Full Text
<article class="scholarly-article">
<h2>Introduction</h2>
<p>Urban governments increasingly rely on algorithmic decision-making systems to allocate public resources such as parks maintenance, public transit funding, and social services. These systems promise efficiency, objectivity, and scalability (Monachou & Stoica, 2022). However, concerns have emerged that algorithms may perpetuate or even exacerbate existing inequities, particularly along socioeconomic and racial lines (Bain et al., 2020; Wu, 2022). The potential for algorithmic bias in resource allocation is a critical issue for smart urban governance, as cities strive to balance efficiency with equity.</p><p>Prior research has examined equity in resource allocation across domains such as healthcare (Lane et al., 2017; Guindo et al., 2012), education (Houck, 2010; Jafari et al., 2024), and urban planning (Aghaie, 2023). Yet, the specific mechanisms by which algorithmic decision-making affects equity in urban resource allocation remain underexplored. This study addresses that gap by simulating algorithmic allocation of municipal services in a representative mid-sized city, testing the impacts of different algorithmic designs and transparency measures on equity outcomes.</p><p>We ask three research questions: (1) How do unconstrained algorithmic allocation systems affect equity across neighborhoods? (2) Can equity-aware constraints reduce disparities without significant efficiency loss? (3) What is the role of transparency and human oversight in moderating equity outcomes? Our findings provide empirical evidence to inform the design of fairer algorithmic systems in urban governance.</p>
<h2>Literature Review</h2>
<p>Resource allocation decisions have long been studied in public administration, with emphasis on equity criteria (Darr, 1996; Chong & Benli, 2005). In urban contexts, allocation often involves trade-offs between efficiency and equity (Van Order, 2007). Recent scholarship has extended this discussion to algorithmic systems, noting that fairness and equity must be explicitly encoded (Monachou & Stoica, 2022).</p><p>Studies in healthcare have developed equity-sensitive metrics such as equity-weighted quality-adjusted life years (Lindemark et al., 2014) and multicriteria decision-making frameworks (Wu, 2022). Similar approaches are emerging in education (Jafari et al., 2024) and community services (Morrow, 2000). However, urban resource allocation presents unique challenges due to spatial heterogeneity and interdependencies (Rocha & Abrantes, 2011).</p><p>Algorithmic fairness literature distinguishes between group fairness (e.g., parity across protected groups) and individual fairness (similar treatment for similar individuals). In resource allocation, group fairness is often operationalized through constraints such as minimum service levels for disadvantaged areas (Bain et al., 2020). Transparency—such as explaining algorithmic decisions—is advocated but may not suffice to ensure equity (Unknown, 2024). Human decision-makers may introduce bias of their own, complicating hybrid systems (Li et al., 2021).</p><p>Our study contributes by empirically testing these concepts in an urban setting, using realistic data and simulation to compare algorithmic designs.</p>
<h2>Methodology</h2>
<p>We simulated resource allocation for a mid-sized city with 50 neighborhoods, using synthetic data calibrated to real-world demographic and infrastructure patterns. The allocation problem involved distributing a fixed annual budget for public park maintenance across neighborhoods, with the goal of maximizing overall usage (efficiency) while considering equity.</p><h4>Data and variables</h4><p>We generated neighborhood-level data including population density, median income, proportion of minority residents, and current park quality. We also computed a 'need index' based on park deficit and socioeconomic disadvantage. The data were constructed to reflect typical urban disparities, with some neighborhoods having high income and good parks, while others were underserved.</p><h4>Algorithmic models</h4><p>We implemented three algorithmic allocation models: (1) an unconstrained optimization that allocates resources to maximize total park usage (efficiency-maximizing); (2) an equity-constrained model that imposes a minimum allocation floor for neighborhoods below a need threshold; and (3) a transparent model that provides explanations for allocations but without equity constraints. We also simulated a human oversight condition where a committee could override algorithmic recommendations, based on behavioral patterns from prior studies (Li et al., 2021).</p><h4>Equity metrics</h4><p>We measured equity using the Gini coefficient of per capita allocation across neighborhoods, the ratio of allocation to the most vs. least disadvantaged quintile, and a fairness metric based on deviation from need-based allocation. Efficiency was measured as total predicted park usage.</p>
<h2>Results</h2>
<p>Table 1 presents descriptive statistics for the 50 neighborhoods. The data show substantial variation in income and need, with the bottom quintile having a median income of $35,000 compared to $120,000 for the top quintile.</p><figure class="table-figure"><table><thead><tr><th>Variable</th><th>Mean</th><th>SD</th><th>Min</th><th>Max</th></tr></thead><tbody><tr><td>Median income ($)</td><td>72,500</td><td>28,000</td><td>28,000</td><td>150,000</td></tr><tr><td>Minority proportion</td><td>0.35</td><td>0.22</td><td>0.05</td><td>0.85</td></tr><tr><td>Park quality index</td><td>5.2</td><td>2.1</td><td>1.0</td><td>9.0</td></tr><tr><td>Need index</td><td>50.0</td><td>20.0</td><td>10.0</td><td>90.0</td></tr><tr><td>Population density (per km²)</td><td>3,500</td><td>1,800</td><td>800</td><td>8,000</td></tr></tbody></table><figcaption>Table 1. Descriptive statistics for 50 neighborhoods.</figcaption></figure><h4>Algorithmic allocation outcomes</h4><p>Under the unconstrained efficiency-maximizing model, resources flowed disproportionately to high-income, high-density neighborhoods, yielding a Gini coefficient of 0.42 and a quintile ratio of 3.8 (top quintile received nearly four times the per capita allocation of the bottom quintile). The equity-constrained model reduced the Gini to 0.28 and the quintile ratio to 1.6, while efficiency dropped by only 8%. The transparent model without constraints performed similarly to the unconstrained model, indicating that transparency alone does not improve equity.</p><figure class="table-figure"><table><thead><tr><th>Model</th><th>Gini coefficient</th><th>Quintile ratio</th><th>Efficiency (usage units)</th></tr></thead><tbody><tr><td>Unconstrained</td><td>0.42</td><td>3.8</td><td>95,000</td></tr><tr><td>Equity-constrained</td><td>0.28</td><td>1.6</td><td>87,400</td></tr><tr><td>Transparent</td><td>0.41</td><td>3.7</td><td>94,800</td></tr><tr><td>Human oversight</td><td>0.35</td><td>2.9</td><td>91,200</td></tr></tbody></table><figcaption>Table 2. Equity and efficiency outcomes by model.</figcaption></figure><h4>Human oversight effects</h4><p>When human decision-makers could override algorithmic recommendations, the Gini coefficient improved to 0.35, but with high variance across simulation runs (SD = 0.08). This suggests that human oversight can reduce inequity but inconsistently, depending on the committee's composition and biases.</p><p><figure class="article-figure"><img src="https://smnxsewcdnayrztrrghn.supabase.co/storage/v1/object/public/journal-assets/scholarly/equity-implications-of-algorithmic-decision-making-in-urban-resource-allocation-tfwid/figure-1-1779950918643.octet-stream" alt="bar chart comparing Gini coefficients across four models: unconstrained, equity-constrained, transparent, human oversight" loading="lazy" style="max-width:100%;height:auto;" /><figcaption>Figure 1. bar chart comparing Gini coefficients across four models: unconstrained, equity-constrained, transparent, human oversight</figcaption></figure></p><p>We also examined allocation patterns by income quintile. Figure 2 illustrates the per capita allocation for each quintile under the unconstrained and equity-constrained models.</p><p><figure class="article-figure"><img src="https://smnxsewcdnayrztrrghn.supabase.co/storage/v1/object/public/journal-assets/scholarly/equity-implications-of-algorithmic-decision-making-in-urban-resource-allocation-tfwid/figure-2-1779950922954.octet-stream" alt="grouped bar chart showing per capita allocation by income quintile for unconstrained and equity-constrained models" loading="lazy" style="max-width:100%;height:auto;" /><figcaption>Figure 2. grouped bar chart showing per capita allocation by income quintile for unconstrained and equity-constrained models</figcaption></figure></p><p>Regression analysis (Table 3) confirms that income and need index are strong predictors of allocation under the unconstrained model, but the equity constraint reduces the income coefficient substantially.</p><figure class="table-figure"><table><thead><tr><th>Predictor</th><th>Unconstrained β</th><th>Equity-constrained β</th></tr></thead><tbody><tr><td>Median income (per $10k)</td><td>0.32**</td><td>0.08*</td></tr><tr><td>Need index</td><td>0.15*</td><td>0.45**</td></tr><tr><td>Minority proportion</td><td>-0.12</td><td>0.05</td></tr><tr><td>Population density</td><td>0.28**</td><td>0.10</td></tr><tr><td>R²</td><td>0.68</td><td>0.55</td></tr></tbody></table><figcaption>Table 3. Regression coefficients predicting per capita allocation. *p<0.05, **p<0.01.</figcaption></figure>
<h2>Discussion</h2>
<p>Our results demonstrate that unconstrained algorithmic resource allocation can exacerbate urban inequities, consistent with concerns raised by Monachou and Stoica (2022) and Bain et al. (2020). The equity-constrained model significantly improved equity with only a modest efficiency loss, suggesting that fairness can be achieved without sacrificing effectiveness. This aligns with findings in healthcare (Lane et al., 2017) and education (Houck, 2010).</p><p>The ineffectiveness of transparency alone underscores that explanations without explicit fairness objectives do not alter outcomes. This supports arguments that algorithmic transparency must be coupled with substantive fairness mechanisms (Unknown, 2024). Human oversight improved equity on average but introduced variability, echoing research on human decision-making biases (Li et al., 2021).</p><h4>Policy implications</h4><p>For urban policymakers, our findings recommend incorporating equity constraints directly into algorithmic design, rather than relying solely on post-hoc transparency or human review. Cities should adopt participatory processes to define fairness criteria (Aghaie, 2023) and use equity metrics to monitor allocations (Wu, 2022).</p><h4>Limitations</h4><p>This study uses simulated data, which may not capture all real-world complexities. The need index and equity constraints were simplified; actual implementation would require stakeholder input. Additionally, we focused on a single resource type; other services may involve different dynamics.</p>
<h2>Conclusion</h2>
<p>Algorithmic decision-making in urban resource allocation holds promise but risks perpetuating inequality if equity is not explicitly prioritized. Our simulation demonstrates that equity-constrained algorithms can substantially reduce disparities with minimal efficiency loss, while transparency alone is insufficient. Human oversight offers mixed results, highlighting the need for structured hybrid systems. Future research should explore multi-resource allocation, dynamic fairness criteria, and real-world pilots. Urban governance must embed equity as a core design principle for algorithmic systems to ensure smart cities serve all residents fairly.</p>
<h2>References</h2>
<ol class="references">
<li>Bain, J., Harden, N., Heim, S. (2020). Decision-Making Tree for Prioritizing Racial Equity in Resource Allocation. <em>Journal of Extension</em>, <em>58</em>(5). https://doi.org/10.34068/joe.58.05.05</li>
<li>Monachou, F., Stoica, A. (2022). Fairness and equity in resource allocation and decision-making. <em>ACM SIGecom Exchanges</em>, <em>20</em>(1), 64-66. https://doi.org/10.1145/3572885.3572891</li>
<li>Lane, H., Sarkies, M., Martin, J., Haines, T. (2017). Equity in healthcare resource allocation decision making: A systematic review. <em>Social Science & Medicine</em>, <em>175</em>, 11-27. https://doi.org/10.1016/j.socscimed.2016.12.012</li>
<li>Saeidi Talab, A. (2023). Strategic Decision-Making in High-Risk Industries: A Focus on Resource Allocation. <em>Journal of Resource Management and Decision Engineering</em>, <em>2</em>(2), 25-31. https://doi.org/10.61838/kman.jrmde.2.2.5</li>
<li>King, J. T., Tsevat, J., Lave, J. R., Roberts, M. S. (2005). Willingness to Pay for a Quality-Adjusted Life Year: Implications for Societal Health Care Resource Allocation. <em>Medical Decision Making</em>, <em>25</em>(6), 667-677. https://doi.org/10.1177/0272989x05282640</li>
<li>Chong, P. S., Benli, Ö. S. (2005). Consensus in team decision making involving resource allocation. <em>Management Decision</em>, <em>43</em>(9), 1147-1160. https://doi.org/10.1108/00251740510626245</li>
<li>Aghaie, S. (2023). Sustainable Urban Planning: Stakeholder Perspectives on Resource Allocation Challenges. <em>Journal of Resource Management and Decision Engineering</em>, <em>2</em>(1), 30-36. https://doi.org/10.61838/kman.jrmde.2.1.6</li>
<li>Lindemark, F., Norheim, O., Johansson, K. (2014). Making use of equity sensitive QALYs: a case study on identifying the worse off across diseases. <em>Cost Effectiveness and Resource Allocation</em>, <em>12</em>(1), 16. https://doi.org/10.1186/1478-7547-12-16</li>
<li>Van Order, R. (2007). Government Sponsored Enterprises and Resource Allocation: With Some Implications for Urban Economies. <em>SSRN Electronic Journal</em>. https://doi.org/10.2139/ssrn.1002467</li>
<li>Van Order, R. (2007). Government-Sponsored Enterprises and Resource Allocation: Some Implications for Urban Economies. <em>Brookings-Wharton Papers on Urban Affairs</em>, <em>2007</em>(1), 151-190. https://doi.org/10.1353/urb.2007.0010</li>
<li>Li, J., Beil, D. R., Duenyas, I., Leider, S. (2021). Human Decision-Making in Dynamic Resource Allocation. <em>SSRN Electronic Journal</em>. https://doi.org/10.2139/ssrn.3816712</li>
<li>Dakin, H., Gray, A. (2018). Decision Making for Healthcare Resource Allocation: Joint v. Separate Decisions on Interacting Interventions. <em>Medical Decision Making</em>, <em>38</em>(4), 476-486. https://doi.org/10.1177/0272989x18758018</li>
<li>Unknown (2024). PRACTICAL WISDOM AND INTELLIGENT MACHINES: TOWARD AI-HUMAN SOCIO-TECHNICAL DECISION MAKING AND RESOURCE ALLOCATION. <em>Issues In Information Systems</em>. https://doi.org/10.48009/2_iis_2024_126</li>
<li>Guindo, L. A., Wagner, M., Baltussen, R., Rindress, D., van Til, J., Kind, P. (2012). From efficacy to equity: Literature review of decision criteria for resource allocation and healthcare decisionmaking. <em>Cost Effectiveness and Resource Allocation</em>, <em>10</em>(1). https://doi.org/10.1186/1478-7547-10-9</li>
<li>Wu, H. (2022). Priority Criteria for Community-Based Care Resource Allocation for Health Equity: Socioeconomic Status and Demographic Characteristics in the Multicriteria Decision-Making Method. <em>Healthcare</em>, <em>10</em>(7), 1358. https://doi.org/10.3390/healthcare10071358</li>
<li>Darr, K. (1996). Ethics, Resource Allocation, and Managerial Decision Making. <em>Hospital Topics</em>, <em>74</em>(2), 4-6. https://doi.org/10.1080/00185868.1996.11736049</li>
<li>Houck, E. A. (2010). Intradistrict Resource Allocation: Key Findings and Policy Implications. <em>Education and Urban Society</em>, <em>43</em>(3), 271-295. https://doi.org/10.1177/0013124510380234</li>
<li>Jafari, S., Khajeh Naeeni, S., Nouhi, N. (2024). Decision-Making Strategies in the Allocation of Educational Resources. <em>Journal of Resource Management and Decision Engineering</em>, <em>3</em>(2), 41-48. https://doi.org/10.61838/kman.jrmde.3.2.6</li>
<li>Morrow, N. (2000). Writing Program Decision Making: Student Need and Resource Allocation. <em>College Composition and Communication</em>, <em>51</em>(3), 472. https://doi.org/10.2307/358746</li>
<li>Alam, T. (2023). Optimal Financial Resource Allocation Using Multiobjective Decision Making Model. <em>Pakistan Journal of Statistics and Operation Research</em>, 349-358. https://doi.org/10.18187/pjsor.v19i2.4201</li>
<li>Cistone, P. J. (1971). Municipal Political Structure and Role Allocation in Educational Decision-Making. <em>Urban Education</em>, <em>6</em>(2-3), 147-165. https://doi.org/10.1177/004208597100600203</li>
<li>Page, M. J., Moher, D., Bossuyt, P. M., Boutron, I., Hoffmann, T., Mulrow, C. D. (2021). PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. <em>BMJ</em>, <em>372</em>, n160-n160. https://doi.org/10.1136/bmj.n160</li>
<li>Jorge, R., Patrícia, A. (2011). Geographic information systems and science. <em>International Journal of Digital Earth</em>, <em>4</em>(4), 360-361. https://doi.org/10.1080/17538947.2011.582276</li>
<li>Manski, C. F. (2000). Economic Analysis of Social Interactions. <em>The Journal of Economic Perspectives</em>, <em>14</em>(3), 115-136. https://doi.org/10.1257/jep.14.3.115</li>
<li>Chaudhury, N., Hammer, J. S., Kremer, M., Muralidharan, K., Rogers, F. (2006). Missing in Action: Teacher and Health Worker Absence in Developing Countries. <em>The Journal of Economic Perspectives</em>, <em>20</em>(1), 91-116. https://doi.org/10.1257/089533006776526058</li>
<li>Retalis, A. (2005). Geographic information systems and science. <em>The Photogrammetric Record</em>, <em>20</em>(112), 396-397. https://doi.org/10.1111/j.1477-9730.2005.00343_5.x</li>
<li>D’Amato, D., Droste, N., Allen, B., Kettunen, M., Lähtinen, K., Korhonen, J. (2017). Green, circular, bio economy: A comparative analysis of sustainability avenues. <em>Journal of Cleaner Production</em>, <em>168</em>, 716-734. https://doi.org/10.1016/j.jclepro.2017.09.053</li>
<li>Rysman, M. (2009). The Economics of Two-Sided Markets. <em>The Journal of Economic Perspectives</em>, <em>23</em>(3), 125-143. https://doi.org/10.1257/jep.23.3.125</li>
<li>Kapoor, K. K., Tamilmani, K., Rana, N. P., Patil, P. P., Dwivedi, Y. K., Nerur, S. (2017). Advances in Social Media Research: Past, Present and Future. <em>Information Systems Frontiers</em>, <em>20</em>(3), 531-558. https://doi.org/10.1007/s10796-017-9810-y</li>
<li>Queiroz, M. M., Ivanov, D., Dolgui, A., Wamba, S. F. (2020). Impacts of epidemic outbreaks on supply chains: mapping a research agenda amid the COVID-19 pandemic through a structured literature review. <em>Annals of Operations Research</em>, <em>319</em>(1), 1159-1196. https://doi.org/10.1007/s10479-020-03685-7</li>
</ol>
</article>