Technology Diffusion Through Worker Mobility

Core Questions

This literature asks when worker mobility is a transmission mechanism for technology and knowledge rather than merely a reallocative labor-market outcome. The central objects are mobile engineers, scientists, inventors, R&D workers, and other knowledge workers whose departure from one firm and arrival at another can carry tacit know-how, project-specific learning, social ties, and problem-solving routines. A large share of the literature asks whether observed “knowledge spillovers” are truly disembodied externalities or instead are embodied in workers and networks, as in Almeida and Kogut (1999), Localization of Knowledge and the Mobility of Engineers in Regional Networks, Song, Almeida and Wu (2003), Learning–by–Hiring: When Is Mobility More Likely to Facilitate Interfirm Knowledge Transfer?, and Breschi and Lissoni (2009), Mobility of skilled workers and co-invention networks: an anatomy of localized knowledge flows.

A second major question is how labor-market frictions reshape the direction, rate, and quality of diffusion. Non-compete enforceability, trade-secret doctrines, firing costs, search frictions, buyout provisions, and geographic mobility costs can all alter whether ideas move, who captures the gains, and whether firms invest in innovation or instead in protection and retention. This branch asks not only whether mobility raises innovation, but also whether restricting mobility can sometimes raise certain kinds of innovation by strengthening incentives to invest, as in models by Shi (2023), Optimal Regulation of Noncompete Contracts, and Johnson, Lipsitz and Pei (2023), Innovation, Inventor Mobility, and the Enforceability of Noncompete Agreements. The broader macro question is whether these frictions raise or lower aggregate productivity, entrepreneurship, firm dynamics, and long-run growth.

A third question is interaction: how firms’ hiring, compensation, retention, secrecy, patenting, and collaboration strategies adapt to mobility conditions, and how those strategic responses feed back into economy-wide diffusion. Papers in this literature increasingly move from reduced-form spillover tests toward equilibrium models of search, poaching, imitation, and endogenous growth, as in Perla and Tonetti (2014), Equilibrium Imitation and Growth, Bilal, Engbom, Mongey and Violante (2022/2023), Firm and Worker Dynamics in a Frictional Labor Market and Labor Market Dynamics When Ideas are Harder to Find, and Heggedal, Moen and Preugschat (2017), Productivity spillovers through labor mobility in search equilibrium. The frontier concern is not just whether mobility matters, but what allocation of mobility is socially efficient and which policy instruments best approximate it.

Key Findings

The empirical consensus is that worker mobility is an important and often quantitatively meaningful conduit for knowledge diffusion, but the mechanism is heterogeneous. Hiring from productive, patenting, or technologically proximate firms tends to raise recipient-firm productivity or patenting, especially when the mover is skilled or has prior inventive experience. This is visible in Stoyanov and Zubanov (2012), Productivity Spillovers Across Firms through Worker Mobility, Parrotta and Pozzoli (2012), The effect of learning by hiring on productivity, Kaiser, Kongsted and Rønde (2014), Does the mobility of R&D labor increase innovation?, Serafinelli (2019), “Good” Firms, Worker Flows, and Local Productivity, and Braunerhjelm, Ding and Thulin (2020), Labour market mobility, knowledge diffusion and innovation. At the same time, the literature repeatedly finds that benefits are not evenly distributed: firms capture a large share of spillover gains, while workers often receive only a small wage premium, as in Stoyanov and Zubanov (2014), The distribution of the gains from spillovers through worker mobility between workers and firms, and Stoyanov and Zubanov (2013), Money on the Table? Firms’ and Workers’ Gains from Productivity Spillovers Through Worker Mobility.

A second robust finding is that legal and contractual restrictions on mobility usually reduce departures, job switching, and outside options, but their innovation consequences are more nuanced. Many papers find that stricter non-compete or trade-secret enforcement lowers inventor mobility and patenting, including Marx, Strumsky and Fleming (2009), Mobility, Skills, and the Michigan Non-Compete Experiment, Johnson, Lipsitz and Pei (2023), Innovation, Inventor Mobility, and the Enforceability of Noncompete Agreements, Contigiani, Hsu and Barankay (2018), Trade secrets and innovation: Evidence from the “inevitable disclosure” doctrine, and Shi (2023), Optimal Regulation of Noncompete Contracts. Yet some work documents composition effects rather than simple level effects: Xiao (2022), Non-competes and innovation: Evidence from medical devices, finds a shift toward more exploitative and fewer exploratory innovations with higher enforceability, while still observing a net increase in total innovations in that setting. This suggests that mobility restrictions may reallocate inventive effort along the exploitation-exploration margin rather than simply suppress innovation uniformly.

The frontier literature increasingly shows that mobility costs affect not only diffusion speed but also the geography and quality of knowledge flows. Papers on inventor diasporas, cross-border migration, and regional networks—Akcigit, Baslandze and Stantcheva (2015/2016/2016/2016/2016), Taxation and the International Mobility of Inventors; Prato (2024), The Global Race for Talent: Brain Drain, Knowledge Transfer, and Growth; Miguélez and Noumedem Temgoua (2019), Inventor migration and knowledge flows: A two-way communication channel?; Bahar, Choudhury, Miguélez and Signorelli (2024), Global Mobile Inventors—show that movement patterns can generate both source- and destination-side spillovers, with stronger effects when migrants remain connected to origin networks or when destination firms can absorb the knowledge. In parallel, macro-theory papers such as Benhabib, Perla and Tonetti (2021), Reconciling Models of Diffusion and Innovation, and Bilal et al. (2022/2023), Labor Market Dynamics When Ideas Are Harder to Find, emphasize that diffusion frictions shape the whole productivity distribution, not just average growth.

Identification & Credibility

The strongest empirical papers rely on quasi-experimental variation in mobility frictions or mobility opportunities. The Michigan non-compete reversal in Marx, Strumsky and Fleming (2009), Mobility, Skills, and the Michigan Non-Compete Experiment, is a canonical design: a difference-in-differences comparison around an inadvertent legal change provides evidence that enforceability attenuates mobility, especially for inventors with firm-specific skills and narrow technical specialization. The key identifying assumption is that absent the policy reversal, Michigan’s inventor mobility would have followed the counterfactual path of similar states after accounting for industry composition and auto-sector shocks. The outcome is cleanly a mobility response, and the limitation is that patenting or productivity effects are harder to isolate from concurrent industry restructuring and from the fact that the policy change may shift both who stays and who exits the state.

A second class of designs exploits staggered state-law reforms, especially changes in non-compete enforceability or adoption of the inevitable disclosure doctrine. Johnson, Lipsitz and Pei (2023), Innovation, Inventor Mobility, and the Enforceability of Noncompete Agreements, and Johnson, Lavetti and Lipsitz (2019/2023/2025), The Labor Market Effects of Legal Restrictions on Worker Mobility, use panel variation across states and cohorts to estimate effects on mobility and wages. These papers credibly identify reduced job-to-job transitions and lower earnings under stricter enforceability, often with larger effects for workers likely to sign non-competes. The main threat is policy endogeneity at the state level: reforms may correlate with broader legal or economic shifts. The better versions address this with rich fixed effects, border discontinuities, and cohort-based comparisons, but the estimates still mostly recover partial-equilibrium labor-market responses rather than general equilibrium innovation or productivity effects.

A third identification strategy is targeted legal shocks that alter the threat of departure or the credibility of disclosure, thereby changing firms’ incentives to patent, hire, or retain. Contigiani, Hsu and Barankay (2018), Trade secrets and innovation: Evidence from the “inevitable disclosure” doctrine, exploit state-level adoption of the doctrine and compare affected and unaffected inventors. The design is strong for inventor-level patenting responses, but the mechanism is not purely mobility: it combines reduced outside options, altered secrecy-versus-patenting incentives, and possibly shifts in signaling or career incentives. Similarly, Kang and Lee (2022), How innovating firms manage knowledge leakage: A natural experiment on the threat of worker departure, use a California court decision affecting enforceability against out-of-state noncompetes to study firms’ patenting and secrecy choices. These papers are valuable because they move from mobility itself to endogenous firm responses, but they remain vulnerable to interpreting patent counts as innovation quality rather than changes in disclosure behavior.

A fourth empirical family measures spillovers from hiring or worker flows using matched employer-employee or patent-assignee data, often with controls or instruments rather than sharp policy shocks. Stoyanov and Zubanov (2012), Productivity Spillovers Across Firms through Worker Mobility, and Kaiser, Kongsted and Rønde (2014), Does the mobility of R&D labor increase innovation?, estimate recipient and donor effects using linked Danish data; their identifying assumption is that observed hiring from productive firms is exogenous after conditioning on worker characteristics, firm fixed effects, and industry trends, which is only partly convincing. These studies are informative on correlational regularities and timing, but less definitive on causality because hiring is endogenous to unobserved demand, strategy, and match quality. By contrast, papers that exploit exogenous shocks to input supply or mobility—such as Prato (2024), the Quota Acts study in Yoon and Doran (2020), or Fackler, Giesing and Laurentsyeva (2019), Knowledge remittances—are stronger on causality because they induce movement through policy or supply shocks, though they often identify international migration effects that may not map one-to-one into firm-level labor frictions.

The theoretical literature is often cleaner on mechanisms than the empirics. Heggedal, Moen and Preugschat (2017), Productivity spillovers through labor mobility in search equilibrium, and Shi (2023), Optimal Regulation of Noncompete Contracts, model mobility frictions, bargaining, and buyouts explicitly. Their main contribution is to show how limited commitment can cause underinvestment in innovation or excessive rent extraction, and to derive policy prescriptions such as subsidizing innovation or restricting non-competes. The limitation is not identification but calibration and structural validity: welfare results depend on how well the search, bargaining, and innovation technologies approximate the institutions studied. Perla and Tonetti (2014), Equilibrium Imitation and Growth, and Bilal et al. (2022/2023) similarly identify logical comparative statics in imitation-based growth models, but their aggregate implications depend on assumptions about search, heterogeneity, and the speed at which ideas are found.

Methods

The empirical literature is unusually data-rich. A central empirical asset is matched employer-employee data, often combined with patent records, which allows researchers to trace movers, compare donor and recipient firms, and distinguish high-skill from low-skill workers. Danish and Swedish administrative panels, U.S. patent-inventor linkages, LinkedIn resume data, European patent records, and state-level firm-worker datasets are the most common sources. Patent citations, patent quality metrics, and firm productivity or patenting outcomes serve as the main measures of knowledge diffusion, while mobility is measured through job-to-job transitions, inventor-assignee changes, spatial moves, or career exits to startups.

Methodologically, the field combines reduced-form quasi-experiments with structural models. Reduced-form papers use staggered legal reforms, court decisions, tax changes, transport improvements, and immigration quotas to isolate changes in mobility or knowledge flows. Structural papers model search, bargaining, on-the-job search, and imitation/innovation choices to map worker movement into equilibrium innovation, wages, and welfare. The strong point of the structural approach is counterfactual policy analysis; the strong point of the reduced-form approach is sharper causal interpretation of specific mobility shocks. The best current work increasingly pairs them, using reduced-form evidence to discipline the mechanism and structural models to extrapolate aggregate effects.

Open Questions & Gaps

The biggest unresolved issue is welfare: the literature is much better at showing that mobility affects innovation and wages than at measuring the full general-equilibrium tradeoff between diffusion gains and investment incentives. Many papers find that non-competes or secrecy protections reduce mobility and sometimes raise firm investment, but it is still unclear when those private incentives translate into social gains rather than rent extraction or misallocation. Shi (2023), Optimal Regulation of Noncompete Contracts, and Heggedal, Moen and Preugschat (2017), Productivity spillovers through labor mobility in search equilibrium, are frontier attempts, but their policy conclusions depend heavily on how buyouts, bargaining power, and search frictions are parameterized. A stronger empirical frontier would combine legal shocks with measures of firm investment, innovation quality, worker outside options, and downstream product-market outcomes in a unified panel.

A second gap concerns heterogeneity in the quality and direction of knowledge diffusion. The evidence is strongest that mobility helps when workers move from more productive, patent-active, or technologically proximate firms, but the field still lacks a settled map of when movement produces exploratory versus exploitative innovation, and when it yields genuine recombination versus copying. Xiao (2022), Non-competes and innovation: Evidence from medical devices, and Wang (2026), Labor Mobility Restrictions and Exploratory Innovation, point to a shift in the composition of innovation under mobility restrictions, while Johnson, Lipsitz and Pei (2023), Innovation, Inventor Mobility, and the Enforceability of Noncompete Agreements, emphasize aggregate patent declines. The missing piece is a common outcome framework that measures novelty, value, technological breadth, and commercialization jointly rather than relying on patent counts alone.

A third gap is the interaction between mobility and firm strategy. We know firms respond to mobility with patents, secrecy, wages, collaboration, and retention, but it remains unclear how these margins are coordinated. Papers such as Kang and Lee (2022), How innovating firms manage knowledge leakage, and Ekinci and Wehrheim (2023), Holdup, Knowledge Transferability, and Productivity, suggest that mobility frictions alter effort and disclosure incentives, while Balasubramanian, Chang, Sakakibara, Sivadasan and Starr (2020), Locked In? The Enforceability of Covenants Not to Compete and the Careers of High-Tech Workers, shows wage and career effects. What is not yet settled is whether firms substitute between compensation, training, patenting, and organizational design in a way that offsets or amplifies the mobility restriction. Better evidence would require matched worker-employer-contract data with legal variation and measures of internal training, promotion, and knowledge-management practices.

A fourth frontier question is scale and geography. Much of the literature is local, regional, or sectoral, while the most policy-relevant questions are aggregate: how do mobility frictions affect business dynamism, regional convergence, and long-run growth? The macro literature—Akcigit and Ates (2020), Ten Facts on Declining Business Dynamism and Lessons from Endogenous Growth Theory; Bilal et al. (2022/2023), Labor Market Dynamics When Ideas Are Harder to Find; Benhabib, Perla and Tonetti (2021), Reconciling Models of Diffusion and Innovation—makes clear that diffusion from frontier to laggard firms matters for aggregate outcomes, but empirical evidence directly linking worker mobility frictions to economy-wide productivity remains sparse. The ideal next step would integrate micro mobility shocks with measures of firm entry, incumbent churn, concentration, and regional productivity over long horizons, so that the literature can move from “mobility matters” to a quantitatively credible answer on how much and through which channels.