Gregor Schubert

Research (by topic)

Effects of Artificial Intelligence and Technology Shocks on Firms and Labor Markets

Generative AI and Firm Values. Journal of Finance. (Forthcoming)

with Andrea Eisfeldt, Bledi Taska and Miao Ben Zhang 

PDF | SSRN |  Occupation GenAI Exposure Data  | Firm GenAI Exposure Data | NBER WP

Best Paper Award, TCU Finance Conference (Previous draft  was also circulated as "The Labor Impact of Generative AI on  Firm Values")

• Media coverage:  Bloomberg, How the World Works Podcast, Wall Street Journal, Financial Times, Barron's Op-Ed, 

• Selected Presentations:  TCU Finance Conference, Labor in the Age of Generative AI Conference at Chicago Booth, Wharton AI and the Future of Work Conference, 2024 Macro Finance Society Workshop, 2024 Duke/UNC Asset Pricing Conference, 2023 NBER Big Data and Securities Markets Fall Conference, Chicago Booth Empirical Finance Conference 2023, KDD Finance Day 2023

Abstract: How do recent advances in Generative AI affect firm value? We construct the first measure of firms' workforce exposures to Generative AI and show that an ``Artificial-Minus-Human'' (AMH) portfolio that is long high-exposure firms and short low-exposure firms earned daily returns of 0.44% in the two weeks following the release of ChatGPT. The labor-exposure effect is more pronounced for firms with greater data assets and is distinct from the effect of firms' product exposures to Generative AI. Highly-exposed workforces can be either substituted for or complemented by Generative AI technologies.  We measure whether the exposed tasks are core or supplemental to assess relative substitutability. Examining firms' labor demand and profitability following the release of ChatGPT supports a labor-technology substitution channel for the increases in exposed firms' values that we document. 

Generative AI and Finance. The Annual Review of Financial Economics. (Forthcoming)  with Andrea Eisfeldt.

SSRN  Published PDF  Twitter thread   Blog summary 

Abstract: Since ChatGPT’s release in 2022, demand for AI-related skills in finance has grown rapidly, as Generative AI drives significant technological changes in both financial research and the broader economy. We show that financial occupations are highly exposed to the productivity effects of Generative AI, review the literature on the impact of ChatGPT on firm value, and provide directions for future research investigating the impact of this major technology shock. Generative AI also holds great potential as a tool for finance researchers and practitioners: we review and describe innovations in research methods linked to improvements in AI tools, along with their applications. We offer a practical introduction to available tools and advice for researchers in academia and industry interested in using these tools.

Organizational Technology Ladders: Remote Work and Generative AI Adoption

Presentations: NBER Organizational Economics Spring 2025 Meeting, CREDA Real Estate Research Symposium 2024 , Wharton Business and Generative AI Conference 2025 (San Francisco); USC AI and Economics Conference; UEA North America Meeting 2025 in Montreal; Stanford Remote Work Conference 2025

Scheduled presentations: SF Fed Micro Macro Labor Conference 2025; German Economists Abroad Conference 2025; CU Boulder Finance Seminar; GSU Finance Seminar

SSRN PDF  

Abstract: What drives inequality in technology adoption among firms? In this study, I provide novel empirical evidence linking two of the biggest technology shocks to firms in the last decade: remote work and generative AI. I develop an IV approach to identify the causal effect of remote work and apply it to detailed job posting data to estimate large positive effects of remote work on generative AI skill demand. Conversely, I provide evidence from a synthetic difference-in-differences approach that firms that were more exposed to generative AI technology reduced their demand for remote workers after ChatGPT was released. I rationalize these results using a task-based model of firm investments in new technology. Moreover, I provide evidence for the mechanism through which this “organizational technology ladder” operates: When firms go remote, they invest in technology skills, which, in turn, enable a more rapid generative AI adoption. Firms that are less able to accommodate remote work because they have lower managerial or communication capabilities, or workers with lower individual decision-making skills, are more likely to adopt generative AI after they go remote, and are more likely to reduce their remote hiring after they are exposed to generative AI tools.

Career Values for Labor Markets: Evidence from  Robot Adoption  (under review) with Maria Petrova, Pinar Yildirim & Bledi Taska

NEW DRAFT  SSRN  Old version: NBER WP 

• Media coverage: Chicago Booth Review

• Presentations: NBER Economics of Artificial Intelligence Conference 2024 (Toronto), Digital, Data, Design institute at Harvard University

Abstract: In this study, I propose that firms move along an ``organizational technology ladder'': adopting one technology transforms hiring and work processes, and builds skills and capital which then change the cost of adopting the next technology. I focus on the impact of firms' adoption of remote work technology on the proliferation of generative AI technology: using detailed job posting data and an  IV approach based on differences in labor market pressure to allow remote work, I show that a 10 pp rise in remote work induced by the instrument in 2021-22 causes an increase in job postings mentioning generative AI skill by 2024 of 0.4 pp across firms and 1.1 pp across occupations within firms. Generative AI adoption also changes firms' use of remote work: a synthetic difference-in-differences estimation around the release of ChatGPT shows that firms with greater exposure to generative AI reduced their remote hiring by 13% relative to a matched control group, with a more negative effect for tech firms than for non-tech firms. I explore the mechanism for these effects and show that  remote work adoption shifts hiring towards technology and management skills that then enable more rapid generative AI adoption. Generative AI adoption skews remaining human tasks towards requiring more decision-making in the tech sector, which can explain the decline in remote work in tech, if decision-intensive tasks are less productive while working remotely. Firms that have low remote productivity, as evidenced by having a return-to-office mandate, are more likely to adopt generative AI after hiring remotely. I formalize this technology ladder mechanism using a task-based model of sequential firm investments in new technologies that change the importance of in-person coordination.

The Household Benefits of Generative AI    (w. Michael Blank and Miao Ben Zhang)

(Draft coming soon)

• Presentations: OpenAI

• Scheduled Presentations: AEA Meeting 2026

Abstract: Who adopts generative AI, and what do households use generative AI for in practice? This paper provides the first comprehensive evidence on households' usage of generative AI, documents inequality in household benefits and changes in household behavior after adopting generative AI tools. We develop new measures of website and household exposure to generative AI benefits and show that they can predict adoption patterns.

Robotization and the Political Response of Politicians. with Maria Petrova, Pinar Yildirim & Bledi Taska (2025). In A. Agrawal, J. Gans, A. Goldfarb, & C. Tucker (Eds.), The economics of artificial intelligence: Political economy. NBER Conference Volume.

  NBER Book  PDF 

Abstract: Politicians respond to increased robotization in the labor market: we find that the Republican presidential candidate in 2016 targeted campaign visits to areas with high robot exposure and that representatives in high-manufacturing areas shifted their policy positions rightward in response to automation.

Labor Mobility

Employer Concentration and Wages: Evidence from Large Firms' Hiring Shocks (w. Matthew Dey,  Anna Stansbury & Bledi Taska) 

Reject & Resubmit, American Economic Journal: Economic Policy

This is  a revised and expanded version of some of the results from the original working paper "Employer Concentration and Outside Options".

 [NEW DRAFT] [Occupational Transition Data] [Non-Technical Write-up] 

We estimate the effect of increases in employer concentration on wages, using a new instrument for employer concentration based on changes in large firms’ national hiring

patterns. We measure employer concentration over mobility-adjusted labor markets: clusters of local occupations identified through asymmetric mobility patterns (using

new, highly granular data on occupational mobility from 16 million resumes). We find that increased employer concentration causally reduces wages: moving from the

median to the 95th percentile of employer concentration as experienced by workers lowers wages by 4 log points. Overall, we estimate that more than one in six U.S.

workers face wage suppression of 2.5% or more as a result of above-median employer concentration. The effects of employer concentration are particularly pronounced in

healthcare occupations.

Partisan Migration and Political Segregation  (w. Ben Pirie & Stuart Gabriel) 

Preliminary draft: Slides

• Presentations: UEA North America Meeting 2025 in Montreal, Wharton Real Estate Brown Bag

Abstract: We show that partisan political identity has become an increasingly important determinant of where Americans choose to live. Using individual‑level voter‑registration records for 2012‑2022, we construct county‑to‑county migration flows by party and quantify the degree to which migration is ``politically-aligned''---moving to a county whose presidential vote share matches one’s own party. We use  this data to estimate party-specific location preferences in a spatial equilibrium model and find a sharp increase in partisan preferences:  a rise in a destination's Democratic vote share raises Democratic in‑migration by and reduces Republican in‑migration. Partisan preferences by movers increase  relative to 2012‑2015.  This change in partisanship can explain a substantial part of the growth in county‑level partisan segregation, as the share of counties  that were similarly attractive to both parties declined substantially during the decade.

Real Estate Finance and Housing Market Dynamics

When Money Moves In: The Consequences of Housing Wealth    (w. Darren Aiello & Jason Kotter)

   [SSRN]

• Overall Best Paper Prize, AREUEA National Meeting 2023

• Presentations: USC Lusk Real Estate Seminar, UEA North America Meeting 2024, Boston College Finance Seminar, Oxford Said Business School, UC Irvine

• Media Coverage:  Wall Street Journal

We construct a novel dataset tracking households across property purchases covering 25 years of moves within the U.S. We find that information frictions in residential real estate markets cause movers with larger exogenous housing wealth to overpay for their next house, relative to both time varying local prices as well as time invariant characteristics of the property itself. These housing wealth driven overpayments are associated with larger positive price impacts to the immediately surrounding neighborhood and are larger for local movers relative to non-local movers. The aggregate effect of housing wealth inflows is to increase county-level house prices.

House Price Contagion and U.S. City Migration Networks

  SSRN Non-technical write-up    Presentation video    Download migration network data

• Media coverage: Mortgage Banker Magazine

• WFA Ph.D. Candidate Award for Outstanding Research, 2021

• Real Estate Cycles Manuscript Prize of the American Real Estate Society, 2021

• Presentations: NBER SI Real Estate 2021, WFA 2021, EUEA 2021, EALE 2021, ARES 2021, UEA 2020, UBC Urban Economics, Univ. of Toronto Urban Economics, Stanford GSB Finance Seminar, ITAM Finance Workshop

How do shocks propagate between city-level housing markets? I propose an explanation based on migration spillovers between U.S. cities. I derive an empirical measure of a city’s migration exposure to other cities, and use the network structure of migration to estimate that a 10 pp house price growth shock in a city’s network causes a 4.3 pp increase in house price growth in the focal city. I show that differences in migration networks can explain differences in local housing cycles, and show that accounting for spillover effects improves the ability to explain housing market effects in multiple applications from the finance literature, including 2002-2006 housing boom dynamics, and the effects of the Covid pandemic.

The Financial Consequences of Wanting to Own a Home  (w.  Caitlin Gorback)

 [Working Paper]  [SSRN]

• Presentations: NBER SI 2023, Real Estate,  Urban Economics Association North America Meeting 2024

We study the causal effects of homeownership affinities on tenure choice, household sensitivity to credit shocks, and retirement portfolios. Exploiting exogenous variation in affinities across U.S. immigrants’ countries of origin, we find that a 10pp higher affinity causes a 1.5pp higher homeownership rate in the U.S. Using exogenous credit-supply shocks, we show that high-affinity households are more responsive to credit availability, and less likely to default on mortgage payments. By retirement, high-affinity households realize higher homeownership, greater total wealth, and larger real estate shares in their portfolios. These effects are largely driven by appreciation.

Research in Progress

Age Structure and Housing Affordability

 (w. Erica Moszkowski, Shiyan Wei)

Presentations: Urban Economics Association North America Meeting 2024

Career Dynamics of Users and Builders of Generative AI

(funded by a grant from Anthropic's Economic Futures Program)

Digital Technology and Startup Teams

(w. John Barrios, Yael Hochberg)

AI Can Give You a Good Price for That: Using LLMs to Understand Heterogenity in Asset Valuations

 (w. D. Aiello and J. Kotter)