Publications

Machine learning tools have been remarkably successful at using published anomalies for creating portfolios with extremely high returns. However, the underlying economic mechanisms behind their performance remains unclear. This paper proposes a theory-based dynamic arbitrage trading strategy to interpret how neural networks select among anomalies over time. Using 153 firm characteristics (anomalies), this strategy ranks them similarly to neural networks and absent the use of microcaps explains nearly 40% of their monthly performance. When unpublished anomalies and microcap stocks are excluded, an economic model based algorithm fully explains almost all of the neural network’s return performance and largely duplicates the anomaly selection. Additionally, we show how the performance of neural networks can be further improved by incorporating aspects of economic principles.

Over the past two decades, the European Commission's enforcement activities against abusive practices often came too late. Investigations only started when companies had already acquired a strong dominant position. It proved difficult to restore competition. In this paper, we advocate an earlier intervention, namely when undertakings start to monopolize markets. This is possible under Art. 102 TFEU with a return to the original definition of dominance and less emphasis on market shares. In line with cases like Hoffman-LaRoche, dominance must be defined as the ability to steer the market into a direction that is detrimental to competition and trading partners. This will allow to investigate the behaviour of firms at an earlier point of time, particularly in markets prone to tipping.

We re-analyze data from four recent papers to demonstrate that the traditional approach for probing
interactions—linear regression followed by Simple-Slopes ("spotlight") or the Johnson-Neyman
procedure ("floodlight")—can lead to qualitatively incorrect conclusions when true relationships
are not linear. Building on proposals to use GAMs (Generalized Additive Models) to probe
interactions, we introduce the Johnson-Neyman 2.0 procedure (JN2). This procedure involves: (1)
using GAMs to probe an interaction of interest, and then (2) verifying the key conclusion of interest
with a t-test or linear regression run on the subset of relevant data. For example, if a GAM
determines an experimental manipulation has a negative effect for moderator values 0.67 SD below
the mean, a t-test is run only among those observations to verify the negative effect.

Nike describes its purpose with reference to three “Purpose Pillars”: People, Planet, and Play. Each pillar sets targets, tracks progress, and assesses outcomes tied to compensation. This approach helps Nike navigate complex social issues and stakeholder relationships, often translating corporate principles into actionable strategies.

Classical Bayesian mechanism design relies on the common prior assumption, but the common prior is often not available in practice. We study the design of prior-independent mechanisms that relax this assumption: The seller is selling an indivisible item to n buyers such that the buyers’ valuations are drawn from a joint distribution that is unknown to both the buyers and the seller, buyers do not need to form beliefs about competitors, and the seller assumes the distribution is adversarially chosen from a specified class. We measure performance through the worst-case regret, or the difference between the expected revenue achievable with perfect knowledge of buyers’ valuations and the actual mechanism revenue. We study a broad set of classes of valuation distributions that capture a wide spectrum of possible dependencies: independent and identically distributed (i.i.d.) distributions, mixtures of i.i.d. distributions, affiliated and exchangeable distributions, exchangeable distributions, and all joint distributions. We derive in quasi closed form the minimax values and the associated optimal mechanism. In particular, we show that the first three classes admit the same minimax regret value, which is decreasing with the number of competitors, whereas the last two have the same minimax regret equal to that of the case n = 1. Furthermore, we show that the minimax optimal mechanisms have a simple form across all settings: a second-price auction with random reserve prices, which shows its robustness in prior-independent mechanism design. En route to our results, we also develop a principled methodology to determine the form of the optimal mechanism and worst-case distribution via first-order conditions that should be of independent interest in other minimax problems.

Two-sided platforms rely on their recommendation algorithms to help visitors successfully find a match. However, on platforms such as VolunteerMatch, which has facilitated millions of connections between volunteers and nonprofits, a sizable fraction of website traffic arrives directly to a nonprofit’s volunteering page via an external link, thus bypassing the platform’s recommendation algorithm. We study how such platforms should account for this external traffic in the design of their recommendation algorithms, given the goal of maximizing successful matches. We model the platform’s problem as a special case of online matching, where (using VolunteerMatch terminology) volunteers arrive sequentially and probabilistically match with one opportunity, each of which has a finite need for volunteers. In our framework, external traffic is interested only in their targeted opportunity; by contrast, internal traffic may be interested in many opportunities, and the platform’s online algorithm selects which opportunity to recommend. In evaluating the performance of different algorithms, we refine the notion of competitive ratio by parameterizing it based on the amount of external traffic. After demonstrating the shortcomings of a commonly used algorithm that is optimal in the absence of external traffic, we propose a new algorithm, adaptive capacity (AC), which accounts for matches differently based on whether they originate from internal or external traffic. We provide a lower bound on AC’s competitive ratio that is increasing in the amount of external traffic and that is close to (and, in some regimes, exactly matches) the parameterized upper bound we establish on the competitive ratio of any online algorithm. We complement our theoretical results with a numerical study motivated by VolunteerMatch data where we demonstrate the strong performance of AC relative to current practice and further our understanding of the difference between AC and other commonly used algorithms.

Environmental jolts can trigger more conducive conditions for driving change in organizations. However, punctuated equilibrium theories of organizational change concentrate on top managers’ implementation of de novo radical changes after jolts. Existing research has not examined frontline-driven, incremental change efforts during these periods of disrupted stasis, despite the value of frontline change ideas. We develop a process model to explain how and when those on an organization’s front lines can leverage a jolt to opportunistically implement long-desired change ideas in ways that promote their retention. We conducted a two-year qualitative field study at a hospital during the Covid-19 pandemic, examining the trajectories of 33 premeditated change ideas raised by frontline staff. By comparing ideas that persisted to become part of normal operations with those that failed to be selected or retained, we identified practices and conditions that promoted the selection and retention of frontline change ideas. Our study suggests that frontline change advocates can seed the long-term retention of “shovel-ready” ideas—as opposed to de novo ideas—after a jolt by rapidly and opportunistically deploying a novel set of practices before the brief window of opportunity created by lessened constraints and increased managerial receptivity closes. Prior theories of change largely assume frontline-driven change to be slow and continuous, proceeding in a one-off fashion; we explain how and when frontline change can instead occur in rapid, opportunistic bursts. This study advances theories of punctuated equilibrium and bottom-up change in organizations by unearthing an alternative way that change can be intentionally accomplished in organizations.

We study non-monetary markets where objects that arrive over time are allocated to unit-demand agents with private types, such as in the allocation of public housing or deceased-donor organs. An agent’s value for an object is supermodular in her type and the object quality, and her payoff is her value minus her waiting cost. The social planner’s objective is a weighted sum of allocative efficiency (i.e. the sum of values) and welfare (i.e. the sum of payoffs). We identify optimal mechanisms in the class of direct-revelation mechanisms. When the social planner can design the information disclosed to the agents about the objects, the optimal mechanism has a simple implementation: a first-come first-served waitlist with deferrals. In this implementation, the object qualities are partitioned into intervals; only the interval containing the object quality is disclosed to agents. When the planner places a higher weight on welfare, optimal disclosure policies become coarser.

We study the socially optimal level of illiquidity in an economy populated by house- holds with taste shocks and present bias with naive beliefs. The government chooses mandatory contributions to accounts, each with a different pre-retirement withdrawal penalty. Collected penalties are rebated lump sum. When households have homoge- neous present bias, β, the social optimum is well approximated by a single account with an early-withdrawal penalty of 1 − β. When households have heterogeneous present bias, the social optimum is well approximated by a two-account system: (i) an account that is completely liquid and (ii) an account that is completely illiquid until retirement.

In March 2020, Alex Funk, Deal Team Lead at Palladium Equity Partners, LLC, a private equity firm, was grappling with what to do with the student transportation company he had purchased just weeks earlier.

When he closed the deal with American Logistics Company (ALC) to acquire its subsidiary, ALC Schools, Funk was excited about the acquisition and eager to grow the company, which provided transportation to children with special needs. Operating largely in the Pacific Northwest, ALC maintained that it had no true competitors and was the industry leader in its market niche. ALC benefitted from federal and state laws which mandated that school districts provide transportation for children with special needs. Traditional yellow buses were often not suitable, and school districts found alternative options such as taxis unaffordable, providing a wide opening for ALC.

Palladium’s due diligence had confirmed ALC Schools’ attractive profitability, impressive operational prowess, and rapidly growing, recurring revenue from long-term, evergreen contracts with school districts. Funk planned to transform ALC into an Uber-like system, establishing a nationwide footprint. The fund was so enthusiastic about ALC’s prospects that it had purchased the firm at a multiple 30-40% higher than its preferred range.

But in March of 2020, the bottom fell out. Schools across the United States were shutting down due to the COVID-19 epidemic and nobody knew when they would re-open. ALC’s revenues dropped to zero. Funk had to come up with a plan to deal with the acquisition. He knew he had three options: kill (take the loss; sell off ALC Schools’ assets and liquidate); freeze (continue funding ALC Schools at minimal levels and wait out the pandemic); or build (invest in growth opportunities despite the pandemic).

Disruptions to government contracts traditionally arise during federal shutdowns when Congress fails to appropriate necessary funding. However, recent shifts in federal real estate policy, marked by lease cancellations and non-renewals, challenge the long-standing perception of federal leases as a secure and stable investment. We investigate how federal lease cancellations impact the pricing of government contract risk premia. Using unanticipated Department of Government Efficiency (DOGE) cancellations as a shock to commercial mortgage default risk, we find that first-loss CMBS bond tranches directly linked to DOGE-notified leases experience a persistent 4% drop in price, with large, negative spillover effects to bond prices, delinquency rates, and rental cash flows tied to nearby public and private-tenant leases. These results reflect that early termination options were previously perceived by investors as a dormant clause. Applying arbitrage pricing models of commercial lease contingencies confirms the underpricing of risk associated with government tenants.

Geoeconomics is the use of a country’s economic strength to exert influence on foreign entities to achieve geopolitical or economic goals. We discuss how concepts of power in the political science and economics literature can be used to guide research on geoeconomics. Economic threats as a form of coercion have seen a recent resurgence. We show how different types of threats can be modeled using simple tools and discuss what channels their potential effectiveness is based on. We discuss important open questions for the future literature to pursue.

The rapid rise in the number of PE funds searching for acquisitions across various sectors and size ranges has greatly enhanced exit options for private owners of businesses. The pool of potential buyers now extends far beyond competitors or other corporate acquirors. Significant competition among PE firms under pressure to invest large pools of capital has increased average purchase multiples to historic highs.

Business owners should not assume, however, that heated PE competition correlates to a high probability of their own successful sale. There is much at stake for business owners in launching a process to market their company for sale. Private owners tend to underestimate the substantial time, money, and distraction involved in a transaction process and may likely overestimate the probability of success. Additionally, the investment bankers pitching their services have an inherent conflict in providing advice regarding the sale process. An investment banker's livelihood depends on earning fees contingent on the completion of a sale, and they have less at stake in convincing an owner to start a process. As a result, a private business owner may likely be inadequately informed and not fully comfortable in making this momentous decision.

The purpose of this note is to help business owners make informed decisions as they contemplate a sale process. I highlight five fundamental questions a seller should consider before initiating a sale process:

• Am I ready for this huge transition?

• Do I fit what a buyout firm wants?

• Should I launch a sale process?

• How should I prepare for a sale process?

• How do I avoid getting taken advantage of?

In collaboration with VolunteerMatch (VM)---the world's largest online platform for connecting volunteers with nonprofits---we designed and implemented a new display ranking algorithm. VM's original ranking algorithm was intended to maximize efficiency (i.e., the total number of connections), but as a consequence it repeatedly displayed the same few opportunities at the top of its ranking, effectively limiting access to volunteers for the other opportunities. To incorporate VM's desire for equity (defined as the weekly number of opportunities with at least one connection) along with efficiency, we propose a modeling framework for online display ranking in settings where it is important to manage the trade-off between the total number of connections and the equitable allocation of these connections. We take an adversarial approach in evaluating the performance of online algorithms and show that a class of algorithms that applies a penalty to opportunities after each connection provides a strong (and, in certain regimes, optimal) performance guarantee. Inspired by our theoretical results yet mindful of practical considerations on VM's platform, we propose SmartSort, a simple score-based ranking algorithm which enjoys comparable guarantees in many regimes. We implemented SmartSort in two experiments, covering Dallas-Fort Worth and all of Southern California. Using a difference-in-differences analysis, we find that the implementation of SmartSort led to a 8-9% increase in the weekly average number of opportunities with at least one connection (consistent across both experiments) without any meaningful decrease in the total number of connections, implying a Pareto improvement for VM. Based on the success of our experiments, SmartSort has now been deployed nationwide. If SmartSort has a similar distributional effect on a national scale, every year, an additional 30,000 connections will go to opportunities that would have otherwise lacked access to volunteers.

A common-sense moral intuition is that bad acts should be condemned according to severity. Yet seven experiments (N = 6,075 U.S. adults) show that the extent to which people differentiate between transgressions hinges on the direction of comparison. When scaling up from a less severe transgression to a more severe one, people readily express stronger condemnation of the worse transgression. But when scaling down from a more severe transgression to a less severe one, they differentiate less, often condemning the lesser transgression just as strongly as one that is transparently worse. Indicating that one transgression is less bad than another can be construed as downplaying such transgressions, signaling bad moral character. Supporting this account, the asymmetry is larger for judgments that implicate moral character and for transgressions that seem especially important to condemn. Observers’ moral-character judgments reveal a similar pattern, suggesting that the asymmetry is reinforced by social incentives.

Over the first two decades of the 21st century, Rio Tinto, a 150-year-old global mining leader, faced significant volatility as it navigated an increasingly globalized and financialized economy. Mining companies, heavily reliant on commodity prices, struggled after the 2008 Great Recession, leading to cost-cutting measures and changes in how they managed their global operations.

In May 2020, Rio Tinto legally blasted two sacred, 46,000-year-old caves at Juukan Gorge in Western Australia to access $135 million worth of iron ore. The decision deeply distressed the Traditional Owners, who had long opposed the action, and sparked widespread criticism from the government, investors and communities. The fallout led to the resignation of Rio Tinto’s CEO, two senior executives, and the board chair. Jakob Stausholm, formerly CFO, became CEO in January 2021. This case provides background and traces the events that precipitated Rio Tinto’s decision to blow the Juukan Gorge caves—and the ensuing stakeholder backlash.

Online platforms are rife with racial discrimination, but current interventions focus on employers, rather than customers. We propose a customer-facing solution: changing to a two-point rating scale (dichotomization). Compared with the ubiquitous five-star scale, we argue that dichotomization reduces modern racial discrimination by focusing evaluators on the distinction between ‘good’ and ‘bad’ performance, thereby reducing how personal beliefs shape customer assessments. Study 1 is a quasi-natural experiment on a home-services labour platform (n = 69,971) in which the company exogenously changed from a five-star scale to a dichotomous scale (thumbs up or thumbs down). Dichotomization eliminated customers’ racial discrimination whereby non-white workers received lower ratings and earned 91 cents for each US dollar paid to white workers for the same work. A pre-registered experiment (study 2, n = 652) found that the equalizing effect of dichotomization is most prevalent among evaluators holding modern racist beliefs. Further experiments (study 3, n = 1,435; study 4, n = 528) provide evidence of the proposed mechanism, and eight supplementary studies support measurement and design choices. Our research offers a promising intervention for reducing customers’ subtle racial discrimination in a large section of the economy and contributes to the interdisciplinary literature on evaluation processes and racial inequality.

We characterize profit-maximizing menus in screening settings in which the agent has one of two privately-known types. We assume that utilities are quasi- linear but impose no other restrictions (such as increasing differences) on the agent’s utility or the set of alternatives. Our characterization clarifies the role of increasing differences in the standard setting and shows when random menus are beneficial. We describe applications to vertical and horizontal differentiation and multi-product bundling.

How should the federal government respond to the collapse of the Silicon Valley Bank (SVB), the 16th largest bank in the United States?

SVB collapsed on Friday, March 10, 2023, after an unprecedented run on deposit during which customers requested $42 billion of withdrawals in a single day. The value of SVB’s securities portfolio had declined severely. The bank had bet on long-term government securities, and when the Fed raised interest rates, the mark-to-market value of this asset had dropped precipitously. Ninety-Four percent of the bank’s deposits were uninsured, so customers reacted quickly when rumors spread that the bank might be insolvent.

The Federal Deposit Insurance Corporation (FDIC) had taken the unusual action of closing the bank on a Friday morning. Now, the Biden administration, the Federal Reserve and the FDIC had the weekend to decide what to do and coordinate a response. SVB’s collapse had sent shock waves through the banking sector. A few other banks with a large percentage of uninsured deposits and exposure to the hike in interest rates were rumored to be in trouble. Financial pundits were raising the question of whether this might lead to a financial crisis as occurred  after the failure of Lehman Brothers in 2008. There were also the SVB depositors who had not been able to get their money out of the bank in time – many of them tech start-ups that were a major driver of the economy.

Senior policymakers at the Treasury, the Fed, and the FDIC faced decisions that were politically fraught. Obviously, a recession would be politically unpopular. But the decision in 2008 to provide subsidies to banks had also proved contentious. Both the left and right had decried what they perceived to be “bank bailouts.”

Policymakers knew they had to react quickly. Before them, they had a number of options that could limit the fallout from SVB’s collapse. They also had to devise a communications strategy that calmed the markets and the public. Finally, they had to start the long process of examining what went wrong at SVB in order to ensure that it did not happen again.