Alan Moreira
Associate Professor of Finance
email: alan.moreira at nyu.edu
Expertise: Asset Pricing, Financial Intermediation
email: alan.moreira at nyu.edu
Expertise: Asset Pricing, Financial Intermediation
Currently, I'm an associate professor of finance at New York University Stern School of Business and an NBER Research Fellow. My research focus on asset pricing, financial intermediation, and macro-finance. Previously I was associate professor of finance at University of Rochester and an assistant professor of finance at Yale University. I did my PhD in Financial Economics at the University of Chicago, graduating in 2011, and my undergraduate and master studies back in Brazil at the Rio de Janeiro Federal University and PUC-RJ respectively.
We extend back to 1890 the volatility implied by options index (VIX), available only since 1986, using the frequency of words on the front-page of the Wall Street Journal. News implied volatility (NVIX) captures well the disaster concerns of the average investor over this longer history. NVIX is particularly high during stock market crashes, times of policy-related uncertainty, world wars and financial crises. We find that periods when people are more concerned with a rare disaster, as proxied by news, are either followed by periods of above average stock returns, or followed by periods of large economic disasters. We estimate that the disaster probability has a half-life of four to eight months and annual volatility of 4% to 6%. Our findings are consistent with the view that hard to measure time-varying rare disaster risk is an important driver behind asset prices.
We build a macro-finance model in which intermediaries issue equity without friction. In normal times, they maximize liquidity creation by levering up the collateral value of their assets, a process we call shadow banking. A rise in uncertainty causes investors to demand liquidity in bad states, which forces intermediaries to delever and substitute toward safe liabilities; shadow banking shuts down, prices and investment fall. The model is consistent with a slow economic recovery especially when intermediary capital is high. It features collateral runs and flight to quality, and it provides a framework for analyzing unconventional monetary policy and regulatory reform proposals.
Managed portfolios that take less risk when volatility is high produce large alphas, substantially increase factor Sharpe ratios, and produce large utility gains for mean- variance investors. We document this for the market, value, momentum, profitability, return on equity, and investment factors in equities, as well as the currency carry trade. Volatility timing increases Sharpe ratios because changes in factors’ volatilities are not fully offset by proportional changes in average returns. Our strategy is contrary to conventional wisdom because it takes relatively less risk in recessions and crises yet still earns high average returns. This rules out typical risk-based explanations and is a challenge to structural models of time-varying expected returns.
We study the portfolio choice decision of a long-horizon investor when volatility and expected return dynamics are estimated using US data. Investors should reduce exposure when volatility increases, and ignoring variation in volatility leads to large utility losses. Volatility timing is more beneficial than expected return timing, par- ticularly when parameter uncertainty is considered. We show that longer-horizon investors should engage in less volatility timing if increases in volatility also increase the amount of mean-reversion in returns. Nevertheless, we provide new empirical evidence that this channel is not strong enough to substantially decrease the gains from volatility timing.
In this paper, I build a model where financial intermediation slows the flow of capital. Investors optimally learn from intermediary performance to allocate capital toward profitable intermediaries. Intermediaries reach for yield, i.e invest in high tail risk assets, in an attempt to drive flows and reduce liquidation risk. Reaching for yield is stronger among intermediaries with weak opportunities, resulting in a reduction in the informativeness of performance; investors take longer to learn and capital flows become less responsive to performance. Capital becomes slow moving because the reach for yield dampens learning. The model predicts capital immobility to be stronger when tail risk is high; when tail risk is underpriced; and in asset classes with large cross-sectional variation in tail risk exposures.
Text data is inherently ultra-high dimensional, which makes machine learning techniques indispensable for textual analysis. Text also tends to be a highly selected outcome—journalists, speechwriters, and others carefully craft messages to target the limited attention of their audiences. We develop an economically motivated high dimensional selection model that improves machine learning from text (and from sparse counts data more generally). Our model is especially useful in cases where the cover/no-cover choice is separate or more interesting than the coverage quantity choice. Our design allows for parallel estimation, making the model highly computationally scalable. We apply our framework to backcast, nowcast, and forecast financial variables using newspaper text, and find that it substantially improves out-of-sample fit relative to alternative state-of-the-art approaches.
At the announcement of a new policy, agents form a view of state-contingent policy actions and impact. We develop a method to estimate this state-contingent perception and implement it for many asset-purchase interventions worldwide. Expectations of larger support in bad states—“policy puts”—explain a large fraction of the announcements’ impact. For example, when the Fed introduced purchases of corporate bonds in March 2020, markets expected five times more price support had conditions worsened relative to the median scenario. Perceived promises of additional support in bad states alter asset prices, risk, and the response to future announcements.
Liquidity provision is a bet against private information: if private information turns out to be higher than expected, liquidity providers lose. Since information generates volatility, and volatility co-moves across assets, liquidity providers have a negative exposure to aggregate volatility shocks. As aggregate volatility shocks carry a very large premium in option markets, this negative exposure can explain why liquidity provision earns high average returns. We show this by incorporating uncertainty about the amount of private information into an otherwise standard model. We test the model in the cross section of short-term reversals, which mimic the portfolios of liquidity providers. As predicted by the model, reversals have large negative betas to aggregate volatility shocks. These betas explain their average returns with the same risk price as in option markets, and their predictability by VIX in the time series. Volatility risk thus explains the liquidity premium among stocks and why it increases in volatile times. Our results provide a novel view of the risks and returns to liquidity provision.
Standard risk factors can be hedged with minimal reduction in average return. This is true for ``macro'' factors such as industrial production, unemployment, and credit spreads, as well as for ``reduced form'' asset pricing factors such as value, momentum, or profitability. Low beta versions of the factors perform close to as well as high beta versions, hence a long short portfolio can hedge factor exposure with little reduction in expected return. For the reduced form factors this mismatch between factor exposure and expected return generates large alphas. For the macroeconomic factors, hedging the factors also hedges business cycle risk by significantly lowering exposure to consumption, GDP, and NBER recessions. We study implications both for optimal portfolio formation and for understanding the economic mechanisms for generating equity risk premiums.
We argue that quantitative easing (QE) and tightening policies constitute a dynamic state-contingent plan instead of a succession of independent interventions. This view changes the main reason QE is effective by adding an insurance channel to the static effect of absorbing bond supply in a given period. QE purchases occur in bad economic states (e.g., 2008-2009 or 2020) when the supply of government debt increases. Increasing long-term bond prices in bad economic states increases their safety, driving up their value and thus lowering ex-ante yields. We estimate that this insurance channel alone lowers long-term bond yields by 75-100 bps. This channel explains the prevalence of low long-term yields, low term premia, and low yield volatility since the introduction of QE, despite the sharp increase in net government debt supply. Consistent with a state-contingent channel, implied volatilities of long-duration risk-free securities fall substantially on QE announcements, even for options with maturities out to 10 years. We calibrate a policy rule for asset purchases to their historical path and include it in a quantitative term structure model. In the model, state-contingent QE offsets term premia fluctuations in long-term bonds. The insurance effect from this channel lowers long-term Treasury yields by 75bps ex-ante, which explains about 75% of the total effect of QE on yields. The calibrated model matches both broad patterns in bond yields and the response to QE announcements.
Many hedge funds restrict investors' ability to redeem their investments. We show that lockups alleviate a delegation friction. In our model hedge funds can enter a long-term trade that increases expected returns but lowers short-term returns. Investors who rationally learn from returns may mistake a skilled manager who pursues the long-term trade for an unskilled manager. Skilled managers therefore have an incentive to avoid the long-term trade to enhance short-term returns. The tradeoff between the benefits of the long-term trade and investors' fears of being stuck with an unskilled manager determines the optimal lockup. We calibrate the model to hedge fund data and show that arbitrage remains limited even with optimal lockups; the average manager sacrifices 146 basis points in expected returns per year to improve short-term returns.
This subsumes an earlier working paper called "Limits to Arbitrage and Lockup Maturities".
We study the problem of an investor that allocates analysts to assets to learn about future asset values. We show that when analysts are better at relative rather than absolute asset valuations the optimal matching of analysts to assets displays a balancedness property in which pairs of distinct assets are covered by a similar number of analysts. A balanced allocation allows the investor to efficiently aggregate information using the relative value between assets, eliminating the effect of the analyst-specific component. We show that the optimal matching of analysts to assets and the optimal portfolio decision depends on the structure of the analyst coverage network - the bipartite graph where the vertices are the firms and the edges are all the pairs of distinct firms that are covered by at least one common analyst. For example, capital is only reallocated between firms that are connected in the network, and the intensity of the reallocations depends on both the value of relative asset recommendations and the strength of the connection between the assets.