Using a novel data set of U.S. financial advisors that includes individuals' employment histories and misconduct records, we show that coworkers influence an individual's propensity to commit financial misconduct. We identify coworkers' effect on misconduct using changes in coworkers caused by mergers of financial advisory firms. The tests include merger‐firm fixed effects to exploit the variation in changes to coworkers across branches of the same firm. The probability of an advisor committing misconduct increases if his new coworkers, encountered in the merger, have a history of misconduct. This effect is stronger between demographically similar coworkers.

We test the predictability of investment fraud using a panel of mandatory disclosures filed with the U.S. Securities and Exchange Commission (SEC). We show that past regulatory and legal violations, conflicts of interest, and monitoring have significant power to predict fraud. Avoiding the 5% of firms with the highest ex ante predicted fraud risk would allow an investor to avoid 29% of fraud cases and over 40% of the total dollar losses from fraud. We examine the ability of investors to implement fraud prediction models based on the disclosure filings, and suggest changes in SEC data access policies that could benefit investors.

We present a novel approach for neuron model specification using a Genetic Algorithm (GA) to develop simple firing neuron models consisting of a single compartment with one inward and one outward current. The GA not only chooses the model parameters, but also chooses the formulation of the ionic currents (i.e. single-variable, two-variable, instantaneous, or leak). The fitness function of the GA compares the frequency output of the GA generated models to an I-F curve of a nominal Morris-Lecar (ML) model. Initially, several different classes of models compete among the population. Eventually, the GA converges to a population containing only ML-type firing models with an instantaneous inward and single-variable outward current. Simulations where ML-type models are restricted from the population are also investigated. This GA approach allows the exploration of a universe of feasible model classes that is less constrained by model formulation assumptions than traditional parameter estimation approaches. While we use a simple model, this technique is scalable to much larger and more complex formulations.

No abstract

We present a reduction of a Hodgkin-Huxley (HH)--style bursting model to a hybridized integrate-and-fire (IF) formalism based on a thorough bifurcation analysis of the neuron's dynamics. The model incorporates HH--style equations to evolve the subthreshold currents and includes IF mechanisms to characterize spike events and mediate interactions between the subthreshold and spiking currents. The hybrid IF model successfully reproduces the dynamic behavior and temporal characteristics of the full model over a wide range of activity, including bursting and tonic firing. Comparisons of timed computer simulations of the reduced model and the original model for both single neurons and moderately sized networks (n < or = 500) show that this model offers improvement in computational speed over the HH--style bursting model.