This paper examines the production and hedging decisions of the competitive firm under price uncertainty when the firm is not only risk averse but also regret averse. Regret-averse preferences are characterized by a modified utility function that includes disutility from having chosen ex-post suboptimal alternatives. The extent of regret depends on the difference between the actual profit and the maximum profit attained by making the optimal production and hedging decisions had the firm observed the true realization of the random output price. While the separation theorem holds under regret aversion, the prevalence of hedging opportunities may have perverse effect on the firm's optimal output level, particularly when the firm is sufficiently regret averse. The full-hedging theorem, however, does not hold. We derive sufficient conditions under which the regret-averse firm's optimal futures position is an under-hedge (over-hedge). We further show that the firm optimally increases (decreases) its futures position when the price risk possesses more positive (negative) skewness.

On Making Sense of Today is a collection of essays trying to make sense of not only today, but of yesterday - and yesterday's actions, and today's consequences of those actions. It seeks to find understanding of loss, discontent, and disconnection, but ultimately searches for and finds hope, empathy, and connection in the human experience. The collection ranges in subject matter from the trivial - how to recover from a forgotten essay idea - to weightier matters - whether any death should be advocated for and celebrated - and many issues and questions in-between. This collection is also a study of uncertainty and vulnerability, especially how examining uncertainty with vulnerability can cultivate a deeper observation of the human condition. Some of the uncertainties considered in this collection are how to respond to a duplicitous compliment, how a parent can reconcile misunderstanding a child's love language, and how to process the complicated emotions of a child disconnecting from a parent. While the uncertainty of each question is unresolved, each essay does examine these uncertainties with vulnerability, and that vulnerability creates a connection to any reader who has struggled with a similar uncertainty and in that process, both author and reader feel a little less lonely in their struggles to make sense of things that cannot always make sense.

<p>Consumers are negatively impacted by the increasingly high rate of product returns. In 2020, an estimated $428 billion in merchandise were returned to retailers post-purchase with $25.3 billion being fraudulent returns (NRF.com). Previous research has stated that consumers undergo various negative emotional and cognitive mechanisms when returning and identified reasons as to why consumers return purchases such as product failure, dissatisfaction, and regret (Lee, 2015). Specifically, regret occurs when an individual second-guesses a chosen product due to the realization that the benefits of the unchosen product outweigh the original choice, which elicits uncomfortable feelings (Zeelenberg et al., 1998). However, how does the process of product acquisition and the outcome of the purchasing decision affect post-purchase consumer regret? The purpose of this study is to investigate how the process of expending consumer resources (e.g., time vs. money) to acquire a product and the outcome of inconsistent product attitudes and behaviors (e.g., cognitive dissonance) can affect post-purchase consumer regret (PPCR). In this mixed factorial design, participants viewed scenarios that presented the ‘time’ and ‘money’ spent in acquiring their chosen product and were asked to read a product review that either elicited low or high dissonant feelings. It was hypothesized that consumers would experience greater PPCR when dissonance is high, and the time spent to acquire the product is primed. The interaction effect was not supported; however, an ad hoc analysis revealed that a consumer experienced less PPCR when dissonance is high, and the time spent to acquire the product is highly convenient. The current findings highlight the importance of understanding the process and outcome of purchase on post-purchase evaluations.</p>

My collection of short stories is set in the fictional town of Lockswood Gap, Tennessee, and centers around the lives of four women. Through various points of view and story lengths, I interweave several story lines to span over a time period of about twenty years. Themes of change and regret are prevalent in these stories, as each of these four women must make, or refuse to make, choices that will impact their lives. I modeled my collection after Jennifer Egan’s A Visit from the Goon Squad, using individual short stories that share the same group of characters to tell a novel-length story. The ten stories included in my thesis will comprise about threeourths of the novel, and I will add several more to it following my graduation.

Exploration-exploitation trade-off is a fundamental dilemma that reinforcement learning algorithms face. This dilemma is also central to the design of various state of the art bandit algorithms. We take inspiration from these algorithms and try to design reinforcement learning algorithms in an episodic setting. In this work, we develop two algorithms which are based on the principle of optimism in face of uncertainty to minimize exploration. The idea is that the agent follows the optimal policy for a surrogate model, named optimistic model, which is close enough to the former but leads to a higher longterm reward. We show extensively through experiments on synthetic toy MDP’s that the performance of our algorithms is in line (even better in the case where the reward dynamics are known) with the algorithms based on the Bayesian treatment of the problem and other algorithms based on the optimism in face of uncertainty principle. The algorithms suggested in this thesis trump the Bayesian algorithms in terms of the variance of the regret achieved by the algorithms over multiple runs. Another contribution is the derivation of several regret lower bounds,such as a problem specific (both, asymptotic and non-asymptotic) and a minimax regret lower bound, for any uniformly good algorithm in an episodic setting. / Avvägningen mellan upptäckande och utnyttjande är ett grundläggande dilemma som övervakade inlärningsalgoritmer handskas med. Det här dilemmat är också centralt i designen av diverse toppmoderna bandit-algoritmer. Vi inspireras av dessa algoritmer och försöker utforma övervakade inlärningsalgoritmer i en episodisk miljö. I det här arbetet utvecklar vi två algoritmer som är baserade på principen om optimism vid osäkerhet för att minimera upptäckande. Idén är att agenten följer den optimala policyn för en surrogatmodell som kallas optimistisk modell, som är tillräckligt nära ursprungsmodellen men leder till en högre långsiktig belöning. Vi visar utförligt genom experiment på syntetiska leksaks-MDP att algoritmernas prestanda är i linje med (till och med bättre när belöningsdynamiken är känd) algoritmerna grundade på den bayesiska behandlingen av problemet och andra algoritmer baserade på optimism vid osäkerhet. Algoritmerna som föreslås i den här avhandlingen presterar bättre än de bayesiska algoritmerna i varians av den ånger som uppnås av algoritmerna över många körningar. Ett annat bidrag är härledningen av flera nedre gränser, såsom en problem-specifik nedre gräns (både asymptotisk och icke-asymptotisk) och en nedre gräns enligt minmax-principen, för en godtycklig uniformt god algoritm i en episodisk miljö.

No description available.

This dissertation consists of three chapters that generally focus on the design of welfare-maximizing treatment assignment rules in heterogeneous populations with interactions. In the first two chapters, I focus on an important pre-step in the design of treatment assignment rules: inference for heterogeneous treatment effects in populations with interactions. In the final chapter, I and my co-authors study treatment assignment rules in the presence of social interaction in heterogeneous populations. In chapter one, I argue that statistical inference of heterogeneous treatment effects (HTEs) across predefined subgroups is complicated when economic units interact because treatment effects may vary by pretreatment variables, post-treatment exposure variables (that measure the exposure to other units’ treatment statuses), or both. It invalidates the standard hypothesis testing technique used to infer HTEs. To address the problem, I develop statistical methods (asymptotic and bootstrap) to infer HTEs and disentangle the drivers of treatment effects heterogeneity in populations where units interact. Specifically, I incorporate clustered interference into the potential outcomes model and propose kernel-based test statistics for the null hypotheses of (a) no HTEs by treatment assignment (or post-treatment exposure variables) for all pretreatment variables values; and (b) no HTEs by pretreatment variables for all treatment assignment vectors. To disentangle the source of heterogeneity in treatment effects, I recommend a multiple-testing algorithm. In addition, I prove the asymptotic properties of the proposed test statistics via a modern poissonization technique. As a robust alternative to the inferential methods I propose in chapter one, in chapter two, I design randomization tests of heterogeneous treatment effects (HTEs) when units interact on a single network. My modeling strategy allows network interference into the potential outcomes framework using the concept of network exposure mapping. I consider three null hypotheses that represent different notions of homogeneous treatment effects, but due to nuisance parameters and the multiplicity of potential outcomes, the hypotheses are not sharp. To address the issue of multiple potential outcomes, I propose a conditional randomization inference method that expands on existing methods. Additionally, I consider two techniques that overcome the nuisance parameter issue. I show that my conditional randomization inference method, combined with either of the proposed techniques for handling nuisance parameters, produces asymptotically valid p-values. Chapter three is based on a joint paper with Young Ki Shin and Seungjin Han. We study treatment assignment rules in the presence of social interaction in heterogeneous populations. We construct an analytical framework under the anonymous interaction assumption, where the decision problem becomes choosing a treatment fraction. We propose a multinomial empirical success (MES) rule that includes the empirical success rule of Manski (2004) as a special case. We investigate the non-asymptotic bounds of the expected utility based on the MES rule. / Dissertation / Doctor of Philosophy (PhD)

The purpose of this study is to conduct a thematic linguistic analysis of survey responses to museum-quality images depicting various Native American and white settler encounters. The survey asked participants to provide written responses (fill in the blank prompts) to a selection of twelve images composed of photographs and paintings representing one or more of three overarching themes: violence, immersion, and goodwill/collaboration. The research focused on four demographic groups: Latter-day Saints, Native Americans, museum employees, and total participants. Each response was individually analyzed by hand and assigned appropriate classification tags based on the types of words their responses contained including one or more of the following categories: positive, negative, neutral, pushed fear/propaganda, guilt, curiosity, questioning image/artist, questioning accuracy, loaded, wanting more information, and connection/empathy. After the initial analysis, I created word frequency corpuses to calculate word frequency for each image and group. The differing word frequency corpuses showed that high frequency 3 words did not change much among gender, age, or location but a large variation did exist among terms used less than five times. The identification markers that showed the most variance between interpretations of the artwork were museum employees and Native Americans.

No description available.

<p> </p> <p>Formulating RL with MDPs work typically works for a single objective, and hence, they are not readily applicable where the policies need to optimize multiple objectives or to satisfy certain constraints while maximizing one or multiple objectives, which can often be conflicting. Further, many applications such as robotics or autonomous driving do not allow for violating constraints even during the training process. Currently, existing algorithms do not simultaneously combine multiple objectives and zero-constraint violations, sample efficiency, and computational complexity. To this end, we study sample efficient Reinforcement Learning with concave objective and convex constraints, where an agent maximizes a concave, Lipschitz continuous function of multiple objectives while satisfying a convex cost objective. For this setup, we provide a posterior sampling algorithm which works with a convex optimization problem to solve for the stationary distribution of the states and actions. Further, using our Bellman error based analysis, we show that the algorithm obtains a near-optimal Bayesian regret bound for the number of interaction with the environment. Moreover, with an assumption of existence of slack policies, we design an algorithm that solves for conservative policies which does not violate  constraints and still achieves the near-optimal regret bound. We also show that the algorithm performs significantly better than the existing algorithm for MDPs with finite states and finite actions.</p>