Paris December 2017

This paper investigates the volatility-of-volatility (VVIX) term structure. The
term structure is in nearly all cases downward sloping. We find that the slope
of the VVIX (SlopeVVIX), defined as VVIX' second principal component, predicts
excess returns of S&P500 and VIX straddles. Its informational content is
incremental to the VIX term structure and to the variance risk premium. To
analyze which type of risk is captured by SlopeVVIX, we develop an affine approximation
for the VVIX based on a VIX option pricing model. We find that
the contributions of continuous vol-of-vol and jump risk to the term structure
vary systematically with the state of the economy. When the latest major
crises hit, continuous vol-of-vol took the lion's share over all maturities. As its
relative contribution is similarly high across all maturities at these times, our
results suggest that investors expected a prolonged period of high uncertainty
in volatility.

This paper investigates the effect of an exogenous credit supply shock triggered by a change in accounting norms on firm accumulation of human capital. In 2005, the introduction of new reporting norms for defined-benefit pension plans in Portugal led to large increases in the value of bank pension liabilities. Affected banks increased both direct contributions to pension plans and prudential deductions from Tier 1 capital, subsequently reducing their supply of credit. We first document that firms in a relationship with affected banks do not perfectly substitute credit. Second, we find that affected firms reduce employment. We show that these employment effects are stronger among unskilled workers, but also among the highly educated ones. Workers holding a college degree, or holding complex occupations, are more likely to leave affected firms. These results suggest that credit supply shocks can affect firm accumulation of human capital, with implications for firm long-term productivity.

U.S. corporations headquartered in states with greater public corruption are prone to more unethical behavior, reflective of a state-level "culture-of-corruption". We test for state-level differences by exploiting passage of Foreign Corrupt Practices Act (FCPA) that curtailed bribery of foreign officials. Firms in corrupt states, especially firms trading with more corrupt countries, suffer greater value (Tobin's Q) and performance (ROA) decline following FCPA, indicating larger losses from restrictions on bribery. Culture-of-corruption is also manifest in greater agency problems: Firms in corrupt states are more likely to manage earnings, face securities fraud litigation and be adversely affected by state-level anti-takeover laws.

To facilitate price discovery, Euronext Paris has always relied on a transparent pre-opening phase and on a call auction to open continuous markets. Fast trading, competition from alternative trading venues and the poor volume at the open (2%) however question the role of these non-trading sessions. Using a unique dataset of stocks cross-traded on Euronext Paris, BATS and Chi-X, we explore the behavior of traders during the preopen based on their speed and nature of orders (proprietary, agency or market-making). We show that slow brokers submit orders very early, and most of them are executed within the day. In contrast, fast prop traders or dedicated liquidity providers only participate in the last half-hour. Interestingly the pre-opening activity of slow brokers is strongly related to the price discovery process across trading venues. Finally, we show that although tentative clearing prices of the preopen contain information, they are followed by a reversal in the following 15 minutes across the different platforms, reflecting price pressure and liquidity issues around the open.

A flat Securities Market Line is not evidence against the CAPM. Under the Roll (1977) critique, the CAPM is a “lost city of Atlantis,” empirically invisible. In a noisy rational-expectations economy, there exists an information gap between the average investor who holds the market and the empiricist who does not observe the market portfolio. The CAPM holds for the investor, but appears flat to the empiricist. This distortion is empirically substantial and explains, for instance, why “Betting Against Beta” works; BAB really bets on true beta. Macroeconomic announcements reduce the distortion—for a fleeting moment the empiricist catches a glimpse of the CAPM.

Governments incentivize retirement saving by allowing individuals to contribute to
tax-advantaged accounts where the returns to financial assets receive special tax treat-
ment. In accounts with "back-loaded" taxation, the individual contributes pretax money
and pays taxes when the money is withdrawn. In accounts with "front-loaded" taxa-
tion, the individual contributes aftertax money and pays no future taxes. Under some
simplifying assumptions, a standard benchmark result is that both the individual and
the government are indifferent between the two types of accounts. We add investment
management fees to the benchmark model and show that the neutrality result breaks
down. Assuming fees are fixed as a percent of assets under management (AUM), we
show that individuals are still indifferent to the timing of taxation but the government
is not. Under back-loaded taxation, the government implicitly owns a share of all re-
tirement accounts and is effectively paying investment fees on this share, something
it avoids under front-loaded taxation. We estimate this to cost the government $15
billion per year, representing a subsidy to the asset management industry. We then ask
whether this result holds in general equilibrium, where fees as a percent of AUM are
allowed to vary. The answer depends both on the nature of the cost function for asset
management services, and on the nature of market competition, but we find that the
result will in general continue to hold: back-loaded taxation is more expensive for the
government and produces a larger asset-management industry. Finally, we use the gen-
eral equilibrium model to examine welfare implications. In a rough calibration of the
model, we find that this increase in the size of the asset management industry reduces
consumer welfare.

Witnessing stock market history in the making leaves behind a vivid story, but does not provide valuable information. Nevertheless, well-versed finance professionals extrapolate from witnessed returns when forming beliefs about expected returns which we show by using a unique dataset regarding professionals' career start in the finance industry. This result is robust to controlling for all publicly available information and interpersonal differences. Additionally, we find that returns witnessed early on in a career are more formative than those witnessed recently. Among the potential channels through which witnessed returns might affect professionals' expectations, a judgmental bias appears the most plausible.

Short selling enhances the price feedback mechanism for corporate investment decisions. Using the staggered introduction and repeal of short selling restrictions around the world, we show that the sensitivity of investment to Tobin's Q is 25% higher when short selling is permitted. Further, short selling strengthens the link between the completion of announced M&A deals and the price reaction on announcement day. The positive impact of short selling on investment-Q sensitivity is higher for firms with low analyst coverage and institutional ownership, mitigated for cross-listed firms, and subsequently results in improved firm performance. The effects do not appear to be driven by a higher total amount of information in prices, faster price discovery, improved firm disclosure, or differences in governance. Our results suggest that short selling incentivizes traders to acquire private information new to firm managers, thus enhancing managers' reliance on stock prices as an information signal for investment decisions.

We exploit detailed transaction and position data for a sample of long-short equity hedge funds to document new facts about the trading activity of sophisticated investors. We find that the initiation of both long and short positions is associated with significant abnormal returns, suggesting that the hedge funds in our sample possess investment skill. In contrast, the closing of long and short positions is followed by return continuation, implying that hedge funds close their positions too early and leave money on the table. As we demonstrate with a simple model, this behaviour can be explained by hedge funds being (risk) capital constrained and facing position monitoring costs. Consistent with our model, we document that the return continuation following closing orders is more pronounced when these constraints become more binding (e.g., after negative fund returns or increases in volatility).

We develop a model that explains the average return and volatility spread between sin stocks and non-sin comparable stocks. We endow agents' preferences with a sensitivity factor to firms' ethicalness. A positive marginal rate of substitution between dividends and ethicalness explains the higher average returns that sin stocks exhibit over non-sin comparable stocks. This result can be obtained either when (i) dividends and ethicalness are substitute goods and investors are less riskaverse than log utility, or (ii) when dividends and ethicalness are complementary goods and investors are more risk-averse than log utility. We empirically show that only the latter case can explain the patterns of the conditional return and volatility spreads between sin and non-sin comparable stocks.

We show that peer firms play an important role in shaping corporate earnings management decisions. To overcome identification issues in isolating peer effects, we use fund flow-induced selling pressure by passive open-end equity mutual funds as exogenous shocks to firms' stock prices. Managers respond to such exogenous price shocks by adjusting earnings management policies. We then measure individual firms' reactions to changes in earnings management at peer firms as a result of such exogenous price shocks. The documented peer effect in earnings management is not only statistically, but also economically significant. Our results are robust to alternative measures of fund flow-induced selling pressure and earnings management, and to estimating instrumental variables regressions in which we instrument peer firms' earnings management with mutual fund flow-induced selling pressure.

A number of competing explanations have been offered as a rationale for the trend in idiosyncratic variance that has been experienced over the past four decades. We establish a theoretical model linking a market microstructure bias with the industry-adjusted idiosyncratic variance (Campbell, Lettau, Malkiel, and Xu, 2001) or the risk-adjusted idiosyncratic variance. Using this model's predictions, we empirically show that the bid-ask spread eliminates the time trend in aggregate idiosyncratic variance. These results are robust across various exchanges, across various risk-based measures of idiosyncratic variance, and through time. Two natural experiments demonstrate that an exogenous shock to the bid-ask spread is associated with a subsequent decline in the aggregate idiosyncratic variance. The microstructure hypothesis dominates any of the alternative explanations, including uncertainty about profitability, earnings shocks, or growth options, for the trend in idiosyncratic variance.

I study market-implied real yields extracted from prices of inflation-linked government bonds for nine developed countries. The liquidity premium is an important component of breakeven inflation rates. Unconditional real yield curves are upward-sloping, providing empirical support for habit models. The cross-country real rate equality is rejected. Across countries, real yields are strongly positively correlated while liquidity premia are moderately positively correlated. Low nominal yields following the Great Recession are mainly due to low real yields, although the inflation risk premia have also decreased.

I develop a dynamic search model of the housing market where house prices are determined in auctions rather than by Nash bargaining as in the housing search model from the literature. The model with auctions generates fluctuations between booms and busts. During the boom multiple buyers compete for one house, while in the bust buyers are choosing among several houses. The model improves on the performance of the model with Nash bargaining by producing highly volatile house prices which helps to solve the puzzle of excess volatility of house prices. Higher volatility arises because of the competition between buyers with heterogeneous values. With heterogeneous valuations, the price determinations becomes important for the quantitative properties of the model. With Nash bargaining, the buyer is chosen randomly among all interested buyers. Then the average of buyers' house values determines the house price. In the auction model, the buyer is chosen by the maximum bid among all interested buyers, so the highest value determines the house prices. During the boom, the highest values increase more than the average values, making the sales price more volatile. This high volatility is constrained efficient since the equilibrium in the model decentralizes the solution of the social planner problem, constrained by the search frictions.

We examine the effect of bond price transparency on the information content of credit ratings. To identify the causal impact of transparency, we use a recent regulation in the U.S. that mandates real-time public dissemination (or disclosure) of over-the-counter transactions in corporate debt securities via the Trade Reporting and Compliance Engine (TRACE). We find that dissemination dramatically reduces the average short-term stock and bond price impact of rating downgrades, suggesting that ratings fill information gaps arising from lack of price transparency. The reduction does not arise from changes in investor confidence in, or the ex post quality of, issuer-paid ratings. Dissemination matters less where the issuer is a CDS reference entity or has outstanding bonds traded on an exchange or where stock analysts issue more reliable forecasts of the issuer's future earnings. We also find that rating downgrades become more sensitive to changes in market-based measures of credit risk after dissemination, consistent with transparent prices making rating inflation more transparent, and therefore increasing the incentives for rating agencies to update ratings in a more timely manner. Finally, we document that credit spreads better predict future defaults after dissemination, consistent with more efficient information aggregation in transparent markets.

This study presents evidence consistent with incomplete contracting theory in the relation-
ship between renegotiation costs and the initial terms of debt contracts. We exploit plausibly
exogenous variation in the taxation of U.S. syndicated loan renegotiation and isolate the ef-
fect of changes in renegotiation costs on initial contract terms. We conjecture and find that
if contracts are incomplete, then as renegotiation costs fall the maturity of debt contracts
lengthens, the likelihood of covenant violation increases, and performance pricing provisions
become less common. We also expect and find that this tax change makes private debt more
attractive to lenders and borrowers.

Interest rate variance risk premium (IRVRP), the difference between implied and realized variances of interest rates, emerges as a strong predictor of Treasury bond returns of maturities ranging between one and ten years for return horizons up to six months. IRVRP is not subsumed by other predictors such as forward rate spread or equity variance risk premium. These results are robust in a number of dimensions. We rationalize our findings within a consumption-based model with long-run risk, economic uncertainty, and inflation non-neutrality. In the model interest rate variance risk premium is related to short-run risk only, while standard forward-rate-based factors are associated with both short-run and long-run risks in the economy. Our model qualitatively replicates the predictability pattern of IRVRP for bond returns.

Using novel data on 1,240 credit agreements for large corporate loans, we show that while inclusion of negative covenants that restrict new debt issuance, payments, asset sales, affiliate transactions and investments is widespread, clauses that weaken these restrictions are almost as common. We measure the deductions for the core covenants in terms of their potential impact on overall leverage and show that they are large, and concentrated in already highly levered transactions. We analyze the cross-sectional variation in contractual weaknesses introduced through deductions and exclusions to negative covenants and show that such contractual provisions are characteristic of leveraged buyouts.