Predicting credit card fraud with Sarbanes‐Oxley assessments and Fama‐French risk factors

This research developed and tested machine learning models to predict significant credit card fraud in corporate systems using Sarbanes‐Oxley (SOX) reports, news reports of breaches and Fama‐French risk factors (FF). Exploratory analysis found that SOX information predicted several types of security breaches, with the strongest performance in predicting credit card fraud. A systematic tuning of hyperparamters for a suite of machine learning models, starting with a random forest, an extremely‐randomized forest, a random grid of gradient boosting machines (GBMs), a random grid of deep neural nets, a fixed grid of general linear models where assembled into two trained stacked ensemble models optimized for F1 performance; an ensemble that contained all the models, and an ensemble containing just the best performing model from each algorithm class. Tuned GBMs performed best under all conditions. Without FF, models yielded an AUC of 99.3% and closeness of the training and validation matrices confirm that the model is robust. The most important predictors were firm specific, as would be expected, since control weaknesses vary at the firm level. Audit firm fees were the most important non‐firm‐specific predictors. Adding FF to the model rendered perfect prediction (100%) in the trained confusion matrix and AUC of 99.8%. The most important predictors of credit card fraud were the FF coefficient for the High book‐to‐market ratio Minus Low factor. The second most influential variable was the year of reporting, and third most important was the Fama‐French 3‐factor model R² – together these described most of the variance in credit card fraud occurrence. In all cases the four major SOX specific opinions rendered by auditors and the signed SOX report had little predictive influence.     

Outside versus inside bonds: A Modigliani–Miller type result for liquidity constrained economies

When agents are liquidity constrained, two options exist – sell assets or borrow. We compare the allocations arising in two economies: in one, agents can sell government (outside) bonds and in the other they can borrow by issuing (inside) bonds. All transactions are voluntary, implying no taxation or forced redemption of private debt. We show that any allocation in the economy with inside bonds can be replicated in the economy with outside bonds but that the converse is not true. However, the optimal policy in each economy makes the allocations equivalent.     

Public knowledge,private knowledge: the intellectual capital of entrepreneurs

This paper focuses on the innovative actions of entrepreneurs, namely their tendency to reveal the intellectual capital that results from their research efforts either in the form of public knowledge (publications) or private knowledge (patents). Using data collected by the National Research Council within the US National Academies from their survey of firm’s that received National Institutes of Health phase II Small Business Innovation Research awards between 1992 and 2001, we find that entrepreneurs with academic backgrounds are more likely to publish their intellectual capital compared with entrepreneurs with business backgrounds, who are more likely to patent their intellectual capital. We also find that, when universities are research partners, their presence complements the tendencies of academic entrepreneurs but does not offset those of business entrepreneurs.     

What Determines the Size of Aid Projects?

Over the last few years, considerable attention has focused on aid fragmentation, the proliferation of donors and projects in developing countries. Aid fragmentation has continued to increase despite international efforts to foster donor coordination. One possible implication of fragmentation is smaller aid projects, potentially with the result of more administrative work for overtaxed recipient governments per dollar of aid received.This paper makes use of AidData data on bilateral aid commitments, sector, and funding agencies to explore the determinants of project size and to better understand the forces driving aid fragmentation. To the extent that project size is driven by the sectoral composition or purpose of aid, the associated administrative costs may be justified. Variations due to other factors, e.g. a donor’s administrative structure or bureaucratic interests, provide a stronger case for reforms.     

Incomplete Credit Markets and Commodity Marketing Behaviour

We use a simple theoretical model of seasonal market participation in the presence of liquidity constraints and transaction costs to explain the ‘sell low, buy high’ puzzle in which some households do not take advantage of inter‐temporal price arbitrage through storage and sell output postharvest at prices lower than observed prices for purchases in the subsequent lean season. We test our model with data from western Kenya using maximum likelihood estimation of a multivariate sample selection model of market participation. Access to off‐farm income and credit indeed seem to influence crop sales and purchase behaviours in a manner consistent with the hypothesised patterns.     

Preventives Versus Treatments Redux: Tighter Bounds on Distortions in Innovation Incentives with an Application to the Global Demand for HIV Pharmaceuticals

Kremer and Snyder (Q J Econ 130:1167–1239, 2015) show that demand curves for a preventive and treatment may have different shapes though they target the same disease, biasing the pharmaceutical manufacturer toward developing the lucrative rather than the socially desirable product. This paper tightens the theoretical bounds on the potential deadweight loss from such biases. Using a calibration of the global demand for HIV pharmaceuticals, we demonstrate the dramatically sharper analysis achievable with the new bounds, allowing us to pinpoint potential deadweight loss at 62% of the global gain from curing HIV. We use the calibration to perform policy counterfactuals, assessing welfare effects of government policies such as a subsidy, reference pricing, and price-discrimination ban. The fit of our calibration is good: we find that a hypothetical drug monopolist would price an HIV drug so high that only 4% of the infected population worldwide would purchase, matching actual drug prices and quantities in the early 2000s before subsidies in low-income countries ramped up.     

Introduction to the Special Issue on “The Intersection Between Industrial Organization and Healthcare Economics”

This article introduces the special issue on the intersection between industrial organization and healthcare economics. This intersection has received increasing research interest, growing from 2% of the industrial-organization literature in 1991 to 6% in 2016. We describe the nine included articles, which cover a range of healthcare markets that include hospital services, insurance, and pharmaceuticals.     

An Empirical Comparison of the Four Channel Flow Model and Adventure Experience Paradigm

The objective of this study was to conduct empirical comparisons between two models of optimal experience within an on-site whitewater kayaking setting using a modification of the Experience Sampling Method. Four concerns are examined: (1) differences in explanatory power between the four channel flow model and the Adventure Experience Paradigm, (2) convergent validity among measures used to determine conditions within these models, (3) differences among measures of perceived challenge and risk between test times of Class I-V river difficulty, and (4) differences among measures of perceived skill and competence between test times of Class I-V river difficulty. Questionnaires were administered in the Cheat River Canyon in West Virginia to 52 whitewater kayakers at eight sites of various levels of river difficulty. Data were analyzed at the experience level, rather than between subjects, using 409 experience sampling observations. Hypothesis testing, performed with statistical analyses (stepwise regression, correlations, and repeated measures ANOVA), suggested that the explanatory powers of the four channel flow model and Adventure Experience Paradigm were similar and indicated support for convergent and ecological validity of measures used to determine conditions within each of the two models.