My research broadly addresses the behavior and impact (intended or unintended) of institutional investors. Institutional investors are an important class of investors for several reasons. First, institutional investors control an increasing majority of financial assets. Mutual funds, which hold individual’s assets, are ultimately controlled by the mutual fund manager, who is a professional. Hedge funds, insurance companies, and pension funds — all of these are run by institutional investors. By way of contrast, an individual trading stocks at home with a personal brokerage account is not an institutional investor.

More specifically, my research investigates institutional investors and their effect on price efficiency in financial markets. This agenda stems from a paradox, described in Stein (2009):

1. Institutional investors are supposed to be sophisticated.

2. Institutional investors are growing as a % of market participants.

3. Thus, more sophisticated investors should imply better, more efficient pricing (i.e. less noise, irrational behavior). Yet,

4. We do not observe this. Why?


There are indeed many people addressing this puzzle, and the ‘answer’ is not likely to be singular. But this is the unifying thread of my research. In particular, my research falls into the following four groups under this broader umbrella. While these groups are a helpful categorization, they are not necessarily distinct.

Do constraints on short selling (either via regulation or cost to borrow) generate inefficient prices? In this line of research, I argue that focusing on a single market is too narrow (Blocher, Reed, and Van Wesep 2013, J. Financ. Econ.). Much like the mortgage market is a factor in house prices (lower interest rate mortgages support higher home prices), the securities lending market plays a role in stock prices. The securities lending market is a place where anyone who wants to sell a stock short must borrow the stock. The price to borrow a stock is set in the marketplace and, as such, is a function of supply and demand. As demand to short outstrips supply, spikes in the cost to borrow in the securities lending market inhibit short sellers.

My paper Supply side short-selling constraints: who is buying when shorts are selling focuses on the supply side in the securities lending market. Short sellers are sophisticated and usually have negative information about a stock. Why buy from them? We show that buyers lend those purchased shares at a rate significantly less than normal, generating a kind of endogenous short-selling constraint. Why? Because they are optimistic and do not want to enable short sellers. Those same stocks become more lottery-like, or exhibit positive skewness.

My paper Two-Sided Markets in Asset Management: Exchange-traded Funds and Securities Lending with Bob Whaley also investigates the supply side of the securities lending market. Here, we show that Exchange-traded Funds (ETFs) operate a two-sided market, much like credit cards. Credit card company have to navigate two markets: consumers (who they need to use the card) and retailers (who need to accept the card for payment). Both are charged a positive or negative fee (i.e. ‘cash back’ programs with credit cards). ETF’s two markets are investors and securities lenders. ETFs need to charge low enough fees to gather substantial assets, which they then turn and lend (at some fee) to stock borrowers. We implement a test by Rochet and Tirole (2006) to show that ETFs operate in a two-sided market and show that ETFs can sometimes make substantially more in securities lending than they make in management fees.

I also have a paper with Matt Ringgenberg called The Limits to (Short) Arbitrage, in which we show that short selling is actually pretty hard. Unconditionally, it isn’t, but conditional on actually wanting to short a stock, it can be quite hard to do so. This is because of the lack of stock options, which play an important role in alleviating short selling constraints when stocks are expensive to borrow. We show that stocks without options become expensive to borrow and stay that way – they are effectively unshortable. Those with options have time-varying lending fees.

Overall, this strand of research focuses on the securities lending market more than short sellers. Another way of saying this is that I focus on the ‘supply side’ of the market (beneficial owners and stock lenders) rather than short sellers, who can be seen as the ‘demand side’ of the securities lending market.

What is the role of management? How do the incentives of mutual funds, exchange-traded funds, hedge funds, and commodity trading advisors (CTAs) fit into this? Do they deliver alpha? If so, how do they do it? How much do we have to worry about principal-agent problems or other conflicts between managers and investors?

There is a great range here from passive, index-tracking exchange-traded funds to hedge funds and CTAs, the ultimate active investments. Do these operate simply on a spectrum, or are they doing really different things?

First off, what do investors want? Berk and Van Binsbergen (2015) and Barber, Huang, and Odean (2016) show that mutual fund investors use the CAPM to evaluate fund performance. I show in my paper The Revealed Preference of Sophisticated Investors that hedge fund investors do also. This was predicted by Berk and Van Binsbergen but is not obvious. Sophisticated investors easily could use more sophisticated models, which abound. But this result deepens the CAPM puzzle: how can investors use the CAPM but the CAPM not explain the cross-section of stock returns?

My paper Two-Sided Markets is about the incentives and behavior of Exchange-traded funds (ETFs). We show that operate a two-sided market model, balancing investor demands (efficient index tracking) with profits from securities lending. Essentially, ETF investors are not the customer, they are the product being sold to the securities lenders.

Network Externalities in Mutual Funds is about how mutual fund managers may get increased (abnormal) performance due to flow-based effects into peer funds (defined as funds with similar holdings). The key here is that these abnormal returns subsequently reverse because they are not based in fundamentals.

In Benchmarking Commodity Returns, we create a Fama-French-like factor model to evaluate commodity funds. Since commodities are a growing asset class, this seems like an important gap we fill in the literature.

We show in Risk shifting or just risk-adjusted returns that existing results establishing risk shifting in hedge funds and commodity trading advisors (CTAs) may be spurious. This is in line with research by Ross (2004) and Carpenter (2001) that show that asset managers should not risk shift, contrasting with empirical findings that they do. We show instead that documented increases in risk are actually efficient increases in Sharpe ratios, and thus represent efficient portfolio allocation, not risk-shifting.

Do network effects in financial markets create hidden effects that impact asset prices? Broadly, in my research, I define ‘network effect’ as an externality or spillover among agents or market participants via a market-based connection. For example, connections can be common asset holdings, common trading patterns, or common valuation benchmarks. In contrast, I do not have any work on the impact of social networks or social connections. While investigating network effects is growing in popularity, it is still a relatively new approach. It is also a challenging approach because network effects are often endogenous (hard to identify) and computationally intensive to measure. But since network effects are also hard-to-observe contemporaneously, I believe it is a fruitful line of research to help understand pricing inefficiencies and market anomalies.

My dissertation was published as Network Externalities in Mutual Funds (J. Financ. Mark. 2016). This looked at exernalities among mutual fund flows. I showed that funds flowing into mutual funds had reinforcing effects on other mutual funds with similar holdings, independent of style effects. I argue that these externalities are a key driver of the so-called “smart money/dumb money” abnormal return-and-reversal patterns we see in the literature.

I continue to look for more ideas to address with a network methodology. I’ve had several that have started promising but not developed into a good contribution. I have a promising early paper looking at stock pricing in a network framework, but it is still a work in progress.

Do certain types of institutional investors (high frequency traders, in particular) cause inefficiencies in financial markets? High frequency traders (HFT) have received a lot of negative attention in the past few years. It is possible that the rise of computer-based trading (using algorithms, often at very high speed) is a key driver of market inefficiency. Are ‘flash crashes’ the new normal? Do high frequency market makers provide ‘phantom’ quotes that disappear when a trader attempts to transact on them? What is the impact of market fragmentation?

The questions around HFT center on liquidity provision – i.e. the ability to easily transact in financial markets. If HFT are generating liquidity, but regulators move to tax or inhibit them, then price efficiency declines. Alternatively, if HFT are harmful to efficient market pricing and liquidity, and are left to grow and develop unchecked, that too is a suboptimal outcome. Understanding the effect of HFT on market function is a first order problem.

My first paper in this area is Phantom Liquidity and High Frequency Quoting (J. Trading 2016). Here, we show with a massive dataset (Entire order book from all S&P500 stocks for all of 2012) that so-called “cancellation clusters” or abnormal cancellation activity seems to be benign. The limit order book is effectively the same after as it was before. This is far from proving that HFT is benign because we cannot, by definition, identify so-called ‘Phantom Liquidity’ – it is like a black hole. What we show is that by looking at everything around the black hole, there doesn’t seem to be much there.

Philosophically, we also argue in that paper that the conflict around HFT is really simply pitting one set of large institutions (HFT firms) against another set of large institutions (non-HFT institutions). The ‘little guy’ here is clearly better off since very small orders get executed immediately at the prevailing quote. So really, the argument here is about which set of large, sophisticated market participants are advantaged or disadvantaged by market rules.

This is an active area of research for me and more is coming here. Stay tuned.


Berk, Jonathan B, and Jules H van Binsbergen, 2016, Assessing asset pricing models using revealed preference, J. Financ. Econ. 119, 1–23.

Barber, Brad M, Xing Huang, and T Odean, 2016, Which Factors Matter to Investors? Evidence from Mutual Fund Flows, Rev. Financ. Stud. 29, 2600–2642.

Stein, Jeremy C, 2009, Presidential address: Sophisticated investors and market efficiency, J. Finance 64, 1517–1548.

Ross, Stephen A, 2004, Compensation, Incentives, and the Duality of Risk Aversion and Riskiness, J. Finance 59, 207–225.

Carpenter, Jennifer N, 2000, Does Option Compensation Increase Managerial Risk Appetite? J. Finance 55, 2311–2331.

My competitive advantage

One of my competitive advantages is that I am able to compute at scale. Much of my research is computationally intensive. Network computations often rely on well-written algorithms to compute at any reasonable speed. Interact that with large datasets, and naively written code can take weeks to execute. High-frequency trading data is enormous (the dataset for the Phantom Liquidity paper is on the order of 6 TB, though Rick and Ben handled that). For my Limits of (Short) Arbitrage paper, I computed both American and European options prices for almost every option in Option Metrics from 1996-2013 using the Cox-Ingersoll-Ross binomial option pricing model. I then used these prices to rigorously compute the early exercise premium for each contract, each day (EEP = American Price – European Price).

I’m able to do this because I’ve developed skills at writing parallelizable code in both SAS and MATLAB, which I can then run on the Vanderbilt Advanced Computing Center for Research and Education computing grid.