Da Huang

This paper shows that the rise of passive investing makes the active mutual fund industry more skilled. Greater passive investing makes it easier for active funds to outperform the benchmark and accelerates the exit of underperforming funds. In response, skilled managers take less risk to outperform more consistently. Since unskilled active managers introduce noise into stock prices, accelerating their exit improves market efficiency. These findings reconcile the rise of passive investing, closet indexing, and fund homogenization, which may imply a lack of skill, with the literature documenting the presence of skills in the active mutual fund industry.

This paper shows that exchange-traded funds (ETFs) "sample" their indexes, systematically underweighting or omitting illiquid index stocks. As a result, arbitrage activity between the ETF and its index has heterogeneous effects on underlying asset markets. Using an instrumental variables approach, we find that the trading activity of ETFs reduces liquidity and price efficiency and increases volatility and co-movement for liquid stocks, but has no effect on illiquid stocks. Our results demonstrate that the effects of passive investing on asset markets depend on how passive funds replicate their target index.

Property tax limitations reduce the inherent pro-cyclicality of property taxes and expose households to greater risk of mortgage distress. We develop a novel measure of tax policy risk using an Arrow-Debreu framework to price property tax regimes' consumption smoothing features and obtain simulated measures of risk that capture all of the key characteristics of states' property tax systems. We simultaneously account for the effects of these policies---including all applicable tax limitations, effective tax rates, and reassessment frequency---on the overall level of taxation. Using a state-border discontinuity design and parcel-level data for residential properties across the continental U.S., we show that a one standard deviation increase in tax policy risk increases the probability of mortgage distress by nearly 30 percent. Variation in the level of taxation due to these same property tax policies is strongly negatively correlated with tax policy risk, and has a somewhat smaller effect on the probability of distress.

Are cryptocurrencies mere speculative bubbles, or do their prices reflect the underlying technology? We address this question by examining flaws in the underlying technology that could lead to cryptocurrency theft and their effect on returns. Most cryptocurrencies are open-source projects hosted on GitHub with public source code, whose cybersecurity flaws are documented to be fixed later. We find that one flaw predicts a 5 basis points decrease in the coin's daily return. A portfolio that longs no-flaw coins and shorts high-flaw ones, which we term the “cybersecurity factor,” earns 30 basis points daily. Our results demonstrate that cryptocurrency prices reflect the quality of its underlying technology.

Does high-frequency trade increase or decrease volatility in financial markets during crises? We introduce a novel intraday volatility measure for ETFs, and find that during the Covid-19 crisis period, the withdrawal of high-frequency trade from large stock ETFs increases intraday ETF volatility net of the fundamental shock from Covid itself by over 30%. The speed of arbitrage activities slows down during the Covid-19 period as high-frequency traders reduce the intensity of their trading. While high frequency traders may serve as de facto market makers during normal times, they withdraw from the market during a crisis, precisely when they are needed most.

In statistics, samples are drawn from a population in a data-generating process (DGP). Standard errors measure the uncertainty in sample estimates of population parameters. In science, evidence is generated to test hypotheses in an evidence-generating process (EGP). We claim that EGP variation across researchers adds uncertainty: non-standard errors. To study them, we let 164 teams test six hypotheses on the same sample. We find that non-standard errors are sizeable, on par with standard errors. Their size (i) co-varies only weakly with team merits, reproducibility, or peer rating, (ii) declines significantly after peer-feedback, and (iii) is underestimated by participants.