Presentation Abstracts

Historically, minority entrepreneurs have been disadvantaged in their access and utilization of technology (Fairlie, 2004).  This is while access to technology can substantially alleviate other barriers (e.g., limited access to financial resources) confronting this community of entrepreneurs (Prieger, 2023). Accordingly, a pivotal step to solving inequalities in business entails making technology not only more accessible but also more usable for minority entrepreneurs, which in this research we refer to as “humanizing technology”. For entrepreneurs in general, humanizing technology involves ensuring that technology serves as a valuable tool that enhances their capabilities, fosters connections, and addresses their specific needs (Sciutti et al., 2018). Utilizing a sample of minority entrepreneurs participating in training program offered by an entrepreneurial support organization (ESO), we first clarify how technology can be humanized for minority entrepreneurs specifically. Some of the identified unique features of humanizing technology for minority entrepreneurs included customizing technology to their needs, providing one-on-one support and mentorship, and fostering a positive mindset and well-being. Using semi-structured interview, we then explored how minority entrepreneurs can fully utilize technology when their perspectives and experiences are considered in tailoring the application of technology to their business practices. The results demonstrated that access to the program which aimed to humanize technology help minority entrepreneurs improve their business digital presence. Moreover, the participating entrepreneurs reported increases in the number of customers served, sales, and earned income. Expanding business support networks and access to resources and partners were among other outcomes of participation in the program. Notably, all of the participants revealed a shared interest in ensuring that the digital improvement of their business operations also supported the well-being of the community.

This study has important policy implications. The findings of the research reported here emphasize the superiority of technological humanization, as opposed to mere technology access, in helping underrepresented entrepreneurs in various ways. Our research provided evidence that when these needs and challenges are understood and are woven into the support program, results are extraordinary. Thus, incentivizing programs that extend beyond providing resources to underrepresented entrepreneurs by helping them use resources effectively will increase their likelihood of success. For example, we highly recommend providing support for programs designed to supplement funding with training aimed to help entrepreneurs prioritize different business needs investments. Moreover, in our research, we identify ESOs’ critical role in humanizing technology for underrepresented entrepreneurs. For example, these organizations’ deep connection with the local entrepreneurial landscape makes them highly effective at providing entrepreneurs with critical resources. Thus, since ESOs understand the unique challenges facing local underrepresented entrepreneurs, provide targeted resources and guidance, foster an inclusive community, and advocate for systemic change, providing support to these organizations can enable them to help minority entrepreneurs overcome various barriers and access new resources. 

Even though entrepreneurship within a region and the region’s overall economic prosperity are considered to be interlinked, the exact nature of this relationship is less understood (Kimmit et al. 2020). Further, the relationship becomes more complex if issues of racial inequity and structural racism involving minority entrepreneur groups also come into play (Singh & Nurse, 2024; Song et al., 2020).

On the one hand, studies have revealed that ethnic minority entrepreneurs are in a “precarious situation” even among ethnic minority individuals, which makes them engage in low-skilled, low-growth, low-return sectors of the economy, such as retailing, restaurants, fast food, and personal services (Olarewaju, 2023; Suarez, 2016). Further, minority entrepreneurs and entrepreneurs of color face discrimination when they attempt to secure resources for starting their ventures (Conley & Bilimoria, 2022; Neville et al. 2018). On the other hand, the intersection of race, class, and gender actually helps shape individuals in minority groups - in the process of “becoming” entrepreneurs (Harvey, 2005). Other studies (e.g., Orozco, 2022; Vallejo & Canizales, 2016, 2023; Wingfield & Taylor, 2016) have suggested that minority group entrepreneurs carefully utilize “counterframes” (Wingfield & Taylor, 2016), the salience of “ethnic identity” (Orozco, 2022), including “social identity” at the intersection of race, class, and gender (Vallejo & Canizales, 2016), as well as “ethnoracial capitalism” (Vallejo & Canizales, 2023) to achieve their entrepreneurial ambitions, especially by harnessing a continuous process of adaptation, learning, and innovation (Wang et al., 2021).

However, the exact process through which an enactment of strategy to foster minority-driven entrepreneurship to achieve economic prosperity within a region has not been studied before. This is the focus of the present research. Utilizing an ecosystems lens (Scott et al., 2022; Spigel & Harrison, 2018), we study the evolution and growth of the entrepreneurial ecosystem in the borderplex region of the southern United States, comprising southern Texas, New Mexico, and the Juarez region of Mexico. We premise that minority (Latino/a) entrepreneurship is fostered if the ecosystem orchestration process focuses on building financial inclusion and capital formation. In the research context we study, two organizations have primarily contributed to the form of an ecosystem orchestration partnership effort, to intensify entrepreneurship efforts and move toward economic prosperity. These two entities comprise the entrepreneurship and incubation center of a public university located in the region and a local venture capital firm that supports technology-based startups in the region. Together, the two entities are expanding financial literacy and savings for economic gain and prosperity, by supporting a select number of technology-driven startups that support other local entrepreneurs to grow their businesses.

In the aftermath of George Floyd’s murder in 2020, multiple corporations aligned themselves with the #BlackLivesMatter movement by releasing statements, highlighting the movement in their advertising, and prominently displaying the #BlackLivesMatter hashtag on their websites. An important question is whether these displays of support translated into favorable outcomes for Blacks. Evidence of bias and discrimination towards Blacks has been largely documented in areas including labor markets, entrepreneurship, and online platforms and has proven challenging to mitigate. In this paper, we study if displaying the #BlackLivesMatter hashtag on a crowdfunding platform’s website had any effect on improving outcomes experienced by Black entrepreneurs.

The use and effectiveness of nudges in the form of visual cues have been proposed as an effective way to correct biased behavior. Although nudges may sometimes appear effective, they may not successfully solve underlying societal problems concerning bias or, in our context, discriminatory funding behavior. In June 2020, in response to the murder of George Floyd, Kickstarter changed the color of its logo to black and added the #BlackLivesMatter (#BLM) hashtag under it. This change lasted for over a year. We leverage this change as a natural experiment to study if including a nudge intended to reduce discrimination in a crowdfunding platform positively affected the outcomes Black entrepreneurs experienced. Moreover, we investigate whether this change affected Black entrepreneurs by examining its effect on their participation in the platform.

We use difference in differences to estimate the effect of #BLM on crowdfunding outcomes for projects on Kickstarter between May 2019 and June 2021. For each project, we labeled the founder’s gender and race using name classification algorithms. In alignment with previous studies, we find evidence of disparate outcomes for Black creators on Kickstarter, as Black creators are less likely to be funded on Kickstarter than non-Black creators. Given the underperformance of projects led by Black creators on Kickstarter, we evaluate whether the presence of #BLM improves funding outcomes for Blacks. We find no effect of #BLM on the crowdfunding performance of projects led by Black creators across several different performance variables. Notably, we find that #BLM had no effect on increasing the likelihood of projects led by Black creators getting funded or improving the aggregate influx of funding to Black-led projects. Instead, we find that the only measurable effect of #BLM was to increase the number of projects launched by Black entrepreneurs, suggesting that the inclusion of #BLM encouraged Black entrepreneurs to launch their projects on the platform.

This work highlights the need for systemic-driven initiatives over individual-level interface design changes in addressing societal problems such as bias towards Black creators seeking crowdfunding. Moreover, companies should think carefully about how nudges influence their stakeholders. In the context of crowdfunding, nudges may harm the communities that it intends to help. To effectively mitigate bias on online platforms, we need a better understanding of the underlying causes of such discrimination.

Introduction 

A recent workshop funded by the National Science Foundation (NSF) convened researchers, doctoral students, and accelerator facilitators in a sustained dialogue about diversity in entrepreneurship. The participants were united in their desire to build a more inclusive innovative entrepreneurial ecosystem. A core idea that emerged was that the lack of diversity in the entrepreneurial ecosystem begins long before someone goes to a university or applies to an accelerator. By altering entrepreneurship education today for very young (8-10 year old) people and their caregivers can help to make tomorrow’s ventures more diverse and representative of the larger population. Policy focused on informal entrepreneurship for very young people and their caregivers is likely to help develop a more diverse entrepreneurial ecosystem. 

Methods

The NSF workshop was guided by a practice theory approach to entrepreneurship and to diversity. Because studies informed by practice theories examine the (re)production of the resilient social order of an organization or an ecosystem, they offer the promise of more effective interventions.

The workshop embraced a reflective dialogue approach. By bringing practitioners into conversation with scholars, the participants were able to envision projects that could bridge the practice / theory gap and stimulate real change in the entrepreneurial ecosystem.

Highlights of the reflective dialogue (Formal results not available) Popular culture presents a racialized and gendered view of entrepreneurs. Investigating the ways that educational and popular cultures are interconnected with matters of diversity in entrepreneurship opens new questions about entrepreneurship-as-practice and suggests new models for entrepreneurial education. Understanding how entrepreneurial abilities are cultivated before individuals get to the point of participating in an accelerator requires the exploration of practices that shape the entrepreneurial futures of young people.

More studies are needed that explore the dynamic role(s) and methods of youth-oriented informal education, and examine the interdependent practices that shape learning opportunities about entrepreneurship for youth, K-12 teachers, afterschool program facilitators, and families. Policy 

Implications

Policy to support informal entrepreneurship education (via media, kinesthetic movement, and other means) for very young people is likely to help develop a more diverse entrepreneurial ecosystem. Similarly, policy to support entrepreneurial learning for caregivers of young people (e.g., teachers, after school program leaders, and parents) may spark a more diverse entrepreneurial ecosystem in both the near and long term.

Introduction 

Traditionally, most migrant businesses in the UK are low value businesses operating within the ethnic economy and are excluded from the mainstream economy (Edwards et al, 2016). However, recent wave of migration and improved socioeconomic conditions are changing this narrative. There are now emerging high value migrant businesses in the mainstream economy and especially in the digital economy. This research focuses on Black migrant entrepreneurs in the UK’s fintech industry. It explores the lived experiences of Black African migrant entrepreneurs navigating a predominantly white entrepreneurial space in the UK fintech industry. Existing studies have largely focused on the racial challenges of minority groups with fintech (Bhagat and Roderick, 2020); however, little is known about the specific racial barriers Black entrepreneurs experience and how they negotiate a non-inclusive fintech ecosystem. Understanding the challenges and specific barriers to fintech innovation among minoritized entrepreneurs is an important policy consideration for making the fintech ecosystem a more inclusive entrepreneurial space for Black and minority entrepreneurs.

Methodology 

This research employed a phenomenological approach to study the lived experiences of Black African entrepreneurs doing entrepreneurship in a predominantly white fintech ecosystem. To explore their experiences, we conducted six in-depth semi-structured interviews with first generation Black African migrant entrepreneurs who provide fintech and digital solution (including payment and banking solutions) in the UK financial services. These entrepreneurs run fintech businesses which include App development, financial platform services and digital payment solutions. At this preliminary phase of the study, the interviews were analyzed using thematic analysis.

Findings 

Preliminary findings suggest that Black African migrant entrepreneurs experience racial barriers along three major themes: (i) institutional barriers (ii) inclusion barriers and (iii) identity barriers. Institutional barriers are barriers Black African migrant entrepreneurs experience from the structural inequality in society and from policies and institutions that disadvantage minority groups. These include difficulty in accessing capital, regulation barrier, racial discrimination, and systemic prejudice. Inclusion barriers are barriers that prevents these cohort of entrepreneurs from gaining access to important entrepreneurial opportunities due to their racial and minority status. These barriers include exclusion from entrepreneurial opportunity such as important network, tokenism, lack of access to important segment of the fintech industry and lack of representation and belonging. Our research found identity barriers to include issues around intersectionality and gender. Because this space is pervasively white, other identity barriers that become heightened include accent and the entrepreneurs' names.

Policy Implications

Although our research still at a preliminary phase, it has potential to contribute to policy and engage with policy makers on the important issue of inclusion in the fintech industry. As the demographics and entrepreneurial landscape of the UK gradually shift from a predominant white society to a more diverse population, it is important fintech policies reflect these changes. Policies that promote inclusive regulation, diversity in venture capital funding and access to opportunities for minority groups will foster an inclusive innovation in the UK fintech ecosystem.

Business models (BMs) have grown into an important area of study in the fields of entrepreneurship, strategy, technology and innovation management, and e-commerce (Zott et al., 2011) due to their perceived utility as a communication, strategic planning, and implementation tool (Morris et al., 2005). Scholarly attention has increasingly been devoted to business model designs – defined here as gestalts or patterns of value creation-delivery-capture enablers, detractors, and their interrelationships – as they have been positively associated with firm performance (Zott & Amit, 2007) and impact firm performance in magnitude comparable to industry effects (Sohl et al., 2020).

The dominant view is Zott and Amit’s (2007) activity systems perspective, which introduces novelty, lock-in, complementarities, and efficiency (NICE; Zott & Amit, 2010) to describe the dominant drivers of value creation. Novelty-centered models are characterized by new ways of engaging in economic exchanges and may involve adopting new activities (content), novel approaches to linking activities (structure), and/or new approaches to governing activities (governance; Zott & Amit, 2007, 2010). Efficiency-centered models, in turn, are focused on improving transaction efficiency through reducing transaction costs between business model participants. While studies have begun to explore antecedents of business model design, knowledge gaps remain as to what they are and how they impact BM design and outcomes (Amit & Zott, 2015).

An adjacent line of inquiry investigates how TMT composition influences key properties of the business model (Knockaert et al., 2011; Ensley & Hmieleski, 2005) as well as team and firm outcomes more generally (Ponomareva et al., 2022). On one hand, the value-in-diversity hypothesis contends diversity within TMTs positively contributes to BM design novelty and firm financial performance due to enhanced information-processing and creative problem-solving capacities (Martins & Sohn, 2022; Cox et al., 1991). On the other, studies of homophily in organizations suggest positive outcomes for teams lacking diversity due to improved coordination, communication, and trust among members (Ertug et al., 2022). Miller et al.’s (2022) review suggests concerning theoretical assumptions and methodological practices, such as using indicators of demographic diversity to infer cognitive diversity, contribute to the inconsistent conclusions collectively presented across the two camps. The present study follows recent suggestions to study surface and deep-level diversity simultaneously (Miller et al., 2022; Jansen & Searle, 2021), and extends the field by considering the differential impact of team members’ participation in an entrepreneurship training program on team-level cognitive diversity.

A fitting setting to disentangle the effects of surface and deep-level diversity on business model design is the I-Corps™ program. Described in detail elsewhere (see Bosman & Garcia-Bravo, 2021; Nnakwe et al., 2018; Huang-Saad et al., 2017), the National Science Foundation (NSF) I-Corps™ program offers an immersive training experience in Lean Startup Methodology (LSM) to academic entrepreneurial teams engaging in technology commercialization. The study employs a one-group pretest-posttest design with data collected via a web-based survey administered in Qualtrics.

The findings bear implications for which dimensions of diversity policies aim to maximize to increase teams innovative output and firm performance.

Theoretical background 

Black women and girls are severely underrepresented in Science, Technology, Engineering and Mathematics (STEM) fields across both educational and career continuums. Their access to STEM pathway opportunities, which could lead them on a path towards future STEM majors and career trajectories, lags significantly behind their White female and male counterparts. Meanwhile, the United States has an urgent need for a highly skilled and innovative STEM workforce and STEM leaders in order to stay competitive in a global marketplace. STEM education access in formal and informal learning spaces is limited for Black girls as a whole and often representative of compounding socio-political issues of racism, sexism, and generational poverty which keep Black girls pushed out of transformative learning experiences. Even if STEM education is available, these opportunities are often disconnected from the historical, cultural, gendered, and daily lives of Black girls. The cultivation of a STEM identity frequently lacks intersectional, and socio-cultural relevance for them. Several asset-based pedagogies and programs are being employed to cultivate Black girls STEM learning throughout their PK-12 education, in order for them to envision the possibilities of pursuing STEM majors and careers in the future, including entrepreneurial pursuits. Here we report on a community based informal after-school program - Black Girls STEAMing through Dance (BGSD). BGSD is a transdisciplinary STEAM program at the intersection of dance, design, and computational learning that builds upon the assets and interests of Black girls in order to make a STEM focused business a reality. As burgeoning social media creatives, the girls are given the opportunity to develop ideas that support the development of entrepreneurial ideas. Several girls have started on their entrepreneurial journey using their making and design skills to draw other girls into this world.

Methods 

Our research methods are grounded in an interpretive/ qualitative tradition, and we use a multimodal data collection and data analysis plan. Key constructs that we explore include STEAM literacies, STEAM identities, and Self-concept.  We have collected data from various BGSD activities covering artifacts from home (journals), dance (captured by video), design, and computational learning. Our data analysis plan consists of various methods of coding transcribed data as well as audio/visual analysis and narrative analysis techniques.

Contribution 

In summary, Black Girls STEAMing through Dance continues to bridge the racial and gender disparities in STEM by incorporating the arts into STEM activities to develop and maintain their STEM interests. By integrating dance throughout all aspects of BGSD, dance serves as an artform and entry point to leverage the cultural wealth of Black girls. STEAM activities are scaffolded from learning to dance (Art), to making and designing costumes (Art), and lastly coding to enhance costumes and to create VR dance environments (CS, Technology, Engineering, Mathematics). This technical approach to activity offers the girls the opportunity to see themselves as creators and not only consumers. BGSD is a model of transitioning from STEM to STEAM that leverages the cultural wealth and daily lives of Black girls.

U.S. cities vary widely in the diversity and robustness of their entrepreneurial ecosystems (EEs). The number and activity of new ventures are important metrics in determining the overall health of surrounding innovative and economic conditions (Cao & Shi, 2021; Guzman & Stern, 2020; Haltiwanger et al., 2013). Additionally, ecosystems benefit from having higher levels of equity in the distribution of entrepreneurial resources (Mazzucato, 2013); yet current conditions are far from equitable. There is a gap in the distribution of venture capital for under-represented entrepreneurs when compared to white men (Robb & Fairlie, 2007). This project analyzes longitudinal changes in equitable entrepreneurship after the introduction of Community Development Financial Institutions (CDFIs) into the surrounding ecosystem.

We build on efforts to uncover the institutional and economic roots of disparities across U.S. urban areas. We apply lessons from Robinson’s (2007) work that identified institutional barriers as a key obstacle for entrepreneurs in urban markets. Specifically, we highlight the role of financial lending institutions in supporting small minority-owned businesses. Foundational work in the field has shown that embeddedness in social and institutional networks can improve local business outcomes (Granovetter, 1985; Uzzi, 1996). Similar findings appear with respect to relationships between firms and financial institutions. And while considerable scholarship has focused on angel or venture capital access (Kerr et al., 2014; Samila & Sorenson, 2011), we shift focus to an essential, yet under-studied, form of financial access – communitybased financial institutions. Uzzi’s (1999) analysis shows a strong influence of a firm’s embedded relationships to banks on firm’s ability to secure capital. Our project also builds upon work aiming to measure the performance and health of entrepreneurial ecosystems by combining large public data sources into relational database management systems (RDBMS) (Johnson et al., 2022).

We find a discernible positive impact of CDFIs on the formation of minority- and woman-owned businesses, especially among high-tech firms, which in turn catalyzes the equitable distribution of entrepreneurial resources. Our empirical study sheds light on the puzzling results of prior research on the measurement and evaluation of CDFI performance (McCall & Hoyman, 2023). For managers, our research provides insight into capital access opportunities for minority and woman-owned establishments, especially young firms that are often severely constrained in their ability to secure the resources needed to survive and grow in competitive markets. For policymakers, our results emphasize the importance of government certification and support in providing financial assistance for CDFIs to strengthen diverse and equitable EEs and stimulate economic growth in LMI communities. Our findings imply that policy reforms that encourage the entry of new CDFIs or expand the geographic footprint of existing CDFIs are likely to boost the formation rate of minority- and woman-owned businesses.

Introduction
Generative Artificial Intelligence (AI) tools are increasingly becoming a cornerstone of societal operations, influencing sectors from healthcare to criminal justice. However, these systems often encapsulate and perpetuate societal biases, leading to unequal outcomes across different demographics. A few examples exemplify the societal biases permeating how Generative AI reflects systemic biases:

Example 1: A southeast student asked GenAI to make her photo more professional – it made her white Caucasian: https://www.msn.com/en-us/news/technology/an-mit-student-asked-ai-to-make-her-headshot-more-professional-it-gaveher-lighter-skin-and-blue-eyes/ar-AA1e4Hwr

Example 2: Gemini (a google genAI tool) tried to be inclusive – when asked to generate images of Nazi soldiers, the images depicted the soldiers as people of color and wearing hijab.

https://www.msn.com/en-us/news/technology/google-apologises-after-gemini-ai-generates-images-of-nazis-as-people-ofcolour/ar-BB1iHJM0

This session aims to elucidate these biases by presenting carefully developed prompts that force a deeper critical examination of Generative AI outputs among students, policymakers, entrepreneurs, and others. The initiative seeks to foster an understanding of bias in AI, thus systemic biases, stimulate dialogue on ethical AI use, and propose pathways towards more equitable technological solutions.

Methods
The session will be structured around a series of pre-developed prompts designed to reveal systemic biases in generative AI outputs. These prompts will cover a range of biases, including but not limited to gender, race, and socioeconomic status. The audience will be engaged through multiple formats (e.g. presentations, workshops, group discussions, panels, etc.) aimed at dissecting the AI-generated content for biases. We will guide participants through the evaluation of these outputs, providing tools and frameworks for critical analysis. By creating a multi-disciplinary dialogue, the session intends to bridge gaps between technical understanding and societal impacts of embedded Generative AI biases.

Expected Results
While specific results will emerge from the conduct of the session, the expected outcomes include an enhanced awareness among participants of the nuanced ways biases manifest in AI systems and the societal repercussions thereof. Participants will leave with a set of analytical tools for identifying biases within AI outputs and a deeper appreciation for ethical AI development practices. The session aims to culminate in the drafting of a collective document that outlines observed biases, discussions from the session, and a set of actionable recommendations for addressing biases in AI, which could be disseminated among stakeholders in the AI ecosystem.

Policy Implications
The insights and recommendations derived from the session are poised to influence policy decisions related to AI and technology governance. Highlighting the critical need for policies that enforce transparency, accountability, and fairness in AI systems, the session's outcomes could advocate for the introduction of bias assessments, diversity mandates in AI development teams, and ethical AI training programs. Such policies would not only aim to mitigate biases but also foster innovation through inclusive and equitable AI development practices. Moreover, the session's findings could encourage the adoption of ethical guidelines and standards across industries, promoting a shift towards responsible AI that benefits all segments of society. Through its multi-faceted approach, the session aspires to contribute to the development of a technologically advanced society that is rooted in fairness and equality.

In today’s rapidly changing economic landscape, digital marketplaces have become integral to our society. Platforms like Airbnb, eBay, Uber, TaskRabbit, and Amazon offer unparalleled flexibility in work schedules and lower barriers to entry, enabling countless individual sellers to provide a wide variety of products and services globally (Vallas & Schor, 2020). However, despite their positive impact on economic opportunities and inclusivity, these platforms still face persistent discrimination related to gender, race, and ethnicity (Doleac & Stein, 2013; Fisman & Luca, 2016). This issue is particularly evident on platforms like Uber and Airbnb, where transactions are personal, and individual sellers significantly influence the consumer experience (Edelman et al., 2016; Fisman & Luca, 2016; Ge et al., 2016).

To counteract such challenges, scholars have called for the careful crafting of governance rules that regulate user interactions, aiming for a more inclusive and equitable marketplace (Troncoso & Luo, 2022). An essential aspect of market design revolves around the information available to users during transactions (Edelman et al., 2017). Because revealing user information can lead to discrimination based on social identities, research suggests that platforms should consider removing identifiable information like names and pictures to mitigate discrimination (Edelman et al., 2017; Edelman & Luca, 2014). However, platform designs often prioritize enhancing information flow to build trust among strangers (Bolton et al., 2013; Dai et al., 2018; Fradkin et al., 2021), a crucial factor in facilitating online transactions that typically occur anonymously and across geographical distances (Molz, 2013). Therefore, paradoxically, the very mechanisms designed to facilitate trust and transparency may have unintended consequences, potentially placing certain groups of individuals in a more vulnerable situation.

To explore this issue further, I leverage a design change on Airbnb, the leading short-term rental platform. In May 2023, Airbnb introduced an innovative “Host Passport” feature alongside each shared space listing to alleviate concerns related to the safety and trustworthiness of staying with a stranger. This “Host Passport” includes biographical information about the host, as well as prompts similar to those found on dating apps, inquiring about their occupation and personality. Prior to this change, users could only view information about the listings in the search results. However, following the change, host information now prominently appears on the search result page. Leveraging this exogenous shock, I predict that the sudden increase in the visibility of hosts’ social identities amplifies the racial gap in performance, as it becomes easier for users to exclude listings offered by minority hosts. I also explore other outcomes in addition to performance that might have important implications for the platform organization, such as hosts’ switch to competing platform. Furthermore, building on previous research that has examined entrepreneurs’ narrative strategies for achieving superior performance in the Airbnb marketplace, I investigate how minority hosts can effectively manage the distinctiveness of their personal profiles and listings through narratives to mitigate these challenges.

This project may contribute to the literature in several ways. Prior research on digital platforms has not explicitly considered of the potentially unintended impact of trust-enhancing mechanisms on racial minorities hosts. Furthermore, extant research has primarily focused on an individual’s performance as the only outcome of platform design, but has not explicitly considered outcomes that would be important for the organizations, such as the host’s loss of commitment or even departure from the focal platform. Additionally, despite the attention on platform design to address discrimination in online marketplaces, there is a notable scarcity of research on how entrepreneurs can utilize their agency to surmount these challenges. This gap is surprising given the extensive body of work on how entrepreneurs can shape consumer perception and outcomes through identity crafting (Radu-Lefebvre et al., 2021; Mmbaga et al., 2020).

Introduction
Automated valuation models (AVMs) offer a streamlined and cost-effective approach to home valuation by eliminating the need for direct appraiser input. Traditionally, home appraisals involve in-person visits, but research suggests that racial bias among appraisers contributes to racial disparities in home values (Howell and Korver-Glenn, 2018). Today, many lenders are switching to AVMs to make the process more efficient and eliminate documented human bias. But there is growing concern that AVMs may inadvertently perpetuate racial disparities in home prices. Our previous research has documented racial disparities in AVMs at the neighborhoodlevel (Neal et al. 2020; Zhu, Neal, and Young, 2022) but this new research uses a raceimputation framework to investigate bias at the individual level.

Methods
We investigate the potential racial bias inherent in AVMs using an innovative imputation algorithm developed by McCartan et al. (2023), Bayesian Instrumental Regression for Disparity Estimation (BIRDiE). BIRDiE improves on traditional Bayesian Improved Surname Geocoding (BISG) (Elliott et al. 2009), which generates probabilities of an individual’s race and ethnicity based on last names, block groups, and self-reported race. BISG assumes that a person’s race does not directly affect the outcome of interest after controlling for name and location, which is a tenuous assumption for us, particularly given our previous findings that race impacts AVM error at the neighborhood level. BIRDiE allows us to relax the assumption that race does not directly affect the outcome, but instead assume that last name does not directly affect the outcome, by incorporating other relevant housing and neighborhood variables into the imputation model. We use individual-level data on property owners in the Memphis and Atlanta metropolitan areas to generate BISG probabilities and then update using the BIRDiE methodology.

After imputation, we draw on OLS regression analysis to examine systemic differences in valuation for homeowners by imputed race and ethnicity.

Results
Our analysis reveals a statistically significant discrepancy in valuation errors: homes of Black homeowners are subject to a 3.8 percentage point higher valuation error compared to those of White homeowners. Furthermore, we uncover an 11.3% undervaluation in AVM estimates for properties owned by Black individuals. This undervaluation poses a substantial risk of impeding housing wealth accumulation among Black homeowners. Our findings suggest that AVMs can manifest racial disparities, even when the algorithm remains agnostic to the neighborhood's majority race or the homebuyer's race. This study spotlights the need for a systematic framework to audit and address the disparate impacts of AVMs, providing a methodological foundation for evaluating and mitigating racial biases at the individual level.

Policy Implications
Policy solutions will vary depending on where the bias comes from, the data or the algorithms itself, and more research is needed to solidify this. But as a first step, improving the transparency and interpretability of AVM algorithms will enable more researchers and regulators to investigate biases and inform policy to limit bias. Our work also illustrates how imputation, when done carefully and with attention to the risks and tradeoffs, can be a policy tool in auditing for algorithmic bias.