Peter John Lambert

Peter John Lambert

I am an economist studying organizations, industries, jobs, and economic growth.

My research leverages large novel datasets (e.g. text, transactions, networks, images) and frontier AI tools / algorithms to study behaviour within and across organizations. Much of my current work measures how AI and automation are changing hiring, work and productivity.

I hold a PhD in Economics from the London School of Economics. I am a Research Fellow in the Department of Economics at the University of Warwick and a Visiting Research Fellow in the Department of Economics at LSE, with affiliations at CEP, CAGE, POID, and RFBerlin. In 2023 I co-founded the Applied Economics using AI (AEAI) Lab.

Please get in touch if you wish to discuss ideas, I’d love to hear from you! My email is p.j.lambert@lse.ac.uk.

Working papers

  1. 2026SSRN

    The Broken Ladder: AI, Remote Work, and Early-Career Hiring

    with Yannick Schindler

    Coverage Financial Times, The AI Shift Newsletter (FT), Business Insider, Fortune, Bloomberg News, Planet Money (The Indicator), Axios, Bloomberg (video), Context Window Podcast (David Deming), The Wall Street Journal

    Abstract

    Is generative AI replacing junior workers? A growing literature answers yes, citing large declines in early-career hiring concentrated in GenAI-exposed occupations. We argue that this verdict is premature because GenAI exposure is strongly correlated with another post-pandemic shock, working from home (WFH). Using two data sources spanning 243 million new hires and 407 million online job postings, collected across the US, UK, Canada, and Australia during 2017-2025, we estimate difference-in-difference designs at the occupation, region, and firm level. When estimated separately, a two-standard-deviation increase in GenAI and WFH exposure each predicts, by 2025, a fall of around 5pp in the junior-share of new hires and around 3pp in the share of job ads requiring limited experience. Estimated jointly, the WFH effect remains, while the GenAI coefficient attenuates sharply and is often statistically indistinguishable from zero. Alternative exposure measures, residualization designs, flexible non-parametric co-treatment controls, and replacing exposure-measures with actual WFH adoption as the treatment all support our finding that WFH is a robust predictor of the decline in early-career hiring.

  2. 2026NBER WP 35552

    The Equilibrium Impact of Credit Frictions: Evidence from Default Risk Using Firm-Level Data

    with Tim Besley, Isabelle Roland, John Van Reenen

    Abstract

    This paper examines the impact of credit frictions arising from firm-level default risk on aggregate economic performance. We build a micro-to-macro model with heterogeneous firms and sector-specific production functions, showing that perceived default risk is a sufficient statistic for credit frictions. Using UK administrative data (2004–2019) matched to S&P risk measures, counterfactual estimates reveal that relaxing frictions raises output by 25% and wages by 23%. Ignoring equilibrium wage adjustments overstates output gains, while fixed-capital misallocation approaches understate them. Most gains reflect aggregate capital accumulation. Credit frictions remain above pre-crisis levels, reshape firm size dynamics, increase misallocation across firms, and dampen productivity growth over time.

  3. 2026CEP DP 2216

    Measuring the Prevalence and Direction of Innovation: A Pilot Approach Using LLMs

    with Prashant Garg, Ralf Martin and Denis Medvedev

    Abstract

    This paper pilots a method for measuring firms’ product and process innovation at scale. Our approach uses LLMs to complete a detailed innovation survey for a sample of 600,000 companies operating across 40 countries. The preliminary results contain 6 million individual instances of innovative activity along with rich textual descriptions. These responses carry signal, for example a one-standard-deviation increase in a firm’s product innovation index is associated with a 5% higher measured TFP residual. The data reveal that innovation is widespread: two-fifths of companies engage in product innovation and one-quarter in process innovation, far exceeding the share of patenting firms. Firms are more likely to have labor-augmenting than labor-saving introductions, while about 14% have green product innovation and 12% have green process innovation. Greenness and net labor augmentation are positively associated across countries. Beyond our specific application, this paper illustrates the potential for using LLMs to generate survey-level insights at population scale.

  4. 2024CEPR DP 19708

    AI-Generated Production Networks: Measurement and Applications to Global Trade

    with Thiemo Fetzer, Bennet Feld & Prashant Garg

    Abstract

    This paper leverages generative AI to build a network structure over 5,000 product nodes, where directed edges represent input-output relationships in production. We lay out a two-step ‘build-prune’ approach using an ensemble of prompt-tuned generative AI classifications. The ‘build’ step provides an initial distribution of edge-predictions, the ‘prune’ step then re-evaluates all edges. With our AI-generated Production Network (AIPNET) in tow, we document a host of shifts in the network position of products and countries during the 21st century. Finally, we study production network spillovers using the natural experiment presented by the 2017 blockade of Qatar. We find strong evidence of such spill-overs, suggestive of on-shoring of critical production. This descriptive and causal evidence demonstrates some of the many research possibilities opened up by our granular measurement of product linkages, including studies of on-shoring, industrial policy, and other recent shifts in global trade.

  5. 2024

    Bad Bank, Bad Luck? Evidence From 1 Million Firm-Bank Relationships (PDF)

    with Yannick Schindler

    Abstract

    This paper studies the effects of bank failure on firm performance. We collect 36 million loan records to build a novel dataset on the credit relationships of 1.8 million US firms, predominantly composed of small and medium-sized enterprises (SMEs). We then implement a staggered treatment difference-in-differences estimation strategy with 179 bank failures from 1990 to 2023 to estimate the effect of bank failure on firm-level survival and employment growth. We find that firms that had a credit relationship to a bank that fails are 6.7 percentage points (44.3%) more likely to fail themselves within five years of the bank failure. Additionally, firms that survive bank failures show 25% lower employment growth compared to firms banking with non-failed banks. These impacts of bank failure on firm performance persist for more than 10 years, are present for bank failures both during and outside the US financial crisis period, and are strongest for smaller enterprises. Our estimated effects are further supported by two natural experiments. Surprisingly, we observe that some bank failures had positive effects on firm outcomes, revealing that bank failure can, in rare cases, actually be fortuitous for affected firms. Overall, our findings suggest that bank failures exert a substantially larger influence on the real economy than previously recognized, possibly requiring a re-evaluation of current regulatory approaches to managing such events.

  6. 2023NBER WP 31007

    Remote Work across Jobs, Companies, and Space

    with Stephen Hansen, Nick Bloom, Steven Davis, Raffaella Sadun, Bledi Taska

    Abstract

    The pandemic catalyzed an enduring shift to remote work. To measure and characterize this shift, we examine more than 250 million job vacancy postings across five English-speaking countries. Our measurements rely on a state-of-the-art language-processing framework that we fit, test, and refine using 30,000 human classifications. We achieve 99% accuracy in flagging job postings that advertise hybrid or fully remote work, greatly outperforming dictionary methods and also outperforming other machine-learning methods. From 2019 to early 2023, the share of postings that say new employees can work remotely one or more days per week rose more than three-fold in the U.S and by a factor of five or more in Australia, Canada, New Zealand and the U.K. These developments are highly non-uniform across and within cities, industries, occupations, and companies. Even when zooming in on employers in the same industry competing for talent in the same occupations, we find large differences in the share of job postings that explicitly offer remote work.

  7. 2025CEP DP 2131

    Anatomy of Automation: CNC machines and industrial robots in UK manufacturing, 2005–2023 (PDF)

    with Aniket Baksy and Daniel Chandler

    Abstract

    Using a novel proprietary survey of UK manufacturing sites, we study the impact on employment of arguably the two most important industrial automation technologies of the past fifty years: computer numerical control (CNC) machine tools and industrial robots. First, we document the growing prevalence of both technologies across a wide range of industries between 2005 and 2023. Second, we use a local-projection difference-in-difference design to show that plants that adopt these technologies for the first time increase their employment by 6% to 9% compared to non-adopting plants in the same industry. Third, we find that for both technologies, automation is associated with an increase in employment among industry-competitor sites, and a positive overall impact on industry-level employment.

  8. 2024CEP

    Has Work from Home Shifted the US Electoral Map?

    with Chris Larkin

    CEP Occasional Paper No. 67

    Abstract

    Since 2020, the dramatic rise in remote work has coincided with increased geographic mobility in the United States. We examine the relationship between these trends and their effects on the electoral landscape. Using IRS microdata, online job postings, and Census surveys, we find that remote work opportunities concentrate in Democratic-leaning areas, with interstate migration strongly linked to individuals who mostly work from home. Our analysis reveals significant population shifts from Democratic to Republican and swing regions, potentially impacting electoral outcomes in key battleground states.

  9. 2019ADB

    Disaster Insurance in Developing Asia: An Analysis of Market-Based Schemes

    with Swenja Surminski and Architesh Panda

    ADB Economics Working Paper Series No. 590

    Abstract

    In recent years, insurance against natural disasters has gained recognition as an important tool for climate risk management that could, if carefully implemented, help increase the resilience of those insured. In response, insurance solutions are increasingly tested and applied in many countries that have no prior experience with insurance or no existing market. This paper analyzes the status, types, and patterns of market-based disaster insurance schemes across emerging and developing countries in Asia. We provide a snapshot of the current use of insurance based on data from Grantham Research Institute on Climate Change and the Environment’s Disaster Risk Transfer Scheme Database (2012–2018). Our analysis shows that although the use of insurance is expanding, there are many countries that still don’t have any kind of cover available. Where insurance mechanisms exist, they often rely on subsidies or bundling strategies. Although a mix of insurance schemes covering risks for governments (sovereign); or at meso (risk aggregators, cooperatives); and micro level currently operate to address a wide variety of climate and disaster risks, without demand-side support, many markets are likely to collapse or, at the very least, experience far lower penetration rates. We conclude with a discussion of the role of these insurance schemes in increasing resilience, which raises important questions for designing new and measuring and evaluating existing insurance schemes.

Publications

  1. 2026OxREP

    Critical Minerals and Industrial Policy: A Network-Based Approach to Supply Chain Risk

    with Thiemo Fetzer

    Oxford Review of Economic Policy, 42 (1), 193–211

    Abstract

    Policy debates on ‘critical minerals’ have multiplied faster than our empirical tools for identifying which products are system-critical. Most existing lists are expert-driven and static; they say little about how upstream raw materials and downstream technologies are knit together in global production networks, or about the asymmetric roles of large demand and supply hubs. This paper proposes a network index of criticality (NIC), built from trade data and a directed production network, that integrates (i) a product’s share in world trade, (ii) exporter and importer concentration, and (iii) its position in a product-input network. Criticality here refers to systemic exposure encoded in tradeable production networks; it is not a welfare metric and it is not a measure of physical scarcity. We show that NIC aligns with revealed policy attention in official critical-mineral lists. We then construct three families of counterfactuals that remove China, the United States, or the European Union from trade on either the export (supply) or import (demand) side. These scenarios yield product-level diagnostics of hub dependence and map directly to policy instrument choice (e.g. recycling standards, permitting and processing investment, strategic reserves, or trusted-partner agreements).

  2. 2023EconPol

    Measuring Remote Work Using a Large Language Model (LLM)

    EconPol Forum, 24 (3), 44–49

Work in progress

Policy writing and media

Open access data

Selected presentations