We show that the relationship between aggregate investment and Tobin’s q has become remarkably tight in recent years, contrasting with earlier times. We connect this change with the growing empirical dispersion in Tobin’s q, which we document both in the cross-section and the time-series. To study the source of this dispersion, we augment a standard investment model with two distinct mechanisms related to firms’ research activities: innovations and learning. Both innovation jumps in cash flows and the frequent updating of beliefs about future cash flows endogenously amplify volatility in the firm’s value function. Perhaps counterintuitively, the investment-q regression works better for research-intensive industries, a growing segment of the economy, despite their greater stock of intangible assets. We confirm the model’s predictions in the data, and we disentangle the results from measurement error in q.

I show that patents are pledged as collateral to raise significant debt financing, and that the pledgeability of patents contributes to the financing of innovation. In 2013, 38% of US patenting firms had pledged their patents as collateral at some point, and these firms performed 20% of R&D and patenting in Compustat. Employing court decisions as a source of exogenous variation in creditor rights, I show that patenting companies raised more debt, and spent more on R&D, when creditor rights to patents strengthened. Subsequently, these companies exhibited a gradual increase in patenting output and the use of patents as collateral.

Most research on firm financing studies debt versus equity issuance. We model an alternative source -- non-core asset sales -- and identify three new factors that contrast it with equity. First, unlike asset purchasers, equity investors own a claim to the firm's balance sheet (the "balance sheet effect"). This includes the new financing raised, mitigating information asymmetry. Contrary to the intuition of Myers and Majluf (1984), even if non-core assets exhibit less information asymmetry, the firm issues equity if the financing need is high. Second, firms can disguise the sale of low-quality assets -- but not equity -- as motivated by dissynergies (the "camouflage effect"). Third, selling equity implies a "lemons" discount for not only the equity issued but also the rest of the firm, since both are perfectly correlated (the "correlation effect"). A discount on assets need not reduce the stock price, since assets are not a carbon copy of the firm.

We present a model rationalizing the economic value of digital tokens for launching peer-to-peer platforms: By using the blockchain to transparently distribute tokens before the platform begins operation, a token sale overcomes later coordination failures between transaction counterparties during the platform operation. This result follows from forward induction reasoning, under which the costly and observable action of token acquisition credibly communicates the intent to participate on the platform. Our theoretical framework demonstrates the applications of digital tokens to entrepreneurship, including initial coin offerings (ICOs), and offers guidance for both practitioners and regulators.

(earlier versions were circulated under the titles "Initial Coin Offerings and Platform Building" and "Regulation of Initial Coin Offerings")

How much do students benefit from student loan subsidies? We investigate this question, exploiting a natural experiment: a demand shock due to the tightening of credit standards in the PLUS program in 2011. We first establish that the Bennett Hypothesis is best explained by colleges charging large markups over their marginal costs, rather than by advantageous selection. Then we use our results to estimate that students plausibly capture less of the resources expended on loan subsidies than colleges do. We discuss alternative approaches that would more directly benefit students.

(earlier version circulated under the title "Student loans, marginal costs, and markups: Estimates from the PLUS program")

We find that housing return volatility is negatively correlated with income at the zip-code level. We rationalize this finding with a model featuring a collateral constraint that translates income volatility to housing return volatility. Collateral constraints are tighter for lower-income areas, causing higher housing return volatility. We validate this mechanism using variation in wealth induced by lagged housing returns, using cross-sectional data on the housing expenditure share, and using state-level non-recourse status to instrument for collateral constraints. Consistent with our model, housing return volatility is negatively correlated with lagged returns, positively correlated with expenditure share, and higher in non-recourse states.

What are the real effects of tight monetary policy during a recession? We provide novel evidence from the United States depression of 1920-1921. Our identification strategy exploits county-level variation in access to the Federal Reserve’s discount window, and hand-collected data on banking and agriculture in Illinois. In the short term, tightened conditions at the discount window decreased bank lending and lowered farm revenues. In the long term, however, they lowered debt-to-output levels and led to greater farm productivity and scale, suggesting an avoidance of debt overhang problems. These findings establish a tradeoff between the short- and long-run effects of tight monetary policy.

We show that venture debt, a growing segment of venture capital, confers flexibility in the timing of equity raising. In the Preqin database, we show that venture debt constitutes 15% of total venture capital since 2010, most of it coming between the Series A and D rounds. Using novel contract-level data on venture loans, we further show that debt is repaid quickly out of subsequent equity rounds; that default is rare, while prepayment risk is large; and that intellectual property collateral and warrant coverage are prevalent features. We present a model in which venture debt avoids dilution and “extends the runway.”

We investigate how loan terms respond to competition between lenders when borrowers have multidimensional private information. In our model, competitive lenders screen borrowers using contracts that consist of an interest rate and a collateral requirement. Compared to a monopolistic lender, a competitive market offers uniformly lower collateral requirements, and a larger share of funded projects are negative-NPV. The competitive market may be more or less efficient than the monopolist, depending on the deadweight cost of collateral and the degree of selection externalities. In principle, efficiency could be greatly improved by an uninformed regulator who completely controls the interest rate schedule. However, simpler policies such as minimum collateral or higher interest rates do little to improve efficiency.