- Asset pricing, and its connections with corporate finance; financial constraints; credit risk; liquidity risk.
- Robustness; financial econometrics; financial machine learning.

**Hui Chen**

MIT Sloan School of Management77 Massachusetts Avenue, E62-616

Cambridge, MA 02139

Tel: (617) 324 3896

Fax: (617) 258 6855

huichen@mit.edu

**Research Interest**

**Working Papers**

Structural models in economics can offer appealing insights but often suffer from a poor fit with the data. In contrast, machine learning models offer rich flexibility but tend to suffer from over-fitting. We propose a novel framework that incorporates useful economic restrictions from a structural model into a machine learning model through transfer learning. The core idea is to first construct a neural-network representation of the structural model, and then update the network using information from real data. In an example application to option pricing, our hybrid model significantly outperforms both the structural model and a conventional deep-learning model. The out-performance of the hybrid model is more significant when the sample size of real data is limited or under volatile market conditions.

We study how investors, firms, and information sellers interact in a market with manipulable information. To better predict the firm characteristics they care about, investors can buy a score from a monopolistic information seller, which aggregates signals that are subject to firm manipulation. The average degree of signal manipulability has no effect on the equilibrium, while the uncertainty about manipulability becomes a new source of noise. Its contribution depends on firms' incentive to manipulate the signals, which in turn depends on the equilibrium price sensitivity to the score. The optimal design of the score weighs signal precision against the endogenous uncertainty due to manipulation. The introduction of mandate investors, who care about the scores on the characteristics and not the characteristics themselves, generates an incentive for information sellers to inflate the scores. When applied to green investing, our model implies that the effectiveness of impact investing on the cost of capital could actually decline as the fraction of green investors or the strength of the mandate keeps rising, because they generate stronger incentives for manipulation.

We propose a measure of valuation gap between debt and equity, the debt-equity spread (DES), based on the difference between actual and equity-implied credit spreads. DES predicts the cross section of stock and bond returns in opposite directions, with stronger results among smaller, less liquid, and more difficult-to-short stocks and bonds, and the predictability cannot be explained by exposures to a variety of risk factors. Furthermore, high-DES firms tend to have more negative growth forecast revisions, are more likely to issue equity and retire debt, and have more insider equity selling. These findings on asset pricing dynamics and corporate financing behavior are consistent with DES capturing relative mispricing between debt and equity. They imply that segmentation between the two markets is prevalent at firm level.

We study how process-focused intangible capital affects firm investment and compensation. We document that, in the cross-section, investment and executive compensation are both increasing in process intangibility, defined as the share of process intangibles in total capital stock. To explain these patterns, we build a dynamic agency model with two key ingredients: (i) physical investment and process innovation are complementary in driving the growth of efficiency-adjusted capital; (ii) due to its opaque nature, process innovation is susceptible to moral hazard problems. The model not only accounts for the positive relations between process intangibility, investment, and compensation, but also predicts that the pay-process intangibility association will be stronger among firms with high physical investments. Finally, it implies that pay-equity measures that restrict the compensation of executives and skilled labor, especially those associated with process innovation, could depress investment and reduce firm value.

We introduce ``deep surrogates'' -- high-precision approximations of structural models based on deep neural networks, which speed up model evaluation and estimation by orders of magnitude and allow for various compute-intensive applications that were previously infeasible. As an application, we build a deep surrogate for a high-dimensional workhorse option pricing model. The surrogate enables us to re-estimate the model at high frequency to construct an option-implied tail risk measure, which is highly predictive of future market crashes. It also enables us to systematically examine the model's out-of-sample performances, which reveals the tradeoffs between structural and reduced-form approaches for option pricing. Moreover, we construct a measure for the degree of parameter instability and connect it to option market illiquidity in the data. Finally, we use the surrogate to construct conditional distributions of option returns, which is useful for risk management and provides a new way to test the model.

We analyze a model of optimal capital structure and liquidity choice based on a dynamic tradeoff theory for financially constrained firms. In addition to the classical tradeoff between the expected tax advantages of debt financing and bankruptcy costs, we introduce a cost of external financing for the firm, which generates a precautionary demand for cash and an optimal retained earnings policy for the firm. An important new cost of debt financing in this context is a \emph{debt servicing cost}: debt payments drain the firm's valuable precautionary cash holdings and thus impose higher expected external financing costs on the firm. Another change introduced by external financing costs is that realized earnings are separated in time from payouts to shareholders, implying that the classical Miller-formula for the net tax benefits of debt no longer holds. We offer a novel explanation for the "debt conservatism puzzle" by showing that financially constrained firms choose to limit their debt usages in order to preserve their cash holdings. In the presence of these servicing costs, a financially constrained firm may even choose not to exhaust its risk-free debt capacity. We also provide a valuation model for debt and equity in the presence of taxes and external financing costs and show that the classical adjusted present value methodology breaks down for financially constrained firms.

**Publications**

The fact that internal liquidity is a key source of corporate funding puts the marginal value of cash at the center of a variety of firm decisions, including investment, payout, financing, savings, and risk management policies. The marginal value of cash is inherently a dynamic concept, because a firm facing financing frictions has to be forward-looking, managing both its asset and liability structures in a unified framework and carefully trading off the use of liquidity across time and states. We present a dynamic framework for corporate liquidity management and survey the related literature, with a focus on the determinants of the marginal value of cash and its ubiquitous role in firm decisions.

Firms tend to compete more aggressively in financial distress; the intensified competition in turn reduces profit margins, pushing themselves further into distress and adversely affecting other firms. To study such feedback and contagion effects, we incorporate strategic competition into a dynamic model with long-term defaultable debt, which generates various peer interactions like predation and self-defense. The feedback effect imposes an additional source of financial distress costs incurred for raising leverage, which helps explain the negative profitability-leverage relation across industries. Owing to the contagion effect, in a decentralized equilibrium, leverage is excessively high from an industry perspective, compromising industry's financial stability.

Market-wide trading halts, also called circuit breakers, have been widely adopted as part of the stock market architecture, in the hope of stabilizing the market during dramatic price declines. We develop an intertemporal equilibrium model to examine how circuit breakers impact market behavior and welfare. We show that a circuit breaker tends to lower the overall level of the stock price and significantly alters its dynamics. In particular, as the price approaches the circuit breaker, its volatility rises drastically, accelerating the chance of triggering the circuit breaker -- the so-called ``magnet efect''; in addition, returns exhibit increasing negative skewness and positive drift, while trading activity spikes up. Our empirical analysis finds supportive evidence for the model's predictions. Moreover, we show that a circuit breaker can affect the overall welfare either negatively or positively, depending on the relative significance of investors' trading motives for risk sharing vs. irrational speculation.

We formalize the concept of ``dark matter'' in asset pricing models by quantifying additional information the econometrician can obtain about the fundamental dynamics from asset pricing cross-equation restrictions. The dark matter measure captures the degree of fragility for models that are potentially misspecified and unstable: a large dark matter measure signifies a model's lack of internal refutability (weak power of specification tests) and external validity (high overfitting tendency and poor out-of-sample fit). The measure can be computed at low cost even for complex dynamic structural models. We illustrate its applications via (time-varying) rare-disaster risk and long-run risk models.

We introduce the decision-aware time-series conditional generative adversarial network (DAT-CGAN), a method for the generation of time-series data that is designed to support decision-making. The framework adopts a multi-Wasserstein loss on decision-related quantities and an overlapped block-sampling approach for sample efficiency. We characterize the generalization properties of DAT-CGAN and in application to a multi-period portfolio choice problem and financial time series data, we demonstrate better training stability and generative quality in regard to both raw data and decision-related quantities than strong GAN-based baselines.

We provide causal evidence for the value of asset pledgeability. Our empirical strategy is based on a unique feature of the Chinese corporate bond markets, where bonds with identical fundamentals are simultaneously traded on two segmented markets with different rules for repo transactions. We utilize a policy shock on December 8, 2014, which rendered a class of AA+ and AA bonds ineligible for repo on one of the two markets. By comparing how bond prices changed across markets and rating classes around this event, we estimate that when the haircut increases from 0 to 100%, the bond yields increase in the range of 39 to 85 bps. These estimates help us infer the magnitude of the shadow cost of capital in China.

We document several facts about corporate debt maturity: (1) debt maturity is pro-cyclical; (2) higher-beta firms tend to have longer debt maturity; (3) shorter maturity amplifies the sensitivity of credit spreads to aggregate shocks. We build a dynamic capital structure model that explains these facts. In the model, leverage and maturity choices are highly interdependent, which reflects the tradeoffs of liquidity discounts of long-term debt, repayment risks of short-term debt, and the benefit of short-term debt as a commitment device for timely leverage adjustments. Additionally, the model quantifies the effects of maturity dynamics on the term structure of credit spreads.

Mortgage refinancing activity associated with extraction of home equity contains a strongly counter-cyclical component consistent with household demand for liquidity. We estimate a structural model of liquidity management featuring counter-cyclical idiosyncratic labor income uncertainty, both long-term and short-term mortgages, and realistic borrowing constraints. We then empirically evaluate its predictions for the households' choices of leverage, liquid assets, and mortgage refinancing using micro-level data. Taking the observed historical paths of house prices, aggregate income, and interest rates as given, the model quantitatively accounts for many salient features in the evolution of balance sheets and consumption in the cross section of households over the 2001-2012 period.

We propose a new measure of financial intermediary constraints based on how the intermediaries manage their tail risk exposures. Using a dataset for the trading activities in the market of deep out-of-the-money S&P 500 put options, we identify periods when the variations in the net amount of trading between financial intermediaries and public investors are likely to be mainly driven by shocks to intermediary constraints. We then infer tightness of intermediary constraints from the quantities of option trading during such periods. We show that a tightening of intermediary constraint according to our measure is associated with increasing option expensiveness, higher risk premia for a wide range of financial assets, deterioration in funding liquidity, and deleveraging of broker-dealers.

Full Text Internet Appendix Data: Option-based constraint measure

We develop a structural credit risk model to examine how the interactions between default and liquidity affect corporate bond pricing. The model features debt rollover and bond-price dependent holding costs for illiquid corporate bonds. Both over the business cycle and in the cross section (across ratings), our model does a good job matching the average default rates and credit spreads in the data, and it captures important variations in bid-ask spreads and bond-CDS spreads. A structural decomposition reveals that the default-liquidity interactions can account for 10% to 24% of the level of credit spreads and 16% to 46% of the changes in spreads over the business cycle. Through this decomposition, we estimate that liquidity-related corporate bond financing costs amount to about 6% of total issuance amount for the period of 1996 to 2015. We also apply our framework to evaluate the impact of liquidity-provision policies for the bond market.

Since corporate debt tends to be riskier in recessions, transfers from equity holders to debt holders that accompany corporate decisions also tend to concentrate in recessions. Such systematic risk exposures of debt overhang have important implications for corporate investment and financing decisions, and for the ex ante costs of debt overhang. Using a calibrated dynamic capital structure model, we show that the costs of debt overhang become higher in the presence of macroeconomic risk. We also provide several new predictions on how the cyclicality of a firm's assets in place and growth options affect its investment and capital structure decisions.

We study an investor's optimal consumption and portfolio choice problem when he is confronted with two possibly misspecified submodels of stock returns: one with IID returns and the other with predictability. We adopt a generalized recursive ambiguity model to accommodate the investor's aversion to model uncertainty. The investor deals with specification doubts by slanting his beliefs about submodels of returns pessimistically, causing his investment strategy to be more conservative than the Bayesian strategy. This effect is especially strong when the submodel with a low Bayesian probability delivers a much smaller continuation value. Unlike in the Bayesian framework, the hedging demand against model uncertainty may cause the investor's stock allocation to decrease sharply given a small doubt of return predictability, even though the predictive variable is large. Adopting the Bayesian strategy can lead to sizable welfare costs for an ambiguity-averse investor, especially when he has a strong prior of return predictability.

Firms face uncertain financing conditions, which can be quite severe as exemplified by the recent financial crisis. We capture the firm's precautionary cash hoarding and market timing motives in a tractable model of dynamic corporate financial management when external financing conditions are stochastic. Firms value financial slack and build cash reserves to mitigate financial constraints. The finitely-lived favorable financing condition induces them to rationally time the equity market. This market timing motive can cause investment to be decreasing (and the marginal value of cash to be increasing) in financial slack, and can lead a financially constrained firm to gamble. Quantitatively, we find that firms' optimal responses to the threat of a financial crisis can significantly smooth out the impact of financing shocks on investments, marginal values of cash, and the risk premium over time. Thus, a firm may still appear unconstrained based on its relatively smooth investment over time despite significant underinvestment. This smoothing effect can be used to disentangle financing shocks from productivity shocks empirically.

Risks of rare economic disasters can have large impact on asset prices. At the same time, difficulty in inference regarding both the likelihood and severity of disasters as well as agency problems can effectively lead to significant disagreements among investors about disaster risk. We show that such disagreements generate strong risk sharing motives, such that just a small amount of optimists in the economy can significantly reduce the disaster risk premium. Our model highlights the "latent" nature of disaster risk: the disaster risk premium will likely be low and smooth during normal times, but can increase dramatically when the risk sharing capacity of the optimists is reduced, for example, following a disaster. The model also helps reconcile the difference in the amount of disaster risk implied by financial markets and international macro data, and provides caution to the approach of extracting disaster probabilities from asset prices, which can disproportionately reflect the beliefs of a small group of optimists. Finally, our model predicts an inverse U-shaped relation between the equity premium and the size of the disaster insurance market.

Non-linearity is an important consideration in many problems of finance and economics, such as pricing securities and solving equilibrium models. This paper provides analytical treatment of a general class of nonlinear transforms for processes with tractable conditional characteristic functions, which extends existing results on characteristic function based transforms to a substantially wider class of nonlinear functions while maintaining low dimensionality by avoiding the need to compute the density function. We illustrate the applications of the generalized transform in pricing defaultable bonds with stochastic recovery. We also use the method to analytically solve a class of general equilibrium models with multiple goods and apply this model to study the effects of time-varying labor income risk on the equity premium.

We propose a model of dynamic investment, financing, and risk management for financially constrained firms. The model highlights the central importance of the endogenous marginal value of liquidity (cash and credit line) for corporate decisions. Our three main results are: 1) investment depends on the ratio of marginal q to the marginal value of liquidity, and the relation between investment and marginal q changes with the marginal source of funding; 2) optimal external financing and payout are characterized by an endogenous double-barrier policy for the firm's cash-capital ratio; and 3) liquidity management and derivatives hedging are complementary risk management tools.

We develop a dynamic incomplete-markets model of entrepreneurial firms, and demonstrate the implications of nondiversifiable risks for entrepreneurs' interdependent consumption, portfolio allocation, financing, investment, and business exit decisions. We characterize the optimal capital structure via a generalized tradeoff model where risky debt provides significant diversification benefits. Nondiversifiable risks have several important implications: more risk-averse entrepreneurs default earlier, but choose higher leverage; lack of diversification causes entrepreneurial firms to underinvest relative to public firms, and risky debt partially alleviates this problem; entrepreneurial risk aversion can overturn the risk-shifting incentives induced by risky debt. We also analytically characterize the idiosyncratic risk premium.

I build a dynamic capital structure model that demonstrates how business cycle variation in expected growth rates, economic uncertainty, and risk premia influence firms' financing policies. Countercyclical fluctuations in risk prices, default probabilities, and default losses arise endogenously through firms' responses to macroeconomic conditions. These comovements generate large credit risk premia for investment grade firms, which helps address the credit spread puzzle and the under-leverage puzzle in a unified framework. The model generates interesting dynamics for financing and defaults, including market timing in debt issuance and credit contagion. It also provides a novel procedure to estimate state-dependent default losses.