that DVP
(UVP) transmits to international markets mostly through nancial integration (real eco-
Aug 18 | 1:45 pm to 2:00 pm
Aug 18 | 2:00 pm to 2:45 pm
Presented by: Ye Li (The Ohio State University)
Delegation bears an intrinsic form of uncertainty. Investors hire managers for their superior models of asset markets, but delegation outcome is uncertain precisely because managers' model is unknown to investors. We model investors' delegation decision as a trade-off between asset return uncertainty and delegation uncertainty. Our theory explains several puzzles on fund performances. It also delivers asset pricing implications supported by our empirical analysis: (1) because investors partially delegate and hedge against delegation uncertainty, CAPM alpha arises; (2) the cross-section dispersion of alpha increases in uncertainty; (3) managers bet on alpha, engaging in factor timing, but factors' alpha is immune to the rise of their arbitrage capital - when investors delegate more, delegation hedging becomes stronger. Finally, we offer a novel approach to extract model uncertainty from asset returns, delegation, and survey expectations.
Aug 18 | 2:45 pm to 3:00 pm
Aug 18 | 3:00 pm to 3:45 pm
Can the Machine Pick Stock Market Winners?
Presented by: Walter Pohl (Norwegian School of Economics)
Aug 19 | 8:30 am to 9:00 am
Check-in | Breakfast
Aug 19 | 9:00 am to 9:45 am
Presented by: Nina Boyarchenko (Federal Reserve Bank of New York)
Aug 19 | 9:45 am to 10:00 am
Aug 19 | 10:00 am to 10:45 am
A Macroeconomic Model of Equities and Real, Nominal, and Defaultable Debt
Presented by: Eric Swanson (University of California, Irvine)
I show that a simple, fully structural New Keynesian model with Epstein-Zin
preferences is consistent with a wide variety of asset pricing facts, such as the size and variability of risk premia on equities, real and nominal government bonds, and corporate bonds—the equity premium puzzle, bond premium puzzle, and credit spread puzzle, respectively. I thus show how to unify a variety of asset pricing puzzles from finance into a simple, structural framework. Conversely, I show how to bring standard macroeconomic models into agreement with a wide range of asset pricing facts.
Aug 19 | 10:45 am to 11:00 am
Aug 19 | 11:00 am to 11:45 am
Economic Uncertainty and Investor Attention
Presented by: Daniel Andrei (McGill University)
Aug 19 | 11:45 am to 1:00 pm
Aug 19 | 1:00 pm to 1:45 pm
How Alternative Are Private Markets?
Presented by: Elise Gourier (ESSEC Business School)
We present a new and flexible methodology to build factors in illiquid markets, from an unbalanced panel of smoothed asset returns. We apply this methodology to a large and unique panel of private market funds. We build a set of eight private factors that capture nearly 60% of the variation in private market returns. Four of these factors command a risk premium above 3%, but half of the variation in their returns is explained by standard listed equity factors. Exposure to these factors can be gained by forming a portfolio with given fund characteristics that include region of investment, industry focus, investment strategy, and fund size.
Aug 19 | 1:45 pm to 2:00 pm
Aug 19 | 2:00 pm to 2:45 pm
An Equilibrium Model with Buy and Hold Investors
Presented by: Tao Wu (Illinois Institute of Technology)
This is the first paper to analyze the effects of buy and hold investors on equilibrium security price dynamics while previous studies mainly dealt with less general restricted participation case. The empirical literature suggests that many investors follow buy and hold strategies by rarely changing asset and flow allocations due to information costs or other frictions. Similar strategies are documented for institutional investors. A buy and hold investor effectively faces an incomplete market and differs in her pricing of risk from a dynamic asset allocator. The equilibrium is solved through the construction of a representative agent with state-dependent utility. The fraction of the stock held by the buy and hold investor emerges as an additional state variable. Characterizations of equilibrium quantities are given analytically as functions of the state variables. In contrast to most previous literature, stock return volatility is solved endogenously in this paper. A simple calibration of our model shows that the economy with buy and hold investors can simultaneously produce a low interest rate and a high Sharpe ratio for the stock. In addition, the buy and hold economy can deliver stock return volatility more than twice that in the limited participation economy, because the stock price is potentially more sensitive to dividend shocks in the buy and hold economy. Moreover, the buy and hold economy achieves this while keeping interest rate volatility at reasonably low levels at the same time. The intuition is related to the smaller impact of dividend shocks on the welfare weight in the buy and hold economy as both types of agents hold stocks. Finally, this paper is also among the first to solve a model with portfolio constraints when investors have potentially non-logarithmic utilities while previous literature mainly considered the case when the constrained investor has logarithmic utility and hence is myopic.
Aug 19 | 2:45 pm to 3:00 pm
Aug 19 | 3:00 pm to 3:45 pm
Size Premium Waves
Presented by: Howard Kung (London Business School)
Small firms earned higher average returns than big firms over the past century. However, the relation between firm size and expected returns has varied significantly over time. The size premium was large and significant before 1970, disappeared in following two decades, and reemerged strongly after 2000. The periods with a significant size effect are preceded by persistently higher microeconomic uncertainty, measured as the cross-sectional dispersion in firm productivity. We explain the size premium waves in an investment-based asset pricing model featuring time-varying microeconomic uncertainty. Small firms are more exposed to macroeconomic risks relative to big firms, and this relation is magnified during states of high productivity dispersion. The model generates a positive unconditional size premium, but can reproduce a statistically insignificant (significant) size premium in periods of low (high) micro uncertainty.
Aug 20 | 8:30 am to 9:00 am
Check-in | Breakfast
Aug 20 | 9:00 am to 9:45 am
Because all Moments Matter: Maximum Likelihood Estimation of Long-Run Risk Models
Presented by: Ole Wilms (Tilburg University)
Aug 20 | 9:45 am to 10:00 am
Aug 20 | 10:00 am to 10:45 am
Endogenous Price War Risks
Presented by: Winston Dou (University of Pennsylvania)
We develop a general-equilibrium asset-pricing model with dynamic games of price competition. Price war risks arise endogenously from declines in long-run growth as firms’ incentive to undercut prices grow stronger with a worse growth outlook. The triggered price wars have amplification effects by narrowing profit margins. In industries with higher capacity of radical innovation, firms compete more fiercely for future market dominance. Their incentives for price undercutting are less responsive to long-run growth shocks, and they are more immune to price war risks and long-run growth shocks. Our results shed new light on the relationship between competition and equity returns.
Aug 20 | 10:45 am to 11:00 am
Aug 20 | 11:00 am to 11:45 am
Market Power in the United States and its Impact on Income and Wealth Distribution
Presented by: Mordecai Kurz (Stanford University)
Aug 20 | 11:45 am to 1:00 pm
Aug 20 | 1:00 pm to 1:45 pm
A Profile Likelihood Ratio Based Test of Return Predictability
Presented by: Robert Erbe (University of Zurich)
Aug 20 | 1:45 pm to 2:00 pm
Aug 20 | 2:00 pm to 2:45 pm
Forecasting in Big Data Environments: an Adaptable and Automated Shrinkage Estimation of Neural Networks (AAShNet)
Presented by: Ali Habibnia (Virginia Tech)
This paper considers improved forecasting in possibly nonlinear dynamic settings, with high-dimension predictors (“big data” environments). To overcome the curse of dimensionality and manage data and model complexity, we examine shrinkage estimation of a back-propagation algorithm of a neural net with skip-layer connections. We expressly include both linear and nonlinear components. This is a high-dimensional learning approach including both sparsityL1 and smoothnessL2 penalties, allowing high-dimensionality and nonlinearity to be accommodated in one step. This approach selects significant predictors as well asthetopology ofthe neural network. We estimate optimal values of shrinkage hyperparameters by incorporating a gradient-based optimization technique resulting in robust predictions with improved reproducibility. The latter has been an issue in some approaches. This is statistically interpretable and unravels some network structure, commonly left to a black box. An additional advantage is that the nonlinear part tends to get pruned if the underlying process is linear. In an application to forecasting equity returns, the proposed approach captures nonlinear dynamics between equitiesto enhance forecast performance. It offers an appreciable improvement over current univariate and multivariate models by RMSE and actual portfolio performance.
Aug 20 | 2:45 pm to 3:00 pm
Aug 20 | 3:00 pm to 3:45 pm
Regularising the Factor Zoo with OWL: A Correlation-Robust Machine Learning Approach
Presented by: Chuanping Sun, Queen Mary University
This paper sheds light on a new perspective of the “factor zoo enigma”, in which factor correlation prevails and worsens the “curse of dimensionality” faced by standard methods such as Fama-MacBeth (FM) regression, LASSO, etc. I introduce a newly developed machine learning tool, OWL, which circumvents complications from correlations in high dimensionality: it can identify and group correlated factors while shrinking off useless/redundant ones. I show OWL estimator is consistent and I derive the grouping condition for correlated factors. Monte Carlo experiments show that OWL outperforms LASSO, adaptive LASSO and Elastic Net (EN) in various settings, particularly when factors are highly correlated. Empirical evidence suggests that liquidity related factors are primary to drive asset prices and out-ofsample Sharpe ratio of OWL hedged portfolios is considerably larger than that of LASSO, EN and FM.