Aaron Sojourner

@aaronsojourner.org

31478 Followers 7403 Following

Labor econ. Formerly White House CEA senior economist for labor, UMN prof, union organizer, bike courier, dishwasher. Minneapolis. Posts mine, not employer's (@upjohninstitute.bsky.social). Fellow @iza.org aaronsojourner.org Be kind. Work hard. Have fun.

Estimating Causal Effects from Data Generated by Stochastic Algorithms Susan Athey, Guido Imbens, and Zoe Ji July 2026 Abstract Recommendation systems and chatbots present content to users, typically using stochastic algorithms that select the content based on user characteristics or context. Examples of content include chat responses, videos, or items available for purchase. Scientists and application developers are often interested in whether characteristics of content increase outcomes such as user engagement. Estimates of such causal effects may guide content providers to generate content that emphasize desirable features. However, in settings with a large content library or where content is generated uniquely for a given user, it can be difficult to use observational data to learn the causal effect of content features, because the content a user sees is tailored to that user, and because content varies in many dimensions. The basic observation that motivates this paper is that because modern AI systems select and generate content stochastically, routine interactions between users and these systems continuously produce micro experiments, and logging two additional elements of each interaction turns those micro experiments into data that identify causal effects of content features. This paper proposes a new method for identifying and estimating the impact of content features using observational data, when the algorithm that determines user exposure incorporates some randomization, and when two additional data elements are logged for each user: (i) the identity of at least one item that could have been exposed to the user, but was not (the unexposed item); (ii) an estimate of the ratio of the probability that the unexposed item would have been shown to the probability that the exposed item was shown.
federal officials admitted CMS had shared data with ICE in January that went beyond what the court order allowed. One dataset of refugees in Minnesota included U.S. citizens
he Price of Corrupt Justice: Evidence from Progressive Era Chicago Presented by Roman Rivera, Assistant Professor of Public Policy at Georgetown University, Faculty Research Fellow at NBER, and Research Affiliate at IZA Thursday July 16, 1 - 2 p.m. ET To receive the Zoom link for this seminar, please click here to register. Abstract: "This paper investigates the spillover effects of judicial independence on police behavior. Despite the adversarial design of the US justice system, lower court judges and local police often share political ties and economic incentives to extract revenue. I study a pivotal Progressive Era reform, the 1905 Municipal Court Act (MCA) in Chicago, which replaced patronage-selected, fee-paid `police justices' with professional, salaried, and elected judges. Crucially, this reform broke the political and economic ties between the courts and police without directly targeting the police department. To study the MCA’s effect, I construct a novel dataset from dozens of archival police, court, and expenditure records. The reform observably increased judicial quality and compensation, but had no effect on police composition or compensation. Nevertheless, the reform caused an over 30% decline in arrests, driven almost entirely by low-quality, highly discretionary misdemeanor arrests. Precinct-level data reveal that this drop was concentrated in the redlight district. Empirical evidence and historical accounts indicate this was due in part to ending a police-justice collusion racket , and I estimate the MCA prevented 9,000 extortive arrests and over $1.2 million (in 2025 dollars) in extorted fees annually. Despite this drastic reduction in enforcement, neither property crime nor murder increased. These findings demonstrate that increasing accountability within a single institution can have positive spillover effects and that severing ideological and economic ties within the justice system can significantly improve local state capacity and publi…
Theoretical Economics: Jul, 2026, Volume 21, Issue 3 News media as suppliers of narratives (and information) https://doi.org/10.3982/TE6424 p. 928-972 Kfir Eliaz|Ran Spiegler We present a model in which news media shape beliefs by providing information (signals about an exogenous state) and narratives (models of what determines outcomes). To amplify consumers' engagement, the media maximize their anticipatory utility. We characterize the optimal monopolistic media strategy, highlighting the synergy between false narratives and biased information. Consumer heterogeneity gives rise to a novel menu‐design problem due to an “equilibrium data externality” among consumers. The optimal menu includes false narratives, and can generate polarized beliefs and choices without any information transmission. False narratives and polarization also feature in a competitive version of our model, albeit coupled with complete information. Nevertheless, competition can make some consumers objectively worse off.
Unknown donors pouring millions into what could be Minnesota’s most expensive Senate primary Untraceable spending, aka “dark money,” is fueling a flurry of ads supporting Angie Craig over Peggy Flanagan. By Allison Kite and Sydney Kashiwagi The Minnesota Star Tribune
When people don’t have access to health insurance, emergency rooms become more crowded and the cost of uncompensated care keeps growing, and inevitably the cost of carrying for a sicker society falls to the taxpayers and leads to higher premiums. “I wish people understood that this isn't just a problem for the people who have to forgo health insurance,” Roberson said. “It's not just their problem. It's the rest of ours because the rest of the healthcare system picks up the costs
Federal judges have spent a full year overwhelmingly rejecting ICE’s effort to lock up people the Trump administration is seeking to deport — finding a combination of illegality, unconstitutionality, incompetence and cruelty at a shocking scale. More than 15,000 times.
Since taking office on June 30, 2016, Philippine President Rodrigo Duterte has carried out a “war on drugs” that has led to the deaths of over 12,000 Filipinos to date, mostly urban poor. At least 2,555 of the killings have been attributed to the Philippine National Police. Duterte and other senior officials have instigated and incited the killings in a campaign that could amount to crimes against humanity. Human Rights Watch research has found that police are falsifying evidence to justify the unlawful killings. Despite growing calls for an investigation, Duterte has vowed to continue the campaign
Close to Home? Five Key Facts about Child Care Customer and Worker Commutes Upjohn Institute Working Paper No. 26-432 Gabrielle Pepin W.E. Upjohn Institute for Employment Research Jonathan Borowsky University of Minnesota Elizabeth E. Davis University of Minnesota Aaron Sojourner W.E. Upjohn Institute for Employment Research June 2026 ABSTRACT Parents often cite location and convenience as important factors in choosing a child care provider. We use data on individual mobility from GPS in cell phones to study commutes to each licensed early childhood care and education (ECE) provider in 14 states separately for center customers, center workers, and customers of home-based child care providers. We document five key facts about ECE customer and worker commutes. One, center customers and workers commute farther than home-based customers to ECE providers. Two, substantial shares of customers and workers commute 10+ miles to their ECE provider. Three, commute distances to ECE providers vary substantially across geographic regions. Four, commute distances, especially for center workers, increase with provider neighborhood socioeconomic status. Five, commute distances vary by the racial and ethnic composition of the provider location, even after controlling for measures of socioeconomic status. Results suggest variation in ECE market size by provider type, geographic location, and workforce characteristics, which has implications for policies intended to increase supply within the sector. Higher-income customers may lift child care prices and wages across a broader region than where they shop, impacting access in less privileged regions. Interventions targeting child care wages may have geographical spillovers outside of child care markets defined by consumer behavior. Key Words: child care, commutes, early childhood care and education workforce, market size This work was supported by a grant from the Washington Center for Equitable Growth.
The teachers getting $50,000 bonuses thanks to a massive Meta data center Story by Ben Eisen, Meghan Bobrowsky
Police Drones: An Empirical Analysis Presented by Miguel de Figueiredo, Professor of Law and Terry J. Tondro Research Scholar at the University of Connecticut Thursday July 2, 1 - 2 p.m. ET To receive the Zoom link for this seminar, please click here to register. Abstract: "While drones are well-known in the context of warfare and foreign surveillance, a relatively new phenomenon has emerged in recent years where domestic police departments are using them for a variety of uses. The phenomenon is growing rapidly. Currently, more than 1,700 police departments use drones, as compared to a few dozen in the mid-2010s. Despite their exploding use, however, we know almost nothing about their effects on police behavior and crime. This Article presents the first rigorous empirical analysis of police drones. We utilize previously untapped data from Montgomery County, Maryland, a jurisdiction that deployed “drones as first responders” (DFRs) more than 3,500 times in a two-year period. By combining that data with more than 1.7 million service calls and crime incident data, we present a comprehensive and detailed picture of how drones operate in a large county that is some 500 square miles and populated with more than 1,000,000 residents. Using a difference-in-differences approach, we are also able to examine the causal effects of police drones on call response time, call clearance time, and crime. Our results show that contrary to the hopes of police departments, municipal and county governments, and drone manufacturers, DFRs have had only modest effects on call response time, call clearance time, and crime. Taken together, our results have critical implications for the design of law and policy designed to regulate the use of police drones, and they contribute to difficult decisions involving the cost-benefit analysis of DFRs when weighing tradeoffs between public safety and citizen privacy." The full Virtual Crime Economics seminar schedule is always available here.
“Moving jobs to people” increases local employment rates by three- fourths more in distressed places than in other places, and these increases persist for decades. n Current “place policies” are dominated by state business tax incentives, but these incentives are often costly per job created, and booming states frequently offer higher incentives than distressed states. n Federal and state “place policies” should target distressed places; focus tax incentives on high-multiplier businesses; and invest in customized small-business advising, public infrastructure, land development, and job skills.
Republican Claims of Massive Fraud Imply Elon Musk Is a Moron It was just 17 months that Trump set Musk loose to weed out all the fraud
For H-2A applications submitted during the first six months of FY 2026, the estimated H-2A cost under observed wage offers and contract durations is approximately $147 million. If all the certified jobs in the first half of FY 2026 had instead been paid the 2025 AEWR of $18.15 per hour, the estimated wage cost would have been approximately $180 million. As a result, employers whose jobs were certified during the first half of FY 2026 saved approximately $33 million relative to what they would have paid if the 2025 AEWR was still in effect
The Economic Costs of Cutting SNAP: Every $1 in SNAP Cuts to Families with Children Costs Society $14 to $20 Anastasia Koutavas, Buyi Wang, Irwin Garfinkel, Elizabeth Ananat, Sophie Collyer, Megan Curran, Robert Paul Hartley, and Christopher Wimer Center on Poverty and Social Policy at Columbia University