Search

We use the best available longitudinal data set, the Health and Retirement Study, and a battery of causal inference methods to provide both central estimates and bounds for the long-term effect of ...

During 2005-2007, the SEC conducted a randomized trial in which it removed short-sale restrictions from one-third of the Russell 3000 firms. Early studies found modest microstructure-related effects of removing the restrictions but no effect on short interest or share prices. More recently, however, many studies have attributed a wide range of substantive indirect outcomes to this experiment. We revisit the principal findings in four recent studies in major journals, using a sample that closely matches the actual experiment and a pre-specified research design, and find no support for any of the reported results. If we instead match their specifications as best we can based on the published descriptions, we still obtain quite different results; only two of 13 outcomes are statistically significant, barely so, and even those results are tenuous. For two papers, we have the authors’ original data and code. We can technically replicate two (different) results, but those results are highly sensitive to specification. Our findings have implications for the robustness of other studies finding indirect effects of the short-sale experiment. More broadly, they have implications for the credibility of “causal” research designs which rely on randomization or on natural experiments. Researchers retain extensive discretion over both the sample and model specification, even for a true randomized experiment. The choices in these studies produced significant results, when other reasonable choices would not. The Online Appendix is available at https://ssrn.com/abstract=3657200. Our pre-specified analysis plan is available at https://ssrn.com/abstract=3415529.

This Online Appendix contains additional results and discussion for Black, Desai, Litvak, Yoo, and Yu, Specification Search in Randomized and Natural Experiments: Lessons from the Regulation SHO Experiment (2020). In particular, it contains further analysis of the various specifications and datasets publicly posted by FHK in response to an earlier draft of the paper, details on sample and other differences between our specification and the best-match specification, and step-by-step results from moving from our specification to the best-match specification for each paper. The underling paper is available at https://ssrn.com/abstract=3657196. Our pre-specified analysis plan is available at https://ssrn.com/abstract=3415529.

This is a pre-analysis plan for The Reg SHO Reanalysis Project, in which we will reassess primary results from selected recent accounting and finance papers, which exploit the randomized trial conducted by the Securities and Exchange Commission (SEC) from May 2, 2005 to July 6, 2007. As part of this trial, the SEC selected every third stock on the Russell 3000 index, based on trading volume during the year before the experiment was announced, and exempted these firms’ shares from the short-sale uptick rule and the test. Early studies of the experiment found minor effects on trading markets (e.g., SEC Office of Economic Analysis (2007), Diether et al (2009)), and led the SEC to rescind these short-sale restrictions. However, more recently, a large body of research has documented evidence for a wide variety of outcomes from the experiment, including increased open short interest, negative returns, reduced investment, reduced earnings management, increased audit fees, and much more. We undertook this project because we viewed most of these outcomes as implausible, given the modest expected impact of the short-sale restrictions on trading markets, limited publicity about the experiment (with no opposition from the affected firms that we could find), and early research consistent with small effects. In this proposed project, we will reassess three prominent, recent studies: Fang, Huang and Karpoff (2016), Grullon, Michenaud and Weston (2015), and Hope, Hu, and Zhao (2017). This pre-analysis plan provides details on our sample selection and the specific analyses we plan to carry out for the main results found in these three papers.

This paper synthesizes recent theoretical and empirical research on the terms of venture capital financings of portfolio companies and limited partnership agreements with investors. The objective is to better understand the ways in which these contracts respond to the conditions of extreme information asymmetry and agency cost among venture capitalists, venture capital investors, and the management of VC-funded firms.

We examine whether the reinvestment choices of public pension funds (PPFs) affect the governance of venture capital funds. We start with a hand-collected dataset of litigation against venture capitalists (VCs) that provides significant shocks to the reputation of VCs. We combine that information with detailed data on limited partner investments in VCs provided by LP Source and test whether PPFs respond differently to the litigation shocks compared to other types of limited partners. Our triple-difference framework reveals that VCs who were defendants in lawsuits suffer a significant subsequent decline in investment by university endowments and several other types of institutional investors, but experience an increase in the investment share of PPFs. Pension funds are about three times more likely to re-invest in post-lawsuit funds offered by litigated VCs. The additional pension fund investments thus partially compensate for the shortfall in post-lawsuit fundraising caused by the exodus of other investors. Our results indicate that the investment choices of PPF managers reduce the effectiveness of reputational penalties imposed by other limited partners in venture capital funds.

We provide the first systematic analysis of the role of reputation in limiting opportunistic behavior by venture capitalists towards four types of counterparties: entrepreneurs, investors, other VCs, and buyers of VC-backed startups. Using a hand-collected database of lawsuits, we document that more reputable VCs (i.e., VCs that are older, have more deals, more funds under management, and syndicate with larger networks of venture capitalists) are less likely to be litigated. We also find that litigated VCs suffer declines in future business relative to carefully selected peers. These negative effects are stronger for more reputable VCs, and when VCs are defendants to multiple lawsuits or sued by entrepreneurs. Our results suggest that reputational mechanisms help deter VC opportunism.

This paper uses a triple difference approach to assess whether the adoption of the Sarbanes-Oxley Act predicts long-term changes in cross-listing premia of affected foreign firms. I measure cross-listing premia as the difference between the Tobin's q of a cross-listed company and a non-cross-listed company from the same country matched on propensity to cross-list (first difference). I find that average premia for firms cross-listed on levels 2 or 3 (subject to SOX) declined in the year of SOX adoption (2002) and remained significantly below their pre-SOX level through year-end 2005 (second difference). Firms listed on levels 2 or 3, which are subject to SOX, experienced larger declines in premia than firms listed on levels 1 or 4, which are not subject to SOX (third difference). The estimated decline is 0.15–0.20 depending on specification. Riskier firms and firms from high-disclosing and high-GDP countries suffered larger post-SOX declines. Firm size predicts smaller declines in premia in well-governed countries. Faster-growing firms in poorly-governed countries experienced smaller declines in premia. The results are robust to the use of different before-and-after periods; the use of annual, quarterly, or monthly data; the use of individual companies' Tobin's q's instead of matched pairs, and different regression specifications. The overall evidence is consistent with the view that SOX negatively affected cross-listed premia, and particularly hurt riskier firms and firms from well-governed countries, while perhaps helping high-growth firms from poorly-governed countries. At the same time, after-SOX, level-23 firms continue to enjoy a substantial premium, estimated at about 0.32.

This paper uses a natural experiment to measure market response to the adoption of the Sarbanes–Oxley Act (ʽʽSOX"). Because SOX applies to all US public companies, US-based studies have difficulty separating the effects of contemporaneous events. However, controlled analysis is available: SOX applies to some cross-listed firms (those listed on level 2 or 3), but not to others (listed on level 1 or 4). By comparing reactions of SOX-exposed foreign firms to reactions of otherwise similar SOX-unexposed foreign firms, we can test investor beliefs about the costs and benefits of SOX in a way that is not cleanly available for US-based studies. We find that stock prices of foreign firms subject to SOX declined (increased) significantly, compared to cross-listed firms not subject to SOX and to non-cross-listed firms, during key announcements indicating that SOX would (would not) fully apply to cross-listed issuers. In cross-sectional tests, high-disclosing firms and firms from high-disclosing countries experienced the strongest declines, while faster-growing companies experienced weaker declines. This evidence is consistent with the view that investors expected the Sarbanes–Oxley Act to have a net negative effect on cross-listed foreign companies, with high-disclosing and low-growth companies suffering larger net costs, and faster-growing companies suffering smaller costs, particularly when they are located in poorly governed countries.

We examine the response of different types of Limited Partners (LPs) to alleged opportunistic behavior on the part of Venture Capitalists (VCs). We use a sample of litigated VCs (identified by Atanasov, et al, 2012, Journal of Finance) to proxy for VC opportunistic behavior. Based on their presumed sensitivity to VC malfeasance and headline risk, we predict that university endowments and economic development authorities will be most likely to respond negatively to potential bad press. To test our hypothesis, we employ differences-in-differences (DiD) analysis and compare the participation of different types of LPs in VC funds before and after litigation relative to the LPs of otherwise similar, matched VCs that are not subject to litigation. We find that endowments reduce by more than 50% their participation in follow-on investment funds offered by litigated VCs relative to other types of LPs. Our results suggest that the threat of university endowment withdrawal of funding can deter VC opportunism.

Recent discussions about Russian foreign policy have generally concentrated on its shift to the right. Along with numerous Western observers, who interpret Russian international behavior as a single-player activity, the Russian foreign minister Andrei Kozyrev has himself attempted to portray his policy in terms of realist theory. Answering numerous accusations on “embracing some policies that [he] once spoke out against” (Rafael et al., 1994), Kozyrev paints an explicit picture of a strategic response by his ministry to objective world conditions. However, analysis shows that world conditions played a very small role in shaping Russian foreign policy.

This paper develops a new, market-based measure of firm-level governance: the degree of correlation between the price of a firm’s minority shares and the market value of firm assets (“beta”). The goal is to capture the extent to which insiders (managers or controllers) translate the value of firm’s assets into the value of minority shares. I hypothesize that insiders vary the levels of asset diversion over time: when the value of firm assets is high, minority investors are content, and insiders divert a higher portion of assets to themselves; when the value of firm assets declines, minority investors start monitoring more closely, and insiders reduce the levels of asset diversion. Thus, firms with lower levels of asset diversion should have a higher correlation between the value of minority shares and the market value of assets. I test this hypothesis on a large panel of oil-producing companies around the world and find supporting evidence. Firms with higher Tobin’s Q have higher betas, as do firms incorporated in Delaware (both proxies for better governance). Firms from countries with better institutional environment have higher betas, both in time series and in cross-section. As country-level institutional environment improves over time, betas rise. I also find evidence inconsistent with the popular “bonding” hypothesis and consistent with the alternative “self-selection” hypothesis: firms from countries with poor regulatory regimes that subject themselves to US or UK regulation by cross-listing on US or UK exchanges have betas more sensitive to changes in their domestic environments than firms from the same countries that did not cross-list.

We use the best available longitudinal dataset, the Health and Retirement Survey, and a battery of causal inference methods to provide both central estimates and bounds on the effect of health insurance on health and mortality among the near elderly (initial age 50-61) over an 18-year period. Those uninsured in 1992 consume fewer healthcare services, but are not less healthy and, in our central estimates, do not die sooner than their insured counterparts. We discuss why a zero average effect of uninsurance on mortality and health is plausible, some selection effects that might explain our full results, and methodological concerns with prior studies.The Appendix that accompanies this article is available on SSRN at http://ssrn.com/abstract=2758692.

I use a natural experiment to identify the effect of labor characteristics on firms’ investment and financial structure. The 1986 immigration reform applied simultaneously to all US firms, significantly increasing the costs of employing undocumented workers, and reducing the pool of potentially available workers currently living abroad. However, not all firms were affected equally: firms located in areas with high levels of immigration used more undocumented labor and therefore experienced a stronger shock to their labor costs and labor characteristics. Likewise, firms from industries that heavily rely on labor of undocumented workers (agricultural production, construction, accommodation and food services) were hit harder than firms from other industries. Using the difference-in-differences approach, I find that after the 1986 immigration reform, firms located in counties with higher levels of immigration significantly reduced their capital investment, dividend payments, debt maturity, cash holdings, and the number of employees, as compared to the control group of firms located in counties with lower levels of immigration. Firms from higher-immigration counties also significantly increased leverage. The changes were concentrated in industries heavily reliant on labor of undocumented workers, whereas industries not using many undocumented workers did not exhibit the same patterns. Other proxies for the use of illegal labor point to the same direction. For a given level of immigration in a county, firms located in counties with higher levels of private-sector union penetration experienced weaker changes in investment patterns. This evidence is consistent with the view that labor market characteristics affect firms’ financial decisions.

This paper uses a natural experiment to shed light on several key questions in securities law - the value of disclosure rules, the beneficiaries of disclosure, and the extent to which low-disclosure markets may deviate from semi-strong market efficiency. On November 6, 2006, the Pink Sheets announced that in May 2007, it would introduce a new classification system: each company would receive a simple sign, reflecting its level of disclosure (including “skull and bones” for firms investigated for fraud; a “stop” traffic sign for firms with no disclosures; a “premier” sign for highly-disclosing firms). The new system would not generate any new information - it would merely summarize already public information in one sign. In the event study of stock market returns, I find that the firms that in May 2007 were assigned into “low-disclosing” categories experienced significant negative returns (7% to 16% decline) following the November 2006 announcement of the upcoming classification, relative to OTC firms that received a “transparent” sign. I also analyze non-disclosure-based predictors of the Pink Sheets classification, and the extent to which investors may have relied on such indirect proxies in their trading. Overall, these results suggest that: (1) prominent summaries of public information can affect stock prices of OTC companies; (2) sophisticated investors predicted subsequent placement of firms into categories and thus were aware of existing levels of disclosure; (3) sophisticated investors expected the new summary disclosure system to affect decisions by unsophisticated investors; (4) sophisticated investors did not expect the new classification to significantly affect the behavior of firms’ insiders; and (5) disclosure by self-regulatory organizations can significantly affect financial markets. This evidence also indicates that prior to the new system, “dark” firms’ prices departed significantly from semi-strong efficiency.

This paper studies the relationship among the U.S. securities laws, the premia that non-U.S. firms obtain by subjecting themselves to U.S. laws, overall U.S. share prices, and a cross-listed firm’s U.S. trading volume. I report three main sets of findings. First, for exchange-traded (NYSE and NASDAQ) cross-listed firms, pair premia and pair returns (premia and returns not explained by valuation and returns for similar non-cross-listed firms from the same home country) are strongly correlated with U.S. stock indices. There is a visually apparent “bubble” in pair premia for these firms, which peaks in early 2000, at the same time as U.S. stock indices. In contrast, pair premia and pair returns for cross-listed firms traded OTC or on PORTAL are not correlated with U.S. indices. The correlation between pair returns and U.S. indices only exists for firms with an above-median ratio of U.S.-based to total trading volume, and is triggered by cross-listing; there is no significant correlation before listing. Second, pair premia for level-23 firms, relative to premia for level-14 firms (“relative pair premia”), exist only in firms with above-median ratio of U.S. to total trading volume. Firms with below-median ratio of U.S. trading have no relative pair premia, regardless of listing level. Third, there are important time variations in relative pair premia. Relative pair premia decline significantly for all firms during the first 6 years after listing, and disappear after year six. The decay is most pronounced for firms with below-median ratio of U.S. trading volume. These results, taken together, weaken the law-based explanation for cross-listing premia (bonding to U.S. securities regime) and strengthen the non-law-based explanations (liquidity and visibility). They also suggest a behavioral explanation: U.S. investors treat high trading volume, exchange traded firms partly like U.S. firms, but treat OTC firm, Portal firms and low-trading-volume exchange-traded firms like other foreign firms.

Prior studies find that foreign firms subject to the Sarbanes-Oxley Act ("SOX") suffered significant declines in market valuation; these declines are too large to be fully explained by the increased costs of direct compliance. This paper investigates one possible source of losses - the charge that SOX discourages corporate risk-taking. I use a triple difference methodology to estimate the effect of SOX on risk-taking by SOX-exposed foreign firms. I match each foreign cross-listed firm to a similar non-cross-listed firm from the same country based on propensity to cross-list. I measure the "pair" risk - the difference between the risk of a cross-listed firm and the risk of its match (first difference). I then estimate the after-minus-before SOX change in pair risk (second difference). Finally, I compare the after-minus-before changes in pair risk for SOX-exposed pairs (where the cross-listed company is listed on level 2 or 3 and is thus subject to SOX) to the change for SOX-unexposed pairs (where the cross-listed company is listed on level 1 or 4 and is thus not subject to SOX) (third difference). I use three sets of proxies for risk: volatility of returns, financial leverage, and cash hording. I find that the risk of SOX-exposed foreign firms declined significantly after SOX on all three measures. High-Tobin's Q firms suffered larger declines in risk. Firms that were riskier before SOX lost more of their value after SOX. This evidence is consistent with the view that SOX induced cross-listed firms to take fewer risks, and placed a particular burden on riskier and better-governed firms.

In November 2006, the Pink Sheets trading system announced the introduction of new disclosure "categories" for all of its companies, proclaiming that "categorizing securities by their level of disclosure will greatly enhance the capital formation process." The new system does not add new disclosure requirements; instead, it simply attaches an easy-to-read sign to each firm based on the existing level of disclosure (a "skull and bones" sign for firms investigated for fraud and spamming their stock; a "stop" sign for firms with no disclosures; a "premier" sign for highly-disclosing firms, and so forth). In the event study of stock market returns surrounding the announcement of the new categorization plan, I find that the "premier" (most-disclosing) category of firms experienced significant positive abnormal returns (24% increase) after the announcement; the category with only limited disclosures experienced significant negative abnormal returns (15% decline). The worst category, firms that have been investigated for fraud, experienced only insignificant declines. These results are consistent with the view that prominent displays of already-known information may affect stock prices, even when such displays add nothing substantively new to the information that is already released to the market. This adds to the growing body of literature studying investors' ability to understand publicly available information.

Venture capital contracts give VCs enormous power over entrepreneurs and early equity investors of portfolio companies. A large literature examines how these contractual terms protect VCs against misbehavior by entrepreneurs. But what constrains misbehavior by VCs? We provide the first systematic analysis of legal and non-legal mechanisms that penalize VC misbehavior, even when such misbehavior is formally permitted by contract. We hand-collect a sample of over 177 lawsuits involving venture capitalists. The three most common types of VC-related litigation are: 1) lawsuits filed by entrepreneurs, which most often allege freezeout and transfer of control away from founders; 2) lawsuits filed by early equity investors in startup companies; and 3) lawsuits filed by VCs. Our empirical analysis of the lawsuit data proceeds in two steps. We first estimate an empirical model of the propensity of VCs to get involved in litigation as a function of VC characteristics. We match each venture firm that was involved in litigation to otherwise similar venture firm that was not involved in litigation and find that less reputable VCs are more likely to participate in litigation, as are VCs focusing on early-stage investments, and VCs with larger deal flow. Second, we analyze the relationship between different types of lawsuits and VC fundraising and deal flow. Although plaintiffs lose most VC-related lawsuits, litigation does not go unnoticed: in subsequent years, the involved VCs raise significantly less capital than their peers and invest in fewer deals. The biggest losers are VCs who were defendants in a lawsuit, and especially VCs who were alleged to have expropriated founders.

Venture capital contracts give VCs enormous power over entrepreneurs and early equity investors of portfolio companies. A large literature examines how these contractual terms protect VCs against misbehavior by entrepreneurs. But what constrains misbehavior by VCs? We provide the first systematic analysis of legal and non-legal mechanisms that penalize VC misbehavior, even when such misbehavior is formally permitted by contract. We hand-collect a sample of over 180 lawsuits involving venture capitalists. Our empirical analysis of the lawsuit data proceeds in two steps. We first estimate an empirical model of the propensity of VCs to get involved in litigation as a function of VC characteristics. We find that poorly performing and early-stage VCs are more likely to participate in litigation. Although we find that older VCs and VCs with more deal flow and larger funds under management are more likely to be litigated, the effect is concave. Second, we analyze the relation between different types of lawsuits and VC fundraising and deal flow. Although plaintiffs lose most VC-related lawsuits, litigation does not go unnoticed: in subsequent years, the involved VCs raise significantly less capital than their peers (matched on age, size, or performance) and invest in fewer and lower quality deals. The biggest losers are VCs who are defendants in a lawsuit and who are alleged to have expropriated founders.

We explore the potential for abuse of startup founders and other common stock shareholders by venture capitalists. We first analyze a set of 26 lawsuits involving venture capitalists and entrepreneurs. Our analysis of lawsuits reveals that VC-related litigation is almost always initiated by founders, and most common allegations are dilution and freezeout of founders, followed by expropriation of company assets via related-party transactions. We document that most of the lawsuits that were not promptly settled end up dismissed by judges on procedural grounds, and yet, after winning, the involved VCs have raised significantly less capital than their peers and have syndicated deals with less reputable partners. We next analyze the founder ownership at the going-public stage in a sample of 390 VC-backed IPOs. We find that founders are less likely to be involved in firm governance and have lower ownership in startups backed by less reputable VCs and where VC investment rounds have been insider dominated. The results suggest that the potential for expropriation of equity holders in venture-backed startups has important implications for entrepreneurial activity.

This essay is prepared for a symposium Bloggership: How Blogs are Transforming Legal Scholarship, held at Harvard Law School on April 27, 2006. It makes three brief points. First, before we ask whether blogs are transforming legal scholarship, we need to put things in perspective. When compared to the impact of other recent developments (availability of data, influx of Ph.D's into the legal academy, long-distance co-authorship, internationalization of legal scholarship and faculties, shift of practitioner-oriented writing to practitioner authors, and so forth), the impact of blogging looks rather pale. Second, I address a popular argument that blogs, like water cooler conversations, affect legal scholarship in many indirect yet important ways - by creating forums for early-stage work. I suggest that blogs are more accurately analogized to bugged water coolers (gathering places that are clearly and openly outfitted with powerful microphones) and ask how the availability of a bugged water cooler may change legal scholarship. Finally, I argue that a successful forum for early ideas must have one of the two properties: privacy or well-specified rules punishing the silence of participants. Bugged water coolers have neither and thus are not likely to succeed in transforming legal scholarship.

This paper has two main components: methodological and substantive. Methodologically, it develops a new way of measuring and coding contractual, statutory, and regulatory variables through series of Monte-Carlo simulations. Substantively, it tests whether the extent to which a firm locks in its capital is related to the level of agency costs. In a study of lock-in choices of venture capital funds, I find that the degree of capital lock-in is inversely related to several measures of expected agency costs. Better venture organizations (the ones with fewer failed and more succeeded companies) lock their capital more tightly, as do funds where managers' compensation is more heavily based on performance and funds that employ alternative governance devices.

When investors sign venture capital partnership agreements, they do not immediately turn over the entire committed capital. Instead, they contribute capital in stages, as needed by the fund. Typically, investors have at least three years to decide whether to honor their commitment obligations fully. To discourage investor defaults, venture funds employ a complicated system of financial penalties. Since venture funds could have eliminated the default problem altogether by demanding the contribution of the entire amount upfront, the question arises: why do venture funds engage into this elaborate staging scheme? I suggest that staged contributions are used as a governance tool: the threat of investor walkaway creates additional incentives for venture capitalists to perform well. The walkaway right, however, comes at a cost of undermining a fund's liquidity and threatening its ability to make investments in a timely fashion. Thus, my hypothesis predicts that the strength of investor walkaway right represents a tradeoff between governance concerns and liquidity concerns. I test this hypothesis by studying venture capital partnership agreements. I find that the strength of investor walkaway right is strongly and positively related to several measures of expected agency costs. Venture funds where managerial compensation is riskier and more heavily tilted toward performance-based component make capital withdrawals more difficult. Larger venture funds (typically run by more prominent managers, who generate fewer mismanagement concerns) give investors weaker walkaway rights.

This paper uses a hand-collected dataset of venture capital partnership agreements to study venture capitalists' compensation. Several new findings emerge. First, VC compensation consists of three elements, not two (management fee and carried interest), as commonly believed. The third element is the value of distribution rules that specify when during the fund's life VCs receive distributions. These rules often generate an interest free loan to VCs from limited partners. A shift from the most popular distribution rule to the second most-popular rule can affect VC compensation as much or more than common variations in management fee (from 2% to 2.5% of committed capital) or carry percentage (from 20% to 25% of fund profit). Second, VC compensation is often more complex and manipulable than it could have been. However, more complex management fee provisions predict lower total compensation, thus complexity is not used to camouflage high pay. Third, common proxies for VC quality predict higher levels of the more transparent forms of VC compensation (carried interest and management fee), but do not predict the levels of opaque compensation (interest-free loan). Fourth, long-term VC performance predicts fund size (which in turn predicts VC pay controlling for fund size), but recent performance does not predict changes in fund size. Finally, VC compensation is less performance-based than commonly believed: for vintage years between 1986 and 1997 (most recent years for fully-liquidated funds), about half of total VC compensation comes from the nonrisky management fee. On average, a 1% increase in fund returns predicts a 0.47% increase in total VC compensation; this pay-performance elasticity is similar to that of public company CEOs during the same years.