Carnegie Mellon University
July 31, 2023

Summer Research Rundown

A collection of cutting-edge research happening at the Tepper School.

Consider Asset Prices and Portfolio Theory Mechanisms When Analyzing Firms With Negative Environmental Externalities

Burton Hollifield, PNC Professor of Finance; Professor of Financial Economics; Associate Dean, Undergraduate Programs

A major obstacle to addressing global warming is a long-recognized source of market failure: Greenhouse gas emissions create a negative externality (i.e., a cost caused by a producer that is not financially incurred or received by that producer). Producers and consumers of carbon-intensive products typically make small individual contributions to the global stock of greenhouse gases, and carbon dioxide emissions rarely have immediate or localized negative impacts. Such emissions are a difficult externality to address because of their global scope and cumulative, long-term harms. Elementary portfolio theory implies that environmentalists optimally hold more shares of polluting firms than do non-environmentalists, and that polluting firms attract more investment capital than identical non-polluting firms through a hedging channel. Pigouvian taxation (a tax on a market transaction that creates a negative externality or an additional cost borne by individuals not directly involved in the transaction, as in the case of tobacco taxes) can reverse the aggregate investment results, but environmentalists still overweight polluters.

In a new article, researchers introduce countervailing motives for environmentalists to underweight polluters, comparing the implications when environmentalists coordinate to internalize pollution or have nonpecuniary disutility (i.e., non-monetary harmful effects) from holding polluter stock. With nonpecuniary disutility, introducing a green derivative may significantly change who invests most in polluters, but has no impact on aggregate pollution. The authors suggest that asset pricing and portfolio theory mechanisms should be part of the analysis of firms with negative environmental externalities.

The article, Asset Prices and Portfolios with Externalities, appears in Review of Finance and is authored by Baker, SD (Federal Reserve Bank of Richmond), Hollifield, B (Carnegie Mellon University), and Osambela, E (Board of Governors of the Federal Reserve System). Copyright 2022 The Authors. All rights reserved.

Rhetorical Strategies To Frame Business Ventures as Low-Risk and Stable

Emily DeJeu, Assistant Teaching Professor of Business Communication

Despite growing interest among students in entrepreneurship education, the limited research on rhetorical strategies for proposing new business ventures has only focused on the argument strategies that startup entrepreneurs use when delivering oral pitches to investors. In a new study that analyzed nearly two dozen real business plans for small business ventures, a researcher explored the topoi, or lines of argument, that entrepreneurs use in written plans created for bank lenders. The study identified nine topoi that small business entrepreneurs use to achieve two rhetorical goals: justifying their ventures, via the creation of stability-focused value propositions, and establishing their entrepreneurial credibility. The study's author argues that small business entrepreneurs use these topoi to frame their ventures as low in risk and stable, which contrasts with startup entrepreneurs' arguments that their ventures are innovative and disruptive.

In addition to learning strategies for highlighting innovation and disruption, students interested in entrepreneurship education would likely benefit from learning rhetorical strategies for minimizing risk and emphasizing stability. The study, although small and therefore preliminary in its conclusions, can inform pedagogy in the field to ensure that students are given an array of rhetorical strategies to meet different kinds of entrepreneurship challenges.

The article, The Topoi of Small Business Entrepreneurship, appears in Written Communication and is authored by DeJeu, EB (Carnegie Mellon University). Copyright 2023 Sage Publications. All rights reserved.

Why the Overlapping Generations Model Is Not the Central Workhorse Model for Macroeconomics

Stephen E. Spear, Professor of Economics

A major issue in macroeconomics is the question of why the overlapping generations (OLG) model did not become the central workhorse model. Introduced in 1958 by Paul Samuelson, OLG postulates an infinite number of finite-lived families. In contrast, the more dominant neoclassical growth model, the infinite lived agent (ILA) model, is based on the assumption that real economies are populated by a finite number of dynastic families. Despite the greater realism of the OLG model and the inherent implausibility of the assumptions underlying the ILA model, the ILA model has become dominant.

A new book explores the evolution of these two competing dynamic, stochastic, general equilibrium models, which have evolved as workhorse models for macroeconomic analysis (defined broadly to include monetary economics, business cycle theory, economic growth, public finance/optimal taxation, and fiscal policy analysis). The authors shed light on the technical and intellectual forces that generated the current situation in which macroeconomics seems to be caught in a standardization trap based on a problematic choice of models.

The book, Overlapping Generations: Methods, Models and Morphology, is authored by Spear, SE (Carnegie Mellon University), and Young, W (Bar Ilan University). Copyright 2023. All rights reserved.

Fast Combinatorial Algorithms for Min Max Correlation Clustering

Benjamin Moseley, Carnegie Bosch Associate Professor of Operations Research

In a new study, researchers introduced fast algorithms for correlation clustering with respect to the Min Max objective. Correlation clustering, a core tool used in the data mining community, is often used for the analysis of social networks and community detection. The Min Max objective aims to ensure fairness for the individuals being clustered. The Min Max objective has been studied extensively to produce fair clusters. Unfortunately, current solutions for Min Max correlation clustering are mostly of theoretical interest because they are too slow to run on networks of more than a few hundred entities. This paper gives the first algorithm that is guaranteed to return a high-quality solution and is fast enough to cluster networks of tens of thousands of entities. The study's algorithms are the first purely combinatorial approximation algorithms for this problem.

Researchers constructed a novel semi-metric on the set of vertices, which they call the correlation metric, which indicates to clustering algorithms whether pairs of nodes should be in the same cluster. The correlation metric unlocks the ability to cluster networks fast and efficiently. Their study demonstrates empirically that, compared to prior work, their algorithms sacrifice little in the objective quality to obtain significantly better run-time. In addition, their algorithms scale to larger networks that are effectively intractable for known algorithms. The study provides new insights into both correlation clustering and mathematical programing representations of the problem.

The article, Fast Combinatorial Algorithms for Min Max Correlation Clustering, appears in Proceedings of the 40th International Conference on Machine Learning and is authored by Davies, S (Northwestern University), Moseley, B (Carnegie Mellon University), and Newman, H (Carnegie Mellon University). Copyright 2023 The Authors. All rights reserved.

Configuration Balancing for Stochastic Requests

Benjamin Moseley, Carnegie Bosch Associate Professor of Operations Research

Researchers considered a configuration balancing problem with stochastic requests, which generalizes well-studied resource-allocation problems (e.g., load balancing, virtual circuit routing): There are m resources and n requests, and each request has multiple possible configurations, each of which increases the load of each resource by some amount. The goal is to select one configuration for each request to minimize the load of the most-loaded resource. Configuration balancing captures many natural resource-allocation problems in which requests compete for a finite pool of resources and the task is to find a "fair" allocation in which no resource is over-burdened. Here, researchers developed both offline and online algorithms for configuration balancing with stochastic requests. They also showed how to leverage adaptivity in the special case of load balancing on related machines to obtain a constant-factor approximation offline and an approximation online. A crucial technical ingredient in all the results was a new structural characterization of the optimal adaptive policy that allowed them to limit the correlations between its decisions.

The article is part of a volume that includes abstracts of papers presented at the 24th annual conference on Integer Programming and Combinatorial Optimization (IPCO), which is under the auspices of the Mathematical Optimization Society.

The article, Configuration Balancing for Stochastic Requests, appears in Integer Programming and Combinatorial Optimization, Proceedings of the 24th International Conference, IPCO 2023, Madison, WI, June 21-23, 2023, Alberto Del Pia and Volker Kaibel (Eds.), Springer, and is authored by Eberle, F (London School of Economics), Gupta, A (Carnegie Mellon University), Megow, N (University of Bremen), Moseley, B (Carnegie Mellon University), and Zhou, R (Carnegie Mellon University).Copyright 2023 The Editors and the Authors. All rights reserved.