Leeds Summer Seminar Series (SSS)
The Leeds Summer Seminar Series (SSS) is an internal talk series that we have done as a school since 2012. The tradition will continue this summer on Zoom with a potential in-person element. We have 6 great speakers lined up, representing all of the divisions in the school. Many of the speakers are new or early in their CU careers and this is a great opportunity to learn about their research.
Talks will be biweekly on Wednesdays at noon starting on June 9th. If you have any questions please email one of the organizers, Jeff York or Clare Wang.
Kai R. Larsen
Associate Professor of Information Management
Title: Global Construct Validation: Increasing Construct Validity Through Construct Similarity and Large-Scale Automation
Abstract: Evaluating the similarity of proposed theoretical constructs to each other and those previously known and studied is imperative in academic research and is required throughout the research process. This paper provides a major update to the construct validation theory and practice. Present methods for construct validation, such as factor analytic techniques, are generally conducted within a narrow scope, such as within a single study, potentially missing constructs introduced in other studies. We present a new notion of global construct validation, which claims construct validity needs to be assessed against all available constructs in a discipline and even across disciplines. Global construct validation adds a new dimension to construct validity – that of construct similarity. Construct similarity is based on semantic closeness of the construct domains and is a novel and key mechanism for performing global construct validation. We introduce the Construct Similarity Framework, which describes the types of relationships among constructs. Equipped with the framework, researchers can now reason why and exactly how construct domains overlap for the constructs of interest. Such reasoning permits more sophisticated handing of construct in studies, including when performing the traditional (typically, local) construct validity techniques.
Zoom Call-In Link: https://cuboulder.zoom.us/j/93786243626
Title: When Does Forecasting GAAP Earnings Entail Unreasonable Effort?
Abstract: SEC rules require that managers reconcile their non-GAAP earnings forecasts with the most directly comparable GAAP forecasts unless doing so would entail ‘unreasonable effort.’ A significant number of managers rely on the unreasonable efforts exception to justify the omission of comparable GAAP forecasts. We analyze firms where managers rely on the unreasonable efforts exception and find that their non-GAAP earnings forecasts are more likely to exclude significant recurring expenses that are not excluded by analysts. Our results suggest that almost a third of managers exploit the unreasonable efforts exception to exclude routine recurring expenses from their earnings guidance.
In-person Presentation: KOBL 352 (NO advance sign-up needed; mask required and no eating)
Concurrent Zoom Call-in Link: https://cuboulder.zoom.us/j/91006540882
Emily Gallagher
Finance
Title: Human Capital Investment After the Storm
Abstract: This paper tests the impact of a major, wealth-destroying natural disaster – Hurricane Harvey (Aug-Sep 2017) – on the use of student debt financing. We find that college-age adults from heavily flooded blocks in Houston are 2.5 percentage points (5.8%) less likely to have student loans than are counterparts from non-flooded blocks. We explain this unexpected decline in student debt by a similar decline in enrollment and graduation rates at more, relative to less, exposed Texas schools. This aggregate decline in enrollment is partially offset by a shift towards college majors that have higher median earnings payoffs, which is suggestive of a scaling back of consumption-based and/or low-return course-taking. Together, results highlight a decrease in both the quantity and the diversity of investments in human capital after the storm.
In-person Presentation: KOBL 352 (NO advance sign-up needed; mask required and no eating)
Concurrent Zoom Call-in Link: https://cuboulder.zoom.us/j/95730576594
Brian Waters
Finance
Title: The Sky's the Limit: Bubbles and crashes when margin traders are all in
Abstract: We analyze a setting in which a risky asset is traded by two types of investors: some are all in and buy up to their margin limit and some buy and sell based on the asset's fundamental value. A higher price of the asset increases all-in investor wealth and they borrow against this wealth to buy more shares. All-in investor demand for shares is therefore upward sloping. If all-in investors have (i) enough wealth, and (ii) access to at least 2:1 leverage, then aggregate demand for shares is S-shaped. If the most recent price was P1, then there are three equilibrium prices, P2>P1>P0. If the price were to rise to P2, then there would again be three equilibrium prices, P3>P2>P0. This is true for any arbitrarily high proposed price. If the price were ever to fall to P0, then P0 would be the unique market-clearing price in all future periods. Our theory provides an explanation for the prominent and sudden surge in prices in January 2021 for retail stocks such as GameStop Corporation.
Full Remote Zoom Call-in Link: https://cuboulder.zoom.us/s/93693647342
Gloria Urrea
Strategy, Entrepreneurship and Operations
Title: The Role of Volunteer Experience on Performance on Online Volunteering Platforms
Abstract: Online volunteering platforms allow humanitarian organizations (HOs) to recruit volunteers to work remotely on projects. While the removal of time and space constraints enables HOs to scale up their volunteer force, HOs must manage greater variation in volunteers’ experience. In this study, we investigate the relationship between volunteers’ experience levels and two performance metrics on these platforms: project completion and volunteer retention. Moreover, we study when experience becomes more relevant to project completion depending on a project’s urgency. To test these relationships, we collected a novel panel dataset from the Humanitarian OpenStreetMap Team Tasking Manager, on which volunteers contribute to mapping projects. Our dataset includes 5,162 online volunteering projects with 2,169,683 contributions by 96,450 volunteers. Overall, our study sheds light on online volunteer management and offers operational insights for HOs as well as for online volunteering platforms.
Full Remote Zoom Call-in Link: https://cuboulder.zoom.us/j/95834023853
Rico Bumbaca
Marketing
Title: Characterizing the Distribution of Heterogeneity: Distributed Markov Chain Monte Carlo for Bayesian Hierarchical Models
Abstract: This article proposes a distributed Markov chain Monte Carlo (MCMC) algorithm for estimating Bayesian hierarchical models when the panel size is extremely large (in the millions of consumers) and the objects of interest are the distribution of heterogeneity and the parameters that characterize it. Extant distributed MCMC methods are inherently inefficient, statistically and computationally, because they require the estimation of both the consumer-level parameters and the distribution of heterogeneity. The approach we present bypasses the estimation of the consumer-level parameters. The two-stage algorithm is asymptotically exact, has excellent variance properties, retains the flexibility of a standard MCMC algorithm, and is easy to implement. The details of the algorithm depend on the form of the prior imposed on the hierarchical model. All three possibilities for the prior are considered: i) nonparametric, ii) exponential family, and iii) nonexponential family, such as a finite mixture. The first stage constructs an estimator of the posterior predictive distribution of the consumer-level parameters, which is also the distribution of heterogeneity. For a nonparametric prior, a second stage is not needed since, by definition, the common parameters that characterize the distribution of heterogeneity are already known. For the two parametric priors (exponential and nonexponential families) for which the common parameters that characterize the distribution of heterogeneity are desired, the second stage draws auxiliary variables from the posterior predictive distribution before directly drawing the common parameters. The proposed algorithm takes particular advantage of exponential family priors by first reducing the auxiliary variables to the sufficient statistics that parameterize the posterior distribution of heterogeneity before drawing the common parameters. Although both stages are embarrassingly parallel, the second stage is sufficiently fast that a serial implementation may be computationally tractable. By avoiding the extensive computational, memory and network resources related to drawing, storing and communicating consumer-level parameters, the algorithm dominates the single-machine benchmark algorithm in computational and statistical efficiency by several orders of magnitude. The algorithm is demonstrated with simulated data and with an application characterizing the purchase behavior of members of a loyalty program.
Zoom Call-In Link: https://cuboulder.zoom.us/j/91361850347
Link to paper: Characterizing The Distribution of Heterogeneity