Causes of Outcome Learning: A causal inference-inspired machine learning approach to disentangling common combinations of potential causes of a health outcome

Nearly all diseases can be caused by different combinations of exposures. Yet, most epidemiological studies focus on the causal effect of a single exposure on an outcome. We present the Causes of Outcome Learning (CoOL) approach, which seeks to identify combinations of exposures (which can be interpreted causally if all causal assumptions are met) that could be responsible for an increased risk of a health outcome in population sub-groups. The approach allows for exposures acting alone and in synergy with others. It involves (a) a pre-computational phase that proposes a causal model; (b) a computational phase with three steps, namely (i) analytically fitting a non-negative additive model, (ii) decomposing risk contributions, and (iii) clustering individuals based on the risk contributions into sub-groups based on the predefined causal model; and (c) a post-computational phase on hypothesis development and validation by triangulation on new data before eventually updating the causal model. The computational phase uses a tailored neural network for the non-negative additive model and Layer-wise Relevance Propagation for the risk decomposition through this model. We demonstrate the approach on simulated and real-life data using the R package ‘CoOL’. The presentation is focused on binary exposures and outcomes but can be extended to other measurement types. This approach encourages and enables epidemiologists to identify combinations of pre-outcome exposures as potential causes of the health outcome of interest. Expanding our ability to discover complex causes could eventually result in more effective, targeted, and informed interventions prioritized for their public health impact.

A New Approach to Risk Attribution and Its Application in Credit Risk Analysis

How can risk of a company be allocated to its divisions and attributed to risk factors? The Euler principle allows for an economically justified allocation of risk to different divisions. We introduce a method that generalizes the Euler principle to attribute risk to its driving factors when these factors affect losses in a nonlinear way. The method splits loss contributions over time and is straightforward to implement. We show in an example how this risk decomposition can be applied in the context of credit risk.

How Fraud Risk Decomposition Affects Auditors' Fraud Risk Assessments

SUMMARY In this paper, we provide a practitioner summary of our paper “The Influence of Judgment Decomposition on Auditors' Fraud Risk Assessments: Some Trade-Offs” (Simon, Smith, and Zimbelman 2018). In that study, we investigate potential unintended consequences from current auditing guidance on risk assessments. Specifically, auditing standards recommend separate assessments of the likelihood and magnitude of risks (hereafter, LM decomposition) when auditors assess risk. Our study involved several experiments, including one with experienced auditors, where we found evidence that LM decomposition leads auditors to be less concerned about high-risk fraud schemes relative to auditors who make holistic risk assessments. Our other experiments involved non-auditing settings and replicated this finding while exploring potential explanations for it. After providing a summary of our study and its results, we offer concluding remarks on the potential implications of our findings.

PRACTITIONER SUMMARY How Fraud Risk Decomposition Affects Auditors' Fraud Risk Assessments

In this paper, we provide a practitioner summary of our paper "The Influence of Judgment Decomposition on Auditors' Fraud Risk Assessments: Some Tradeoffs" (Simon, Smith, and Zimbelman 2018). In that study, we investigate potential unintended consequences from current auditing guidance on risk assessments. Specifically, auditing standards recommend separate assessments of the likelihood and magnitude of risks (hereafter, LM decomposition) when auditors assess risk. Our study involved several experiments, including one with experienced auditors, where we found evidence that LM decomposition leads auditors to be less concerned about high-risk fraud schemes relative to auditors who make holistic risk assessments. Our other experiments involved non-auditing settings and replicated this finding while exploring potential explanations for it. After providing a summary of our study and its results, we offer concluding remarks on the potential implications of our findings.

Coskewness Risk Decomposition, Covariation Risk, and Intertemporal Asset Pricing

We develop an intertemporal asset pricing model where cash-flow news, discount-rate news, and their second moments are priced by the market. This model generalizes the market-return decomposition framework, showing that intertemporal considerations imply a decomposition of squared market returns (coskewness risk). Our model accounts for 68% of the return variation across portfolios sorted by size, book-to-market ratio, momentum, investment, and profitability for a modern U.S. sample period. Further, our findings highlight the importance of covariation risk, that is, the risk of simultaneous unfavorable shocks to cash flows and discount rates, in understanding equity risk premia.

Portfolio Optimization with Risk Decomposition

The paper offers the modification of traditional portfolio optimization approach to construct the portfolio with possibility to control both systematic and specific risk (portfolio with risk decomposition). Built on modern econometric tools, the author estimates and forecasts the dynamics of alphas and betas of stocks in the frame of CAPM model, which are further applied for portfolio optimization. The closing weekly prices of 10 Australian stocks and ASX Index as the market index during the period from July 2000 to July 2016 were used. Within the sample there is no evidence of arbitrage on the Australian equity market employing neutral beta portfolio. The study confirms that portfolios with risk decomposition outperform Markowitz’s one according to various performance indicators.