Benjamin Kessler

Costello Research Sustainable Operations

Costello Research Evaluating Performance

Costello Research Non-Financial Disclosure

Management Faculty Research

Costello Research Digital Platforms

Planting fake customer reviews hurts businesses big time

Katelynn C Hipolito — Tue, 18 Aug 2026 15:13:22 +0000

Planting fake customer reviews hurts businesses big time Katelynn C Hipolito Tue, 08/18/2026 - 11:13

The capital of any business comes in (at least) two categories.

There’s the kind that has a price—land, cash holdings, etc.—and a less tangible, priceless kind. Reputational capital falls squarely into the latter category. Try as they might, companies cannot buy positive public opinion, even with the largest of marketing budgets. But that doesn’t prevent some from engaging in dark marketing tactics, such as planting or buying fake “customer reviews” on influential platforms like Yelp and Amazon. Crossing that line can backfire if the company is caught. When its ethical lapses are brought to light, the company undergoes reputational damage, essentially auto-cannibalizing some of its intangible capital.

Yi Cao, assistant professor of accounting at Costello College of Business at 91��. Photo provided by Yi Cao.

And that bite stings more, and for longer, than you might think, suggests a recently published working paper from Yi Cao, assistant professor of accounting at Costello College of Business at 91��. The research plumbs the financial consequences of flagged false reviews on crowdsourcing platform Yelp, a key reputational broker for local businesses throughout the United States.

The paper was co-authored by Sean Wang (formerly of Southern Methodist University), John Bai of Hong Kong Polytechnic University, and Chi Wan of San Diego State University.

To punish businesses that it deems guilty of gaming the customer-review system, Yelp places a prominent “Consumer Alert” banner on all of their associated listing pages. These are typically 90-day penalties. (As of 2023, more than 4,900 businesses had been disciplined on Yelp in this way.)

“We don’t know every signal Yelp uses to identify fake or paid-for reviews, but from what we know, Yelp relies substantially on reports from consumers and business owners,” says Cao. “There are also reports that they also have an algorithm running in the background to screen reviews for AI-generated language—which is a sign of possible review inflation.”

The researchers accessed Yelp data for the years 2019 to 2024—288,426 firm-month observations in all, including 16,837 observations of flagged firms and their demand-side peers, during the six months surrounding an alert. They analyzed this data-set alongside monthly foot traffic reports for the observed businesses from global location data firm SafeGraph. For some of the businesses, they were also able to obtain credit card transaction data through Consumer Edge, a customer intelligence company.

Cross-referencing these three data sources, the researchers traced the rise and fall of businesses fingered for false reviews. The general demand premium for businesses with a high proportion of four- and five-star reviews equates to about 2.9 percent more foot traffic than lower-reputation rivals in their local area. This difference in demand is one objective way to measure the material benefits of intangible capital.

“This is a classical theory of a trade-off between the benefits of manipulating reputation and the risks of getting caught,” says Cao. “When the risk is relatively low, the business owners decide the incremental benefits are worth the risk.”

Once publicly censured by the platform, however, businesses endured a surprisingly durable—and measurable—dip in demand. Almost immediately upon exposure, foot traffic went down by eight percent on average relative to matched peers, a stark indicator of reputational collapse that was mirrored in the credit card transaction data (an even better reflection of bottom-line outcomes). The decline was greater under conditions of so-called “intermediation intensity,” such as holiday periods when consumers were more reliant on the intermediary (Yelp) to help them plan their free time.

What’s more, the demand penalty outlasted not only the 90-day alert itself, but also (in many cases) the multi-year observation period of the research study. On the whole, foot traffic estimates for the censured group of businesses were still well below pre-alert levels even after a year and a half, implying recovery was nowhere on the horizon.

Part of the problem, the researchers found, is that the manipulated-review alert instantly degrades the informational environment on Yelp around that business. In other words, fewer reviews are posted, and those that do go up tend to be more negative on average. This makes it harder for businesses to rebuild their reputation by offering positive trustworthy signals.

For Cao, the research points to a basic conundrum that has worsened in the digital age: Once depleted, intangible capital is very difficult to restore. “Businesses nowadays heavily rely on digital platforms such as Yelp and YouTube,” he says. “Because the information reaches so many different corners, they are so widespread that you cannot ignore them. This increases not only the benefit of high performance, but also the reputational penalty if you are found manipulating your content.”

In This Story

Yi Cao

Topics

Accounting - Costello

Costello Research Retail

Costello Research Social Influence

Accounting Faculty Research

Costello College of Business

Sustainability commitments have serious consequences

Katelynn C Hipolito — Fri, 07 Aug 2026 17:29:09 +0000

Sustainability commitments have serious consequences Katelynn C Hipolito Fri, 08/07/2026 - 13:29

Amid increasingly stiff competition for capital, one way for a business to stand out is to make strong social impact pledges.

Yi Cao, assistant professor of accounting at Costello College of Business at 91��. Photo provided by Yi Cao.

For example, companies can voluntarily commit to environmental impact goals in hopes of attracting sustainability-minded investors and lowering cost of capital. However, it isn’t easy to tell whether these commitments, in and of themselves, make a difference to the marketplace—or, for that matter, to the planet.

“In regular settings, it’s really hard to evaluate the real economic effect of providing environmental commitments, and also establish that in making the commitment, the company is catering to real stakeholder preferences as opposed to their own idiosyncratic preferences,” says Yi Cao, assistant professor of accounting at Costello College of Business at 91��.

Cao’s paper in Review of Accounting Studies capitalizes on a unique setting that helps isolate the impact of environmental commitments. It was co-authored by John (Jianqiu) Bai of Hong Kong Polytechnic University, Xiumin Martin of Washington University in St. Louis, and Chi Wan of San Diego State University. In 2018, leading crowdfunding platform Kickstarter added an optional “Environmental Commitments” section to project pages where creators could post about their sustainable practices (e.g., responsible sourcing, recyclable packaging, and/or low-emission manufacturing). The new feature was rolled out gradually across product categories, enabling the researchers to compare not only project-to-project differences, but also overall changes within entire Kickstarter verticals.

The researchers scraped information from Kickstarter pages to form a data-set comprising 173,874 observations of unique projects during the years 2016 to 2021. They also gathered data on a host of outcome variables, such as whether the project was successfully funded, how much money each project ultimately raised, how long it took for fully funded creators to deliver products to backers, etc.

The researchers found that, on average, about one-third of creators chose to declare an environmental commitment once doing so became possible in their category. Several of the largest categories (e.g., Design, Technology and Games) saw significantly higher take-up rates.

On the whole, projects that included environmental commitments were 13.2 percent more likely to be funded and increased the pledged amount relative to the funding goal by 8.7 percent.

Using Google BERT, a natural language processing model, the research team separated projects into two categories: those with largely local appeal, and those with a national market base. “We pre-trained the AI with all projects from 2010 to 2016 that had at least 50 backers, and measured the geographic dispersion of these projects’ backers traced by their IP address,” Cao explains.

The purpose was to spare the human researchers the tedious and painstaking work of hand-classifying every project in the data-set. The BERT results showed that national projects with eco-commitments received 62.2 percent more backers from more environmentally aware states—that is, states whose residents reported above-average levels of concern about climate change on the Yale Climate Opinion Maps—than local projects that had made eco-pledges.

For Cao, these findings add crucial nuance to the general idea that pledges of social responsibility help attract capital. In the Kickstarter example, projects of national scale benefited much more from eco-commitment because they were pitched to a larger and more diverse backer base, including environmentally aware consumers from other states.

Additionally, projects offering physical products, as opposed to intangible services, confronted tougher economic trade-offs resulting from e-commitment. “First and foremost, the Kickstarter creators want to know whether [e-commitment] will benefit them,” Cao says. “Because this is pretty costly, based on our analysis. It delays delivery and increases costs, because recycled and other eco-friendly materials, for example, are a little bit more expensive.”

The fact that entrepreneurs were clearly grappling with these economic challenges suggests that e-commitment amounts to more than mere “greenwashing.” “We find consistent evidence that when these small businesses are trying to access environmentally aware consumers through e-commitment, they do incur cost. It isn’t cheap talk. And that means that consumers are attracted by a genuine commitment, rather than lip service or a nice-sounding pitch,” Cao says.

In sum, Kickstarter’s e-commitment feature was a net improvement for the ecosystem, giving creators (especially those with national aspirations) access to an untapped source of consumer demand—namely, environmental awareness. At the same time, creators had to weigh the financial benefit against the increased costs of making good on environmental commitments.

Cao suspects that the same basic rules might apply to other types of pledges related to issues that animate large groups of consumers. “For example, AI is especially polarizing nowadays. You could appeal to consumers or investors who are polarized against the technology with a declaration that no AI was used in the creation of a product, etc. It could be anything where there are differing preferences, and a broader impact on the social perception of those preferences,” he says.

In This Story

Yi Cao

Topics

Accounting - Costello

Costello Research Non-Financial Disclosure

Entrepreneurship - Costello

Costello Research Sustainable Operations

Costello College of Business

Sustainability

entrepreneurship

Why quick responders have the edge on the job hunt

Katelynn C Hipolito — Wed, 22 Jul 2026 16:25:41 +0000

Why quick responders have the edge on the job hunt Katelynn C Hipolito Wed, 07/22/2026 - 12:25

In some settings, playing hard-to-get may telegraph value. But that doesn’t neatly translate to the labor market, says Einav Hart, assistant professor of management at Costello College of Business at 91��.

Einav Hart, assistant professor of management at Costello College of Business. Photo provided by Einav Hart.

“Faster replies yield a higher likelihood of being hired, across different types of employers and employment—and every minute can count,” Hart says.

Hart’s recently published paper in Management Science draws upon both real-world and experimental studies, confirming a dramatic “speed premium” for job candidates.

The paper was co-authored by Eric M. VanEpps of Vanderbilt University, Ovul Sezer of Cornell University, and On Amir of UC-San Diego.

The researchers analyzed more than 11 million exchanges from freelancer marketplace Fiverr, which included pre-hiring conversations between employers and service providers.

The results showed an inverse correlation between the amount of time it took for a service provider to respond to an employer’s message and that provider’s likelihood of being hired. Even a one-hour delay in replying nearly halved a freelancer’s chance of being hired, and replying after more than 24 hours was equivalent to not replying at all. This negative effect of delay held even when only one freelancer was contacted for the position, showing that it is not simply about availability but about what speed signals.

“Faster replies yield a higher likelihood of being hired, across different types of employers and employment—and every minute can count.”

— Einav Hart, assistant professor of management at Costello College of Business at 91��

“Across every dimension we examined—job type, project size, prior platform experience, technology, and more—even a brief delay of five to ten minutes could dramatically lower the likelihood of hiring the provider,” Hart says.

Experimental studies with over 8,600 participants replicate this effect and illuminate a key reason for it: “Faster replies yield a higher likelihood of hiring because reply speed serves as a signal of employees’ future responsiveness,” says Hart. “If you respond to me very quickly, I expect that you will also respond to my needs and concerns if I hire you, which is a big signal of hireability, and increases my desire to work with you.”

Experimental study participants instructed to act as “employers” were more likely to hire providers who responded more quickly (e.g., after one hour versus several hours), even in specialized professions such as catering and medicine. Moreover, participants rated providers who responded quickly as more warm, competent, and more likely to be responsive in the future. Thus, speed acted as a signal of the provider’s characteristics that, in turn, affected hiring decisions. In other words, it wasn’t the speed per se that swayed employers; instead, it was being used as a proxy for other qualities deemed highly desirable in an employee.

Moreover, one of the studies showed that messages with highly responsive language (e.g., “I’ll make sure to stay in close contact”) produced a higher hiring likelihood, even when delayed, than a more neutrally worded message fired off quickly. This suggests that the content of communications can help compensate for a modest delay. Still, the same content sent faster always led to higher hiring probability, underscoring the speed premium.

Although the paper largely focused on contract-based employment, Hart expects her findings would generalize to more permanent, full-time employment. “Certainly, speed is not the only signal,” she says. “Over a longer relationship, the more you know about a person, the less weight speed might get in those decisions. However, both contingent and potentially long-term hiring begin from the same place of uncertainty on the part of the employer. In fact, if you’re hiring someone to be employed indefinitely, it’s all the more important for them to respond to your needs in a timely manner.”

These findings may tempt job-seekers to automate their responses to be fast. However, Hart argues against this in her recent Harvard Business Review article (co-authored by VanEpps). Automating responses or using AI to respond quickly is unlikely to be beneficial: When Hart’s study participants suspected the replies were automated or otherwise not genuine, there were no benefits to speed.

"An algorithm's reply speed does not signal anything about the employee's warmth, competence, or responsiveness and so, a fast auto-reply can even backfire and make you less hireable,” says Hart.

To distinguish themselves in a competitive labor marketplace, Hart recommends that job-seekers “take their time to craft your materials, prepare your application, and consider your future responses if and when you are later contacted. But once you get that first bite from an employer, that’s when the clock starts and speed starts to be a much bigger signal of responsiveness and thus of higher hireability.”

In This Story

Einav Hart

Topics

Management Faculty Research

Future of Work and Leadership - Costello

Costello Research Careers

Costello Research Recruiting

Costello Research Evaluating Performance

A.I. and Innovation - Costello

EV test drives are supposed to boost adoption—but they can backfire

Katelynn C Hipolito — Tue, 14 Jul 2026 14:38:54 +0000

EV test drives are supposed to boost adoption—but they can backfire Katelynn C Hipolito Tue, 07/14/2026 - 10:38

One of the biggest obstacles to wider adoption of electric vehicles (EV) is completely invisible and unquantifiable. It’s the belief in many consumers’ minds that the range restrictions of EV make it a bad option for them. Doubts about mobility, whether or not they are well-founded, can run neck-and-neck with curiosity in the minds of consumers, keeping them from making the switch from ICE (internal combustion engine) to EV.

Hang Ren, professor in the information systems and operations management. Photo provided

These doubts are beyond the remedial reach of automakers and sustainability-minded policymakers. Within the EV ecosystem, auto dealerships are the ones most directly responsible for allaying buyers’ concerns. Newly published research by Ioannis Bellos and Hang Ren, professors in the information systems and operations management area at Costello College of Business at 91��, focuses on one of the main levers at the dealers’ disposal: extended test drives. Bellos and Ren’s paper in Production and Operations Management was co-authored by Vishal Agrawal of Georgetown University.

Test drives have become a standard addition to the EV dealer’s toolkit. For example, BMW's Extended Test Drive Program lets customers use an EV for several days, and Volvo's Test Drive+ lends cars for up to four days. In some cases, government-sponsored initiatives such as Drive Electric USA (a partnership with the U.S. Department of Energy) have worked at the local level to facilitate test drives intended to educate consumers about EV.

“A car’s advertised range is public, but the range a driver actually gets depends on personal habits and conditions they can only learn through hands-on experience—which is why dealer demonstrations like extended test drives matter,” say Bellos and Ren. In other words, test drives can prevent range-based doubts from overtaking curiosity. Indeed, if consumers are pleasantly surprised by their experience with the EV, they can leave doubt in the dust, as they zoom ahead all the way to an accomplished sale.

But it can also work the opposite way, as the researchers found using a game-theoretic model of plausible buying conditions. Included in the model were buyers with a range of mobility needs. In the hypothetical scenario, consumers’ natural (pre-test drive) inclination was to favor EV if their needs were high, since high-needs drivers spend more on fuel and EV is gas-free. As in real life, however, even these simulated consumers would not purchase an EV if their overall impression was unfavorable.

Ioannis Bellos, professor in the information systems and operations management. Photo provided

“Offering demonstration service does not always mean that you’ll be convincing the customer to switch to EV,” Bellos says. “As a matter of fact, it may do the opposite—if the test drive convinces the customer that the achievable range is not enough to meet their need.”

The reseachers’ model shows that as EV technology begins to mature, becoming cheaper to produce and thus more profitable, the dealer’s incentive is to offer as many extended test drives as possible to boost sales. But once production costs fall below a certain threshold, the profit potential of EV becomes so high that dealers would be wise not to risk turning a percentage of customers away (or toward conventional ICE vehicles) by offering test drives.

In addition to implications for dealer profits, this dynamic has environmental significance. “If we consider two customers—one who takes a test drive and decides EV is not a good fit, and another who is convinced to purchase an EV after a test drive—the net effect on adoption is zero. They cancel each other out,” says Ren. “Yet the net emissions can increase, because that first customer probably drives more—that’s the reason they were interested in EV in the first place.”

Test drives, then, can be net-negative for the planet to the extent that they deter heavy-use drivers from switching to EV by resolving their range-based doubts.

The researchers also found that test drives can negate the impact of tightening CAFE standards from the U.S. Department of Transportation, which set mandated fuel-economy levels for all automakers. One result of CAFE standards is to encourage production of more fuel-efficient EV as opposed to ICE vehicles. CAFE does not apply to dealers, however. As higher fuel-economy standards pressure automakers to raise wholesale prices for ICE vehicles, the expected switch from ICE to EV is inhibited by the unfavorable impression left in some consumers’ minds by the test drive. The researchers found this can flip the policy's intended effect: Without test drives, tougher CAFE standards reliably push EV adoption up, but once dealers offer test drives, a strict-enough standard can actually leave adoption lower than it would otherwise be.

Bellos and Ren conclude that “dealer demonstration is a powerful but double-edged lever that must be deployed strategically.” The environmental and financial benefits of try-before-you-buy are likely tied to the EV market’s growing pains, and may diminish as it matures.

In This Story

Ioannis Bellos

Hang Ren

Topics

Costello Research Sustainable Operations

Costello Research Business Model Innovation

Costello Research Retail

Machine learning teaches asset traders not to sweat the small stuff

Katelynn C Hipolito — Mon, 15 Jun 2026 14:40:20 +0000

Machine learning teaches asset traders not to sweat the small stuff Katelynn C Hipolito Mon, 06/15/2026 - 10:40

Financial markets are governed by a combination of rational and irrational forces, statistical probabilities, and “animal spirits.” It takes fluency in both to understand the market, let alone to beat it. Yet market actors, including asset traders, now frequently use machine-learning techniques to help generate predictions of future asset prices.

Bo Hu, assistant professor of finance at the Costello College of Business at 91��. Photo provided by Bo Hu.

Scholars such as Bo Hu, assistant professor of finance at the Costello College of Business at 91��, are researching how these machine-learning tools are changing the decision-making processes that move the market, for better or worse.

The subject of Hu’s recent paper in Management Science is a well-known machine-learning technique called LASSO (least absolute shrinkage and selection operator), which has been widely adopted by financial practitioners since its introduction in 1996 by statistician Robert Tibshirani.

“If you look at the original paper, it describes an approach created by adding a regularization penalty to the least-squares regression method,” Hu says. Translation: “The power of LASSO is that it can screen out (i.e., penalize) weak signals while capturing stronger, potentially profitable ones. A LASSO-type trading strategy involves an ‘inactive zone’ for smaller-scale activity, in which the trading strategy is to do nothing.”

The paper was co-authored by Wen Chen of Texas Tech University and Liyan Yang of the University of Toronto.

Despite LASSO’s popularity and power, the soundness of its economic rationale remains unclear. Traders are presumably seeking any edge, however small, in the pursuit of outsized returns. How could it make sense for them to adopt a system designed to relegate signals of lesser magnitude to an ignored “inactive zone”?

To resolve this question, the researchers developed a theoretical framework to model a financial market in which multiple agents (read: traders) use an asset’s price history to forecast its return and make trading decisions.

In the benchmark case, when traders know the trading environment and do not face model uncertainty, they act according to an alternative to LASSO known as MSE (mean squared error). “MSE is essentially a Bayesian learning approach grounded in economic rationality,” Hu says. “It means that rational agents use Bayesian learning to update their beliefs and design their trading strategies. That stands in stark contrast to LASSO estimation, which filters out weak signals.”

However, the researchers found that when traders faced substantial ambiguity about the distribution of asset values, the trading calculus shifted. Ambiguity-averse agents will adopt a robust, LASSO-like strategy, refraining from trading in response to weak or intermediate market signals. With linear constraints imposed on the allowable trading strategies, the equilibrium decisions exactly matched LASSO estimates.

As an equilibrium trading strategy, LASSO can improve aggregate profits relative to the edge-seeking Bayesian alternative in the benchmark case, because the more conservative positions dictated by the “inactive zone” soften competition among traders. In a large market, aggressive competition drives the aggregate profits of traders using the conventional MSE strategy toward zero. By trading less aggressively, LASSO traders may preserve positive aggregate profits—a mechanism the researchers describe as “implicit collusion,” even though the traders do not communicate or explicitly coordinate.

However, Hu underscores that LASSO’s usefulness as a hedge against ambiguity depends on how well traders’ beliefs match the market’s true distribution of risk. The profitability of a LASSO strategy therefore hangs in the balance between traders’ biases and their enhanced market power (due to LASSO’s conservatism).

This balance is especially important when traders must distinguish between temporary fads and persistent trends. “Recently, the semiconductor index had a record-breaking run of consecutive gains. That’s very strong momentum, but if you are a contrarian, you might want to bet against that trend,” Hu says. “A trader must decide whether to follow the trend or take a contrarian position. LASSO’s inactive zone can help prevent overreaction to weak evidence, but it may also delay action when an emerging trend is genuine.”

There is also the possibility that a LASSO strategy could increase market volatility when combined with the other objectives and constraints of market makers, including high-frequency traders. “These traders need to control their inventory,” Hu says. “If they follow a LASSO-type strategy with an inactive zone, they will accommodate the liquidity demands of the market until their inventory reaches a certain threshold. At that point, they may start trading like momentum traders and cause the liquidity dry-up in financial markets. This dynamic is part of what fuels incidents like the ‘flash crash’ in 2010.”

In This Story

Bo Hu

Topics

A.I. and Innovation - Costello

Costello Research Machine Learning

Costello Research Fintech

Finance - Costello

Costello Research Financial Crises

Costello Research Risk Management

Costello Research Business Model Innovation

Artificial Intelligence

91�� professor probes ‘Labubu economics’

Katelynn C Hipolito — Tue, 02 Jun 2026 15:03:55 +0000

91�� professor probes ‘Labubu economics’ Katelynn C Hipolito Tue, 06/02/2026 - 11:03

The billion-dollar Labubu phenomenon broke a cardinal rule of retail: Consumers need to know what they’re buying before they open their wallet. Most new Labubu sales took the form of “blind boxes,” where purchasers found out which type of doll they’d purchased only after the fact.

Zhechao Yang, assistant professor of information systems and operations management at Costello College of Business at 91��. Photo by Jeffrey Porovich/Costello College of Business.

Zhechao Yang, assistant professor of information systems and operations management at the Costello College of Business at 91��, says that the “blind box” strategy, or “probabilistic selling” in academic terminology, is a rapidly growing global phenomenon that goes way beyond Labubu. Her recently published paper in Manufacturing & Service Operations Management examines how, and why, this emerging strategy has worked for toymakers, car rental companies and booksellers alike—while delighting customers willing to pay for a surprise.

The paper was co-authored by Hongseok Jang of Tulane University and Xiajun Amy Pan of University of Florida.

The researchers used game-theoretic modeling to analyze the interactions of suppliers, retailers and customers in a market where a manufacturer sells through a retailer. In this realistic setup, a supplier offers an assortment of products ranging from high- to low-quality at wholesale prices, which retailers seek to sell to consumers at the highest possible profit. Either the retailer or the supplier can choose to initiate probabilistic selling (PS), selecting some mixture of high- and low-quality products to go in the “blind boxes.”

Yang and her co-authors noted that PS can improve profits for retailers and suppliers in a number of ways. “One is the market expansion effect,” Yang explains. “If there is an excess of high-quality products that consumers are not buying, they can be combined with low-quality products to form a new product, via PS…Another way is through strategic differentiation, essentially preventing the excess high-quality products from experiencing price convergence with lower-quality products.”

But PS’ profitability potential is far from uniform. As the paper clarifies, it can vary greatly with the context, based particularly on two elements: whether PS is retailer- or supplier-led, and whether high-value products are abundant or scarce.

The researchers found that transaction costs—the extra costs of creating and selling blind boxes—were one of the main determining factors. Whichever party initiates PS will have to bear certain costs, which may include the labor required to assemble “blind boxes,” and additional fulfillment, inventory and logistics costs. Additionally, Yang says, “There could be financial and accounting costs associated with adding new products.”

When retailers take it upon themselves to introduce PS, the associated transaction costs can reshape upstream pricing. Anticipating those costs, the supplier may lower the wholesale price of high-quality products to make the blind-box strategy viable for the retailer. This can help the retailer, but it also reduces the supplier’s margin and may make retailer-led PS unattractive to the supplier under some conditions.

When retailers take the lead, they will adjust the product mix for PS based on balancing transaction costs against the likelihood of product cannibalization—a trade-off that may give rise to more shortsighted choices.

On the other hand, supplier-led PS gives the supplier more control over both the blind-box mix and the wholesale terms. Also, by adjusting the number of high-quality products that will go in the “blind boxes,” the supplier can maximize price differentiation, thus increasing profits.

“When the supplier leads the strategy, it can reduce the channel inefficiency,” Yang says. “In the model, the supplier can control both the product mix inside the blind box and the wholesale terms offered to the retailer. That can make it easier to manage cannibalization and preserve the value of high-quality products.”

The importance of supplier-led PS becomes especially clear when high-value products are scarce. In that setting, the paper shows that firms should use the limited high-value capacity on “blind boxes” rather than sell those products separately.

Thus, the paper concludes that supplier-led PS, unlike the retailer-led variety, can create a “win-win-win” scenario where the supplier and retailer both earn higher profits, market coverage expands, and consumers benefit from access to a broader set of purchasing options.

However, power plays a significant role in real-world decisions around PS. “In reality, major retailers like Amazon may have more bargaining power, more pricing power. The high-level assumption we give this model is that when the retailer is more powerful, the retailer gets the first opportunity to decide whether to introduce PS., but that doesn’t mean it’s the best option for the whole supply chain,” Yang says.

“It’s better for the two parties to have some conversation before deciding who will take the power at the very beginning. Large retailers may have some advantages, but sometimes it’s better to give this power to the supplier.”

In This Story

Zhechao Yang

Topics

Operations - Costello

Costello Research Retail

ISOM Faculty Research

Costello Research Careers

The mindset shift that will move you from “manager” to “leader”

Katelynn C Hipolito — Wed, 29 Apr 2026 17:55:07 +0000

The mindset shift that will move you from “manager” to “leader” Katelynn C Hipolito Wed, 04/29/2026 - 13:55

Most managers are promoted because of their good individual performance. Such a mindset—that one’s own performance is the most important—is devilishly difficult to change, however, and employees suffer.

Kevin Rockmann. Photo by Jeffrey Porovich/Costello College of Business.

“It’s a hard lesson to learn, but the job of the manager actually is to help others do their job: Select them, guide them, help them, train them, but ultimately let them thrive and achieve. Not easy for someone who has reached the pinnacle by doing things themselves,” says Kevin Rockmann, assistant dean for research, professor of management, and the CGI Corporate Partner Faculty Fellow at Costello College of Business at 91��.

Rockmann’s research focuses on the brighter side of employee relationships—bonds that, without crossing any lines of propriety, promote trust and mutually beneficial collaborations. It is both common knowledge and common sense that organizations that prioritize such relationships benefit from greater employee engagement, higher retention, and enhanced productivity, among other positive effects.

No wonder business leaders wax rhapsodic about the importance of healthy working relationships between managers and employees. But Rockmann's research shows that in most cases, such talk exists in inverse proportion to the amount of attention managers actually devote to relationships and relationship dynamics. In the battle for managerial bandwidth, relationships routinely lose out to more immediate bottom-line issues (e.g., whether the "work" is getting done).

“It’s a hard lesson to learn, but the job of the manager actually is to help others do their job: Select them, guide them, help them, train them, but ultimately let them thrive and achieve. Not easy for someone who has reached the pinnacle by doing things themselves.”

— Kevin Rockmann, assistant dean for research, professor of management, and the CGI Corporate Partner Faculty Fellow at Costello College of Business at 91��

“Most of what bosses say about their ‘collaborative’ and ‘close-knit’ corporate culture is a form of gaslighting. And employees know it, which serves only to alienate them. That’s a major reason why disenchantment and disengagement are running rampant in today’s organizations,” Rockmann says.

In a recent executive development training involving a national fast-casual dining chain, Rockmann and his team took steps to address this. Participants were assigned to focus on the relational aspects of their job, intervene where necessary and document their results. Such outcomes not only were inspiring but also translate easily to any organization. “Giving systematic attention to relationships not only improved the interpersonal atmosphere in their units but also helped them personally and professionally in four main ways,” Rockmann says.

#1 Deeper, higher-level conversations. As Rockmann tells it, concentrating on relationships changed the nature of the participants’ conversations at work. “Instead of seeing things purely from their own individual point of view, they gained a more all-around perspective based on the input of others,” he says. Managers asked first for thoughts and ideas regarding problem areas rather than punishing or criticizing. This elevated their awareness from a limited focus on what most directly affected them to a broader mindset encompassing more of their context and environment.

#2 Less reactive decision-making. Relational attention separated stimulus from response, leading to less reactionary and more thoughtful responses during meetings and email exchanges. One regional manager was able to do this consistently with his people and found himself “not as angry” and that his people were taking greater ownership over their actions. As a result, decision-making became less knee-jerk, more deliberate, and ultimately more effective from a strategic standpoint.

#3 Reinventing restrictive roles. The new orientation enabled participants to engage in more intentional “relational job crafting,” whereby they began to embrace unprecedented partnerships and collaborations. One of the participants in HR connected with another in IT during the program and started to plan out trainings they could build together. Being together in a learning environment created possibilities for these types of solutions, simply by virtue of paying attention to interdependencies between individuals and teams.

#4 Escaping middle-manager entrapment. The participants were able to escape the trap that ensnares many a middle manager and prevents them from rising—namely, getting stuck playing caretaker or being the “answer man” for their teams. For example, one participant noted that her tendency was to solve problems right away, especially for other people. By relaxing this she was able to empower others in her team and grow their confidence. Forced to look outward and focus on how their teams were relating, they readily identified trusted and reliable team members who were natural candidates for greater responsibility. “This removed the sense of risk that often prevents managers from delegating day-to-day and, by extension, keeps them putting out fires instead of aiming higher. They moved from saying ‘fix things this way’ to asking ‘what do you think is the best way to address this issue?’, Rockmann says.

While these individuals in the exercise were in the food industry, Rockmann believes the leadership lessons to be learned here are universal.

Rockmann notes: “When managers take the focus off themselves and their own achievements so that they can pay close attention to others, they effectively step off the stage and enter the balcony. They remove themselves from the play and can see better what is going on around them. They become attuned and responsive to the complex relational environment and can confidently coach their people through it, ultimately bringing themselves, their teams, and their organization more success.”

Topics

Costello Research Teams

Costello Research Evaluating Performance

Management Faculty Research

Future of Work and Leadership - Costello

In This Story

Kevin Rockmann

Explore the Costello College of Business

Learn more about Research at George Mason

Can machine learning make the world a fairer place?

Katelynn C Hipolito — Wed, 22 Apr 2026 15:08:00 +0000

Can machine learning make the world a fairer place? Katelynn C Hipolito Wed, 04/22/2026 - 11:08

A paradox hovers over our increasingly AI-dependent world. On the one hand, artificial intelligence can make the world a better place (or so we’re told). On the other hand, algorithms have no imagination or consciousness, and thus can know only the status quo—as reflected in the data they are trained on. And our current world is far from perfectly meritocratic or fair.

Jingyuan Yang. Photo by Jeffrey Porovich/Costello College of Business.

Jingyuan Yang, assistant professor of information systems and operations management at Costello College of Business at 91��, suggests that the paradox is compounded by conventional thinking around AI. “The standard view is that fairness is a tax on efficiency. The way conventional systems are structured, fairness checks are added almost as an afterthought that is assumed to negatively impact system performance,” she says.

Is the “better,” optimized world of AI destined to replicate, or perhaps even exacerbate, existing inequalities? Yang’s ongoing research—in collaboration with Pengzhan Guo of Duke Kunshan University and Keli Xiao of Stony Brook University—points to an appealing alternative. It uses AI systems as a proving ground for a theorized “fairness-performance complementarity”—the idea that, under certain conditions, fairness and performance reinforce one another.

“Our 'fairness-by-design’ framework utilizes reinforcement learning, which is a type of machine learning (ML). But unlike most machine learning algorithms, ours includes multiple agents competing for finite resources in a dynamic environment, not a static one,” Yang says. “That makes our paradigm much more structurally similar to many real-world environments in which various people compete over time for finite resources.”

Fairness was integrated in two stages. First, the framework was designed to “nudge” high-performing agents towards exploratory choices that might maximize their rewards. As Yang explains, “In this framework, high-performing agents are held in an exploratory mode for longer, while lower-performing agents settle into stable paths sooner.” Second, options that were abandoned as a result of agents’ reward-seeking behavior were redistributed, with lower-performing agents getting first crack at the best opportunities.

As Yang summarizes, "The exploratory activity of the high performers releases opportunities that the system channels down toward the weaker performers. Theoretically, this increases fairness while retaining individual choice and without constraining performance.”

“Our ‘fairness-by-design’ framework utilizes reinforcement learning, which is a type of machine learning (ML). But unlike most machine learning algorithms, ours includes multiple agents competing for finite resources in a dynamic environment, not a static one. That makes our paradigm much more structurally similar to many real-world environments in which various people compete over time for finite resources.”

—Jingyuan Yang, assistant professor of information systems and operations management at Costello College of Business at 91��

To test out the framework, the researchers used a data-set comprising detailed information on the job histories of 6.5 million professionals across a 20-year timeframe. “In the real-world data, we see a high degree of disparity, without very much redistribution of elite opportunities from relatively advantaged to disadvantaged employees,” Yang says.

The algorithm converted the real-world job information into opportunities offered to hypothetical agents. The resulting career paths were analyzed in terms of both performance and fairness. Performance was defined by aggregate rewards earned by all agents across all periods. Fairness was defined by the degree to which initial performance disparities were resolved over successive decisions.

The “fairness-by-design” framework’s results—for both fairness and performance—were better than those of eight alternative ML methods drawn from three different methodological families.

The researchers also adjusted the system to account for people’s changing preferences. Early-career professionals tend to value employer reputation and advancement potential; in late career, rewards pertaining to job stability and security are more salient. Even with these restrictions implemented, the framework functioned as intended—improving the average quality of overall career paths while fueling upward mobility.

In a follow-up study utilizing the New York Yellow Taxi Trip record database, the framework was tasked with generating route recommendations to hypothetical “agents,” i.e. cab drivers, with varying performance records. In this domain, the choice-set was much smaller (263 locations, as compared to 4,282 companies), and the timeframe far shorter (two hours as opposed to 20 years). As with the career-planning example, the taxi study found that more equitable distribution of high-quality routes led to higher average income per minute for the system as a whole.

“Because the framework proved adaptable to different domains and agent preferences, we think it could be used in future as a governance mechanism for a variety of AI contexts,” Yang says. Health care scheduling, course registration in higher education and provision of digital services are a few areas Yang sees as likely candidates.

While emphasizing that her research is still ongoing, she argues that it poses a serious challenge to standard ways of thinking about AI. “Our formal proof establishes the conditions under which fairness and performance reinforce each other, and our experiments show those conditions are achievable in realistic settings. That gives our work both theoretical and experimental grounding," Yang says.

Learn more about AI at George Mason

Dive into Research at George Mason

In This Story

Jingyuan Yang

Topics

ISOM Faculty Research

Costello Research Artificial Intelligence

Costello Research Machine Learning

Costello Research Digital Platforms

A.I. and Innovation - Costello

Inside the competition for capital at some of the world’s biggest banks

Katelynn C Hipolito — Wed, 08 Apr 2026 16:08:43 +0000

Inside the competition for capital at some of the world’s biggest banks Katelynn C Hipolito Wed, 04/08/2026 - 12:08

As the U.S. economy becomes more consolidated, the strategic decisions of senior leaders at leading companies carry ever-greater weight. A lot is riding on how these companies are run, yet their day-to-day decision-making remains, in most cases, obscure.

Barbara Su. Photo by Jeffrey Porovich/Costello College of Business.

But the banking industry is an exception. As Barbara Su, assistant professor of accounting at Costello College of Business at 91��, notes, “Because the banking industry is heavily regulated, it allows us to have access to subsidiary banks’ financial information. We can observe how much money parent companies take from each subsidiary, as well as the internal capital allocation between subsidiaries by headquarters.”

Su’s forthcoming paper in Management Science finds that in multi-bank holding companies (MBHCs), especially those with multiple subsidiary banks or geographically dispersed operations, it’s essential to bridge the information gap between HQ and local subsidiaries. That is why corporate leaders strongly rely on internal accounting information, as a complement to financial performance, to help them make intramural investment decisions.

Su says that although recent rounds of deregulation and interstate banking consolidation have reduced MBHCs’ economic profile, studying their behavior is relevant “because they provide a clean empirical setting. With multiple legally distinct bank subsidiaries, we can directly observe how the parent allocates capital across units.”

“We often hear that bank holding companies should act as a ‘source of strength’ for subsidiaries. There is some truth to the ‘source of strength’ idea, but our paper also observes that holding companies don’t just allocate funds to save subsidiaries from failing. They’re also allocating funds for the better-performing ones, when the accounting reporting quality is good, to help them thrive.”

—Barbara Su, assistant professor of accounting at Costello College of Business

The paper was co-authored by Scott Liao of University of Toronto and Allison Nicoletti of University of Pennsylvania.

The researchers zeroed in on the subsidiary banks’ loan loss provision—an accounting adjustment reflecting anticipated loan defaults—as an especially meaningful indicator.

Their dataset included financial statement information from 3,031 bank subsidiaries affiliated with 799 bank holding companies, for the period 1996 to 2019.

Su’s hypothesis was that “the accuracy of the loan loss provision coming from the subsidiaries will, or should, guide the internal capital allocation at the holding company level.” In other words, HQ would be wise to interpret valid accounting information as a general quality index, helping them identify worthy targets for investment.

The evidence confirmed Su’s suspicions. The average accuracy of subsidiaries’ loan loss provisions was positively associated with HQ’s ability to pick winners, i.e., to invest in subsidiaries that would go on to achieve relatively high ROA or return on assets.

A follow-on study showed that this association was strongest for two subsets of holding companies: those more dependent on lending for their revenue, as well as those where HQ was most distant (either geographically or economically) from the local branches. Banks with troubled subsidiaries saw the weakest association between accounting information accuracy and wise internal investments. This amounts to further supporting evidence for the utility of loan loss provision as an in-house investment index for dispersed organizations, assuming the investment thesis has to do with picking winners rather than subsidizing struggling subsidiaries.

For Su, this research complicates commonly held beliefs about bank consolidation producing slow-moving behemoths getting by on sheer market power. “We often hear that bank holding companies should act as a ‘source of strength’ for subsidiaries,” she says. “There is some truth to the ‘source of strength’ idea, but our paper also observes that holding companies don’t just allocate funds to save subsidiaries from failing. They’re also allocating funds for the better-performing ones, when the accounting reporting quality is good, to help them thrive.”

Su also surmises that the decision-making method covered in her paper—HQ referencing internal accounting information as a gauge of quality—could apply outside banking. For example, it could come into play when evaluating subsidiary managers’ performance prior to a turnover decision.

Nonetheless, Su recommends that leaders maintain a balanced informational environment. Internal accounting information should be used to augment, not replace, standard performance metrics such as ROA. “I guess we could roughly think about it in a two-by-two matrix,” she says. “The ideal scenario would be high ROA, with high information quality. You could also have high ROA and low information quality—that means you’re cooking your numbers.”

Explore Costello's College of Business

In This Story

Barbara Su

Topics

Accounting - Costello

Accounting Faculty Research