Two new non convex penalty functions - Laplace and arctan - were recently introduced in the literature to obtain sparse models for high-dimensional statistical problems. In this article, we study the ...theoretical properties of Laplace and arctan penalized ordinary least squares linear regression models. We first illustrate the near-unbiasedness of the non zero regression weights obtained by the new penalty functions, in the orthonormal design case. In the general design case, we present theoretical results in two asymptotic settings: (a) the number of features, p fixed, but the sample size,
n
→
∞
, and (b) both n and p tend to infinity. The theoretical results shed light onto the differences between the solutions based on the new penalty functions and those based on existing convex and non convex Bridge penalty functions. Our theory also shows that both Laplace and arctan penalties satisfy the oracle property. Finally, we also present results from a brief simulations study illustrating the performance of Laplace and arctan penalties based on the gradient descent optimization algorithm.
The Spike-and-Slab LASSO Ročková, Veronika; George, Edward I.
Journal of the American Statistical Association,
01/2018, Volume:
113, Issue:
521
Journal Article
Peer reviewed
Despite the wide adoption of spike-and-slab methodology for Bayesian variable selection, its potential for penalized likelihood estimation has largely been overlooked. In this article, we bridge this ...gap by cross-fertilizing these two paradigms with the Spike-and-Slab LASSO procedure for variable selection and parameter estimation in linear regression. We introduce a new class of self-adaptive penalty functions that arise from a fully Bayes spike-and-slab formulation, ultimately moving beyond the separable penalty framework. A virtue of these nonseparable penalties is their ability to borrow strength across coordinates, adapt to ensemble sparsity information and exert multiplicity adjustment. The Spike-and-Slab LASSO procedure harvests efficient coordinate-wise implementations with a path-following scheme for dynamic posterior exploration. We show on simulated data that the fully Bayes penalty mimics oracle performance, providing a viable alternative to cross-validation. We develop theory for the separable and nonseparable variants of the penalty, showing rate-optimality of the global mode as well as optimal posterior concentration when p > n. Supplementary materials for this article are available online.
Financial penalties for delayed enrollment could be useful tools to encourage people to enroll earlier in health insurance markets, but little is known about how effective they are. We use a large ...administrative dataset for a 10% random sample of all Australian tax‐filers to study how people respond to a step‐wise age‐based penalty, and whether the effect has changed over time. Individuals must pay a 2% premium surcharge for each year they delay enrollment beyond age 31. The penalty stops after 10 years of continuous hospital cover. The age‐based penalty creates discontinuities in the incentive to insure by age, which we exploit to estimate causal effects. We find that people respond as expected to the initial age‐penalty, but not to subsequent penalties. The 2% premium loading results in a 0.78–3.69 percentage points (or 2.1%–9.0%) increase in the take‐up rate at age 31. We simulate the penalty impact and implications of potential reforms, and conclude that modest changes around the policy make little difference in the age distribution of insured, premiums or take‐up rates. Our study provides important evidence on an understudied area in the literature and offers insights for countries considering financial penalties.
Fentanyl and fentanyl analogs are increasingly prevalent in the nation’s illicit drug supply. While fentanyl-related deaths were previously confined largely to states east of the Mississippi, they ...are now increasing rapidly throughout much of the United States. Contaminants other than fentanyl are also present in illicitly obtained drugs. Using drug checking equipment to determine the contaminants that may be present in these drugs is a promising avenue for reducing overdose-related harm.
Systematic legal review in which three trained legal researchers collected, reviewed, and coded all US state laws that specify whether the possession and distribution of drug checking equipment is generally legal, whether is it legal in the context of syringe services programs, and, where possession or distribution of drug checking equipment is not clearly legal, the potential penalties for violation.
We find that it is clearly legal to possess some or all drug checking equipment in 22 states, and clearly legal to distribute it to adults in 19 states. In 14 states where distribution of drug checking equipment is not clearly legal generally, it is legal when that equipment is obtained from a syringe services program. Potential penalties for violations range from small civil fines to multi-year jail sentences.
There are currently great variations between states regarding the legality of drug checking equipment. Clarifying, modifying, or repealing drug paraphernalia laws would likely improve access to these promising technologies, potentially reducing overdose deaths.
•Street drugs are increasingly contaminated with fentanyl and other adulterants.•Drug checking equipment can reduce overdose risk by identifying adulterants.•State laws on the legality of drug checking equipment vary widely.•Modifying state laws to legalize drug checking equipment may reduce overdose.
Drawing upon rational choice and investor attention theories, we examine how accusations of corporate bribery and subsequent investigations shape market reactions. Using event study methodology to ...measure loss in firm value for public firms facing bribery investigations from 1978 to 2010, we found that total market penalties amounted to $60.61 billion. We ran moderated multiple regression analysis to examine further the degree to which the unique characteristics of bribery explain variations in market penalties. Companies committing bribery in less corrupt host countries and with the involvement of compromised executives experienced greater market penalties than did other companies. After partitioning share value losses into components for regulatory penalties, class action settlements, and loss to reputation, we found that reputational penalties account for 81.8c̷ of every dollar of share value loss. Omission of reputational penalties in rational choice calculus underestimates bribery costs by 4.5 times. The results suggest that firms should not underestimate the importance of market-imposed reputational penalties by merely considering regulator-imposed fines and sanctions.
Rethinking the Risks of Poverty Brady, David; Finnigan, Ryan M.; Hübgen, Sabine
The American journal of sociology,
11/2017, Volume:
123, Issue:
3
Journal Article
Peer reviewed
Open access
This article develops a framework for analyzing the risks of poverty in terms of prevalences (share of the population with a risk) and penalties (increased probability of poverty associated with a ...risk). A comparison of the four major risks (low education, single motherhood, young headship, and unemployment) across 29 rich democracies reveals there is greater variation in penalties than prevalences. The United States has high poverty partly because it has the highest penalties despite below average prevalences. Also, U.S. poverty in 2013 would be worse with prevalences from 1970 or 1980. There is little evidence that penalties discourage prevalences, while welfare generosity significantly moderates the penalties for unemployment and low education. The authors conclude that a focus on risks does not provide a convincing explanation of poverty, single motherhood may be the least important of the risks, and studies based solely on the United States are constrained by potentially large sample selection biases.
ABSTRACT
I develop a structural model to quantify the costs of tax avoidance. In the model, the firm trades off tax savings with tax audit risk, financial reporting considerations, and operational ...frictions imposed by tax avoidance, the last of which I label as nontax costs. The estimated parameters suggest nontax costs, which are difficult to observe, decrease pretax income by 6.4 percent or $58 million per firm-year. The large magnitude of this estimate can explain why firms appear to underutilize tax avoidance strategies. Through counterfactual analysis, I estimate the effect of tax audit risk and financial reporting considerations to find that financial reporting considerations have an effect on tax avoidance similar to the penalties imposed by tax authorities. Overall, the estimated parameters help explain the “undersheltering puzzle.”
JEL Classifications: G14; H21; H25; H26; M41; M48.
This study examines the importance of social perception of corporate social responsibility (CSR) and irresponsibility (CSI). Drawing from social psychology literature on stereotypes, we argue that ...two fundamental dimensions of social perception-warmth and competence-help explain the underlying processes and conditions under which CSR leads to specific outcomes. We propose that firms engaging in CSR are perceived as higher in warmth and, by default, competence; moreover, different perceptions of the organization's warmth and competence can moderate CSR rewards and CSI penalties. To demonstrate this, we conduct three experiments. Experiment 1 links CSR with perceptions of warmth and competence, showing that warmth perceptions mediate the relationship between CSR and important outcomes, such as purchase intentions and reputation. Experiment 2 adds information on firms' countries of origin, revealing that CSR rewards and CSI penalties differ depending on the (mis)alignment of CSR strategy with country stereotypes. Experiment 3 replicates these findings using behavioral paradigms. We find that firms from high-warmth countries (the United States, Sweden, Portugal) receive lower CSR rewards and pay higher CSI penalties than firms from low-warmth countries (Germany, Pakistan) but this effect is moderated by competence. Our micro–macro study advances social evaluation, strategic CSR, and international management literatures.
How economic sanctions are enforced undoubtedly affects their chances of success. Yet, to date, a lot remains unknown about how governments actually enforce their sanctions. To shed light on the ways ...governments enforce their sanctions, we conduct an empirically driven case study of how the lead agency for enforcing economic sanctions in the United States, the Treasury Department's Office of Foreign Assets Control (OFAC), has enforced US sanctions from 2003 to 2018. Our study begins with an empirical exploration of how OFAC's enforcement actions originate, what factors correlate with enforcement actions being taken, and what factors contribute to the financial penalties imposed as part of enforcement actions. We find that OFAC engaged in two distinct enforcement strategies during our period of analysis: a "fishing" strategy that involved taking a lot of enforcement actions and imposing small fines during the administration of George W. Bush and a "whale-hunting" strategy that involved pursuing fewer cases but imposing enormous fines during Barack Obama's administration. Our analysis indicates that the shift in foreign policy emphasis from Cuba to Iran during the Bush administration and how that motivated new legislation to enhance penalties for violating sanctions paved the way for the OFAC to adopt its innovative whale-hunting strategy during the Obama administration. Our study also yields a wealth of new empirical insights into how the United States enforces its sanctions that can contribute to future theory-building efforts.
On-demand delivery through sharing platforms represents a rapidly expanding segment of the global workforce. The emergence of sharing platforms enables gig workers to choose when and where to work, ...allowing them to do so in a flexible manner. However, such flexibility brings notorious challenges to platforms in managing the gig workforce. Thus, understanding the incentive and behavioral issues of gig workers in this new business model is inherently meaningful. This paper investigates how the incentive mechanisms of sharing platforms—earnings, ratings, and penalties—affect the working decisions of gig workers and their nuanced relationships. To achieve this goal, we use data from one leading on-demand delivery platform with more than 50 million active consumers in China and implement a two-stage Heckman model with instrumental variables to estimate the impact of earnings, ratings, and penalties. We first show that better ratings motivate gig workers to work more. However, interestingly, when ratings are employed together with earnings, the two positive effects of ratings and earnings can be substitutes for each other. Second, we reveal that higher past penalties discourage workers from working more, whereas, interestingly, workers with higher past penalties tend to be more sensitive toward an increase in earnings. Finally, we conduct follow-up surveys to understand the underlying mechanisms of the observed moderating effects from both psychological and economic perspectives. The ultimate goal of this work is to provide managerial implications to help platform managers understand how earnings, ratings, and penalties work together to affect gig workers’ working decisions and how to manage high- and low-quality workers.
This paper was accepted by David Simchi-Levi, entrepreneurship and innovation.
Funding:
This work was supported by the National Natural Science Foundation of China Grants 72192823, 72172169, 71821002, 91646125, 72071206, 72231011, 72025405, and 72088101 and Program for Innovation Research at the Central University of Finance and Economics.
Supplemental Material:
The data files and online appendices are available at
https://doi.org/10.1287/mnsc.2023.4761
.
PDF
