In this work, the authors create machine learning (ML) models to forecast home equity credit risk for individuals using a real-world dataset and demonstrate methods to explain the output of these ML models to make them more accessible to the end user. They analyze the explainability for various stakeholders: loan companies, regulators, loan applicants, and data scientists, incorporating their different requirements with respect to explanations. For loan companies, they generate explanations for every model prediction of creditworthiness. For regulators, they perform a stress test for extreme scenarios. For loan applicants, they generate diverse counterfactuals to guide them with steps toward a favorable classification from the model. Finally, for data scientists, they generate simple rules that accurately explain 70%–72% of the dataset. Their study provides a synthesized ML explanation framework for all stakeholders and is intended to accelerate the adoption of ML techniques in domains that would benefit from explanations of their predictions.

We represent risk factors as sums of orthogonal components capturing fluctuations with cycles of different length. The representation leads to novel spectral factor models in which systematic risk is allowed—without being forced—to vary across frequencies. Frequency- specific systematic risk is captured by a notion of spectral beta. We show that traditional factor models restrict the spectral betas to be constant across frequencies. The restriction can hide horizon-specific pricing effects that spectral factor models are designed to re- veal. We illustrate how the methods may lead to economically meaningful dimensionality reduction in the factor space.

Development of curative treatments for glioblastoma (GBM) has been stagnant in recent decades largely because of significant financial risks. A portfolio-based strategy for the parallel discovery of breakthrough therapies can effectively reduce the financial risks of potentially transformative clinical trials for GBM. Using estimates from domain experts at the National Brain Tumor Society (NBTS), we analyze the performance of a portfolio of 20 assets being developed for GBM, diversified across different development phases and therapeutic mechanisms. We find that the portfolio generates a 14.9% expected annualized rate of return. By incorporating the adaptive trial platform GBM AGILE in our simulations, we show that at least one drug candidate in the portfolio will receive US Food and Drug Administration (FDA) approval with a probability of 79.0% in the next decade.

We propose a heuristic approach to modeling investor behavior by simulating combinations of simpler systematic investment strategies associated with well-known behavioral biases—in functional forms motivated by an extensive review of the behavioral finance literature—using parameters calibrated from historical data. We compute the investment performance of these heuristics individually and in pairwise combinations using both simulated and historical asset-class returns. The mean-reversion or momentum nature of a heuristic can often explain its effect on performance, depending on whether asset returns are consistent with such dynamics. These algorithms show that seemingly irrational investor behavior may, in fact, have been shaped by evolutionary forces and can be effective in certain environments and maladaptive in others.

Academic institutions play a central role in the biotech industry through technology licensing and the creation of startups, but few data are available on their performance and the magnitude of their impact. Here we present a systematic study of technology licensing by one such institution, the Massachusetts Institute of Technology (MIT). Using data on the 76 therapeutics-focused life sciences companies formed through MIT’s Technology Licensing Office from 1983 to 2017, we construct several measures of impact, including MIT patents cited in the Orange Book, capital raised, outcomes from mergers and acquisitions, patents granted to MIT intellectual property licensees, drug candidates discovered and US drug approvals—a key benchmark of innovation in the biopharmaceutical industry. As of December 2017, Orange Book listings for four approved small-molecule drugs cite MIT patents, but another 31 FDA-approved drugs (excluding candidates acquired after phase 3) had some involvement of MIT licensees. Fifty-five percent of the latter were either a new molecular entity or a new biological entity, and 55% were granted priority review, an indication that they address an unmet medical need. The methodology described here may be a useful framework for other academic institutions to track outcomes of intellectual property in the therapeutics domain.

BACKGROUND: Parkinson’s disease (PD) is neurodegenerative, causing motor, cognitive, psychological, somatic, and autonomic symptoms. Understanding PD patients’ preferences for novel neurostimulation devices may help ensure that devices are delivered in a timely manner with the appropriate level of evidence. Our objective was to elicit preferences and willingness-to-wait for novel neurostimulation devices among PD patients to inform a model of optimal trial design.

METHODS: We developed and administered a survey to PD patients to quantify the maximum levels of risks that patients would accept to achieve potential benefits of a neurostimulation device. Threshold technique was used to quantify patients’ risk thresholds for new or worsening depression or anxiety, brain bleed, or death in exchange for improvements in “on-time,” motor symptoms, pain, cognition, and pill burden. The survey elicited patients’ willingness to wait to receive treatment benefit. Patients were recruited through Fox Insight, an online PD observational study.

RESULTS: A total of 2740 patients were included and a majority were White (94.6%) and had a 4-year college degree (69.8%). Risk thresholds increased as benefits increased. Threshold for depression or anxiety was substantially higher than threshold for brain bleed or death. Patient age, ambulation, and prior neurostimulation experience influenced risk tolerance. Patients were willing to wait an average of 4 to 13 years for devices that provide different levels of benefit.

CONCLUSIONS: PD patients are willing to accept substantial risks to improve symptoms. Preferences are heterogeneous and depend on treatment benefit and patient characteristics. The results of this study may be useful in informing review of device applications and other regulatory decisions and will be input into a model of optimal trial design for neurostimulation devices.

We compare and contrast the expected duration and number of infections and deaths averted among several designs for clinical trials of COVID-19 vaccine candidates, including traditional and adaptive randomized clinical trials and human challenge trials. Using epidemiological models calibrated to the current pandemic, we simulate the time course of each clinical trial design for 756 unique combinations of parameters, allowing us to determine which trial design is most effective for a given scenario. A human challenge trial provides maximal net benefits—averting an additional 1.1M infections and 8,000 deaths in the U.S. compared to the next best clinical trial design—if its set-up time is short or the pandemic spreads slowly. In most of the other cases, an adaptive trial provides greater net benefits.

The regulatory approval process for new therapies involves costly clinical trials that can span multiple years. When valuing a candidate therapy from a financial perspective, industry sponsors may terminate a program early if clinical evidence suggests market prospects are not as favorable as originally forecasted. Intuition suggests that clinical trials that can be modified as new data are observed, i.e., adaptive trials, are more valuable than trials without this flexibility. To quantify this value, we propose modeling the accrual of information in a clinical trial as a sequence of real options, allowing us to systematically design early-stopping decision boundaries that maximize the economic value to the sponsor. In an empirical analysis of selected disease areas, we find that when a therapy is ineffective, our adaptive financing method can decrease the expected cost incurred by the sponsor in terms of total expenditures, number of patients, and trial length by up to 46%. Moreover, by amortizing the large fixed costs associated with a clinical trial over time, financing these projects becomes less risky, resulting in lower costs of capital and larger valuations when the therapy is effective.

Biomedical innovation suffers from a “funding gap” between the needs of drug development firms and the availability of funds. The requirement of large investments for drug development projects and the high pipeline risk associated with FDA approval causes this funding gap in part. In this paper, we propose a new financial instrument—the “FDA hedge”—that pays off upon FDA approval failure. We develop a theory to show that the FDA hedge can help eliminate the funding gap. Using novel project-level data, we establish empirically that FDA hedge risk is idiosyncratic, and show how better sharing this risk can spur welfare-enhancing R&D.

The interaction between product market competition, R&D investment, and the financing choices of R&D-intensive firms on the development of innovative products is only partially understood. We hypothesize that as competition increases, R&D-intensive firms will: i) increase R&D investment relative to existing assets in place; ii) carry more cash; and iii) maintain less net debt. Using the Hatch–Waxman Act as an exogenous shock to competition, we provide causal evidence supporting these hypotheses through a differences-in-differences analysis that exploits differences between the biopharma industry and other industries, and heterogeneity within the biopharma industry. We also explore how these changes affect innovative output.