Working papers:
Skewness and networks
+Info
We establish four empirical facts or regularities:
1. Economic activity is skewed left in both levels and growth rates
2. Skewness is greater at the aggregate than the sector level
3. The cross-sectional volatility of sector growth rates is countercyclical
4. Sectors become more central following negative shocks
Those facts are naturally generated in a production network that features complementarity across inputs.
Three leading alternative models -- skewed aggregate shocks, concave decision rules, and fixed adjustment costs with uncertainty shocks -- fail to match all four facts.
+Info+Data
This paper provides the first time-series of option-implied conditional skewness at the firm and market levels. Firm skewness is significantly procyclical and leads the cycle. Market skewness is acyclical. Firm skew and aggregate volatility are closely linked, implying they are driven by a common factor.
This data is from the current draft of the paper and is subject to revision until final publication.
Version date: 5/31/2022.
- Monthly firm-level, market-level, and idiosyncratic skewness, 1980–2021: Data
+Info
We show how state prices and risk preferences can be inferred from returns on synthetic options, which are just dynamic portfolios on the market and risk-free asset. While exchange-traded options earn significantly negative alphas, implying risk aversion rises as wealth falls, synthetic options have zero or positive alphas, implying roughly constant risk aversion. Over the last two decades, returns on exchange-traded options have converged (up) to those on synthetic options, consistent with a model in which prices are driven by intermediary frictions that have shrunk over time.
+Info
This paper describes the response of the economy to large shocks in a nonlinear production network. A sector's tail centrality measures how a large negative shock transmits to GDP. In a benchmark case, it is measured as a sector's average downstream closeness to final production. Increases in interconnectedness can simultaneously reduce the sensitivity of the economy to small shocks while increasing the sensitivity to large shocks. Tail risk is related to conditional granularity, where some sectors become highly influential following negative shocks.
Uncertainty and volatility
+Info
We give a general formula for the dynamics of beliefs -- the moments of agents' conditional distributions -- for very general dynamic processes for fundamentals. The implied belief dynamics fit the behavior of US aggregate stock returns surprisingly well, including the joint behavior of prices, volatility, and skewness. The model additionally shows that long memory in volatility should be pervasive, and explains why.
+Info
This is a review article. It discusses recent research on uncertainty in financial markets and the real economy. The three major areas of focus are the term structure of uncertainty, time-variation in the variance risk premium -- including evidence that it has been declining since 2010 -- and the behavior of conditional skewness. We also present novel data on commodity options in the 1920's and crash cliquets in recent times.
+Info
+Data
We use options on individual stocks to construct a measure of cross-sectional (as opposed to aggregate) uncertainty back to 1980. The index is much less variable than aggregate uncertainty, it is acyclical and it does not forecast aggregate output, all of which are inconsistent with recent models emphasizing the importance of uncertainty shocks. The results imply firm-level uncertainty is not a major "headwind" policymakers should worry about.
+Info
+Data
+Online appendix
We construct option portfolios that directly hedge uncertainty about the macroeconomy, as opposed to the financial sector (e.g. the VIX). Those portfolios historically earn positive average returns,
which is inconsistent with the view that uncertainty shocks are "bad" in the sense of being associated with high marginal utility.
- Archive with returns on straddles and strangles for each market plus at-the-money IVs
+Info
Uncertainty shocks can be identified as news about second moments. Our identified uncertainty shocks are not associated with recessions. The realization of volatility, however, as opposed to its expectation, is.
+Data
- Replication files
- Data for replicating just the main VAR
- Data on monthly S&P 500 implied volatility (VIX) extended to 1983 (daily is available on request)
The price of variance risk With Stefano Giglio, Anh Le, and Marius Rodriguez. Journal of Financial Economics, 2017. Lead article.
+Info
The only significantly priced risk in the variance market over the period 1996–2014 was transitory realized variance. News about future variance was unpriced, suggesting it is not an important driver of the real economy. The results allow us to distinguish among consumption-based models.
+Data
Information and investment across frequencies
On the effects of restricting short-term investment With Nicolas Crouzet and Charles G. Nathanson. Review of Financial Studies, 2019. Lead article and editor's choice.
+Info
We analyze restrictions on short-term investment in a noisy rational expectations model of a futures market. In our simple frictionless setting, the restriction has zero effect on the information content of prices at lower frequencies. While the entrance of high-frequency investors hurts those who invest at lower frequencies, restricting high-frequency investment does not restore the old equilibrium – in fact it makes the situation even worse.
+Info
Allowing people to choose what aspect of income dynamics (i.e. what autocovariances) to learn about has important effects for consumption. Generates excess sensitivity to transitory shocks but consumption growth that is white noise in the long run. People's approximate models for income are most likely to be accurate at low frequencies.
+Info
The long-run risk model (a small, persistent component in consumption growth) is the natural model for investors to use for asset pricing if they are unsure of the true dynamics of the economy.
+Info
+Vox EU column
The price of risk for a shock depends on its dynamic effects on the economy. We derive the relationship between risk prices and dynamic impacts in a range of theoretical models and also estimate it empirically.
+Info
+Replication files
I develop a new long-run variance estimator and use it to estimate the long-run variance of consumption growth. Point estimates are lower than standard long-run risks calibrations, but the more conservative calibrations cannot be ruled out. The estimates are useful more generally for calibrating models with recursive preferences.
Business cycles
+Info
+Replication files
A New-Keynesian model with time-varying risk aversion can fit bond yields nearly as well as an unrestricted three-factor model. Including bond prices in the estimation makes investment technology shocks look much less important.
+NBER WP version
+Vox EU column
+French translation
Older Unpublished Papers:
+Info
I combine Epstein–Zin preferences with habit formation. With production, the model generates a large and volatile equity premium.
+Info
+Replication files
First-order approximations related to perturbation that allow for time-varying risk aversion and volatility. Useful for macro-finance models.
+Vox EU column
Data sets and replication files:
Firm-level, market-level, and idiosyncratic skewness from "Real-time forward-looking skewness over the business cycle"
The file contains monthly option-implied skewness. At the firm level it is the value-weighted average across firms, and at the market level it is for the S&P 500.
This data is from the current draft of the paper and is subject to revision until final publication.
Version date: 5/31/2022
Aggregate and cross-sectional uncertainty from "Cross-sectional uncertainty and the business cycle: Evidence from 40 years of options data"
The files contain historical data for option implied volatility for the S&P 500, averaged across firms (weighted by market capitalization), and the measure of cross-sectional or firm-specific
implied volatility described in the paper.
Returns on straddles in 19 financial and commodity option markets from "Hedging macroeconomic and financial uncertainty and volatility"
The file contains two-week returns on delta-neutral straddles as decribed in the paper. Monthly at-the-money implied volatilities for the markets are also included
S&P 500 VIX extended back to 1983; from "Uncertainty shocks as second-moment news shocks"
The file contains monthly data for one- and six-month implied volatilities, calculated using the VIX-type formula, for 1983–2014. Daily data is available on request.
- Full replication files for the paper
- Data for replicating just the main VAR
- Monthly data for one- and six-month implied volatilities, calculated using the VIX-type formula, for 1983–2014
Variance swap prices from "The price of variance risk"
The file contains monthly prices for S&P 500 variance swaps for 12/1995–9/2013
Summary:
Stabilisation policy should focus on the frequencies consumers care most about. This column presents evidence from stock-market returns suggesting that consumers are willing to pay the most to avoid – and are therefore most concerned about – fluctuations that last tens or hundreds of years. Modern macroeconomic theory tends to view the role of monetary policy as smoothing out inflation and unemployment over the business cycle. The authors' findings suggest that resources would be better spent on policies that smooth out longer-run fluctuations.
Abstract:
I review the history of executive compensation disclosure and other government policies affecting CEO pay. In so doing, I also review the literature on the effects of these policies. Disclosure has increased nearly uniformly since 1933. A number of other regulations, including special taxes on CEO pay and rules regarding votes on some pay packages have also been introduced, particularly in the last 20 years. However, there is little solid evidence that any of these policies have had any substantial impact on pay. We can conclude that policy changes have helped drive the move towards more use of stock options, but there is no conclusive evidence on how policy has affected the level or composition of pay otherwise. I also review evidence from overseas on "Say on Pay," recently proposed in the US, which would allow nonbinding shareholder votes on CEO compensation. The experiences of other countries have been positive, with tighter linkages between pay and performance and improved communication with investors. Mandatory say on pay would be beneficial in the US.
Summary:
Only the top 10% of US earners have seen their incomes grow faster than productivity since 1966. Part of the top-earner income growth is driven by market forces (superstar economics); the only feasible pro-equality policy here is more progressive taxation. For top corporate executives, however, non-market forces (CEO-Board complicity in pay setting) are important, so other policies are warranted. Increased disclosure and improved corporate governance would distribute economic gains more evenly across society and boost firms' value.
Summary:
Europe's jobs outlook has brightened over the past decade. Recent research suggests that about half the rise in job creation is due to labour market reforms, but much of the rest is due to changing social norms concerning female and immigrant labour force participation. But what's good for European job creation seems to be bad for labour productivity growth – a trade-off that European policymakers must be willing to acknowledge and address.