Innovation Job Market Papers 2023
Dozens of paper abstracts
In this special edition of What’s New Under the Sun, we have a big bundle of the titles, abstracts, and links to innovation-related PhD job market papers from 2023 that I either found or were sent to me in response to last week’s solicitation (thank you!). This is not an exhaustive list - I am sure I have missed many great papers. If you have a paper that you think belongs on this list, please send it my way following the instructions here, and I’ll add it.
I enjoyed reading all these abstracts, and am excited to dig into the papers. Back to our regular programming next week!
Titles Index
Titles are presented in random order.
Do Standard Error Corrections Exacerbate Publication Bias? by Patrick Vu
Machines and Superstars: Technological Change and Top Labor Incomes by Donghyun Suh
I, Google: Estimating the Impact of Corporate Involvement on AI Research by Daniel Yue
Returnee Inventors and Home Country Innovation by Sherry Xue
Executive contracts for sustainable innovation: incentivising gains in wealth and health by Slavek Roller
Measuring Knowledge Capital Risk by Pedro H. Braz Vallocci
Multinational Production and Innovation in Tandem by Jin Liu
Staggered Rollout for Innovation Adoption by Ricardo Fonseca
Spillovers and the Direction of Innovation: An Application to the Clean Energy Transition by Eric Donald
Technology Adoption, Learning by Doing, and Reallocation by T. Jake Smith
The Effect of Funding Delays on the Research Workforce: Evidence From Tax Records by Wei Yang Tham, with Joseph Staudt, Elisabeth Ruth Perlman, Stephanie Cheng
Batman Forever? The role of trademarks for reuse in the US comics industry by Franziska Kaiser, with Alexander Cuntz, and Christian Peukert
Race and Science by Gaia Dossi
When are Patents Traded and Why: A Dynamic Structural Model of Drug Development and Patent Trading by Jie Fang
The Effect of Robot Assistance on Skills by Sungwoo Cho
Worker Mobility, Knowledge Diffusion, and Non-Compete Contracts by Jingnan Liu
Equilibrium IPR Protections, Innovation and Imitation in A Globalized World by Leo C.H. Lam
Public R&D Spillovers and Productivity Growth by Arnaud Dyèvre
Optimal Skill Mixing Under Technological Advancements by Elmer Zongyang Li
STEMming the Gender Gap in the Applied Fields: Where are the Leaks in the Pipeline? by Shasha Wang
Reluctant to Grow: The Unintended Effects of R&D Tax Credits Targeting Small Firms by Alexandre Lehoux
Technological Change and Unions: An Intergenerational Conflict with Aggregate Impact by Leon Huetsch
Innovation-Facilitating Networks Create Inequality by Cody Moser, with Paul Smaldino
Embracing the Future or Building on the Past? Growth with New and Old Technologies by Bernardo Ribeiro
Intangible Assets, Knowledge Spillover, and Markup by Yusuf Ozkara
Money, Time, and Grant Design by Wei Yang Tham, with Kyle Myers
The Effects of the Affordable Care Act on Pharmaceutical Prices, Demand and Innovation by Zhemin Yuan
Multidimensional Skills in Inventor Teams by Hanxiao Cui
Return Innovation: The Knowledge Spillovers of the British Migration to the United States by Davide M. Coluccia, with Gaia Dossi
Information provision and network externalities: the impact of genomic testing on the dairy industry by Victor Funes-Leal, with Jared Hutchins
Reveal or Conceal? Employer Learning in the Labor Market for Computer Scientists by Alice H. Wu
Innovation and Technological Mismatch: Experimental Evidence from Improved Crop Seeds by Sergio Puerto
Relying on Intermittency: Clean Energy, Storage, and Innovation in a Macro Climate Model by Claudia Gentile
Intellectual Mobility Frictions by Jordan Bisset, with Dennis Verhoeven
Consequences of Indian Import Penetration in the US Pharmaceutical Market by Jinhyeon Han
Markups, Firm Scale, and Distorted Economic Growth by Jean-Felix Brouillette, with Mohamad Adhami, Emma Rockall
Teacher-directed scientific change: The case of the English Scientific Revolution by Julius Koschnick
Strategic Network Decisions and Knowledge Spillovers: Evidence from R&D Collaborations of U.S. Firms by Kippeum Lee
The Market Effects of Algorithms by Lindsey Raymond
Decline in Entrepreneurship: A Tale of Two Types of Entrepreneurs by Angelica Sanchez-Diaz
Scale-Biased Technical Change and Inequality by Hugo Reichardt
The Effect of Inventor Mobility on Network Productivity by Brit Sharoni
Bread Upon the Waters: Corporate Science and the Benefits from Follow-On Public Research by Dror Shvadron
Titles and Abstracts
Do Standard Error Corrections Exacerbate Publication Bias?
Patrick Vu
Over the past several decades, econometrics research has devoted substantial efforts to improving the credibility of standard errors. This paper studies how such improvements interact with the selective publication process to affect the ultimate credibility of published studies. I show that adopting improved but enlarged standard errors for individual studies can lead to higher bias in the studies selected for publication. Intuitively, this is because increasing standard errors raises the bar on statistical significance, which exacerbates publication bias. Despite the possibility of higher bias, I show that the coverage of published confidence intervals unambiguously increases. I illustrate these phenomena using a newly constructed dataset on the adoption of clustered standard errors in the difference-in-differences literature between 2000 and 2009. Clustering is associated with a near doubling in the magnitude of published effect sizes. I estimate a model of the publication process and find that clustering led to large improvements in coverage but also sizable increases in bias. To examine the overall impact on evidence-based policy, I develop a model of a policymaker who uses in- formation from published studies to inform policy decisions and overestimates the precision of estimates when standard errors are unclustered. I find that clustering lowers minimax regret when policymakers exhibit sufficiently high loss aversion for mistakenly implementing an ineffective or harmful policy.
Machines and Superstars: Technological Change and Top Labor Incomes
Donghyun Suh
I construct a model of production hierarchies in which agents and machines differ in skill levels. The skill level of an agent determines the difficulty of work tasks she can perform. Relatively low-skill agents become workers and high-skill agents become managers who help workers perform difficult tasks. Machines can either augment or substitute workers. Two main findings emerge: First, whether machines augment or substitute workers depends on the highest skill level of machines. If machines can perform sufficiently difficult tasks, then machines substitute workers and augment managers. However, if machines can only perform relatively easy tasks, then machines augment workers. Second, for sufficiently advanced machines, technological change increases income concentration at the top. This occurs as gains from technological change are greater for those with higher skills, thus benefiting the most skilled managers the most. By contrast, if machines augment workers, technological change has the opposite effect on top income shares. The paper further examines the implications of Artificial Intelligence (AI) for managerial functions. If machines can perform more difficult tasks than any worker, they substitute managers. I find that management by machines most significantly raises the wages of the least skilled workers. On the other hand, managers' wages fall, with the decline most pronounced among the least skilled managers. Therefore, while less inequality between workers and managers leads to lower top income shares, the inequality among managers increases.
I, Google: Estimating the Impact of Corporate Involvement on AI Research
Daniel Yue
While corporate involvement in modern scientific research is an indisputable fact, the impact of corporate involvement on scientific progress is controversial. Corporate interests can lead to constraints that redirect research activities into applied problems in a way that benefits the company but reduces scientific impact. However, corporations also provide resources such as funding, data sets, collaborators, engineers, and technical problems that researchers may otherwise be unable to access or know about, spurring knowledge creation. This paper empirically assesses the impact of corporate involvement on scientific research by focusing on dual-affiliated artificial intelligence researchers located at the intersection of academia and industry. After controlling for the researcher's quality and topic preferences, I find that corporate involvement leads to up to a 44% increase in field-weighted citations received by a paper. I document evidence that this effect arises because the average benefit of a firm's scientific resources exceeds the cost of that firm's scientific constraints. Specifically, I show that corporate involvement significantly increases the likelihood of a breakthrough paper and that these effects are magnified by the involvement of firms with greater resources. However, corporate involvement also alters the direction of the dual-affiliate author's research to be more aligned with the firm's commercial interests. This is the first large-scale quantitative study of any field of science to demonstrate a direct positive effect of corporate involvement on science or to describe the underlying mechanism.
Returnee Inventors and Home Country Innovation
Sherry Xue
I analyze the innovations produced by Chinese companies and research organizations (”receivers”) after hiring returnee inventors – Chinese inventors who returned from abroad. Following their return, receivers significantly increase patenting and the number of involved inventors in technological fields where the returnee has experience. However, the new patents receive fewer citations, especially from abroad. Additionally, there is a decreased proportion of patents involving foreign inventors, indicating that returnee inventors substitute for international collaboration. These findings continue to hold when the timing of the return is plausibly exogenous, and the effects are stronger when the returnee is from the United States. My results suggest that returnee inventors play a significant role in knowledge transfer to China while also reducing the flow of knowledge from China to other countries through reduced collaboration.
Executive contracts for sustainable innovation: incentivising gains in wealth and health
Slavek Roller
I propose a framework of contracting for sustainable innovation. The magnitude of innovation (minor vs. major) reflects competitiveness, measured as appropriated net social surplus (profit) from the new market share gained due to innovation. In contrast, the magnitude of sustainable innovation reflects value creation, measured as net social surplus gained in the economy - without limiting said surplus to material prosperity gains. I consider how sustainable finance regulations and practices can influence sustainable innovation via the design of executive contracts. I apply this framework of contracting for sustainable innovation to explain the patterns of social value creation in the pharmaceutical industry. To this end I create a novel dataset in which the therapeutic improvement of each novel medicine proxies net social surplus. I classify innovation metrics based on the beneficiary of the value that the innovation ought to create: shareholders (incentives linked to appropriated social surplus) or society (incentives linked to generated social surplus). The former either explicitly tie innovation to economic success of new products (“sales for new products launched”; “product pipeline with the long-term growth of the company in mind”; “new molecular entities with favourable potential impact on long-term revenue growth rate”) or in their quantity (“number of reimbursable product approvals”; “registrational volume”; “deliver two new molecular entity (NME) launches per year”). The latter tie innovation to patient benefit (“delivering more years of life and quality of life for people around the world”; “patient value”) or therapeutic advance (“obtain breakthrough designation from the FDA”; “highly statistically significant treatment effect”, “number and level of innovation of the products”). I find that companies using executive compensation incentives linked to social surplus creation (with performance metrics measuring sustainable innovation) have a higher rate of subsequent medical advances - medicines with significant or breakthrough therapeutic improvement - than companies using incentives linked to social surplus appropriation. This finding points to a possibility for an optimal innovation policy that instead of directly targeting innovating firms (patent laws, taxes, government-funded innovation prizes, etc.) targets the decision-making agents in those firms via, for example, sustainable finance regulations and corporate governance provisions therein.
Measuring Knowledge Capital Risk
Pedro H. Braz Vallocci
This study proposes a methodology to identify firms that are vulnerable to knowl- edge capital-related risks, relying on a textual analysis of the risk factors disclosed in their annual reports. Further, the paper quantifies these risks through an examination of firms’ concurrent return patterns.
Multinational Production and Innovation in Tandem
Jin Liu
Multinational firms colocate production and innovation by offshoring them to the same host country or region. In this paper, I examine the determinants of multinational firms’ production and innovation locations. I find complementarities between production and innovation within host countries and regions, exploiting plausibly exogenous variation in tariffs. To evaluate manufacturing reshoring policies, I develop a quantifiable multicountry offshoring location choice model. I allow for rich colocation benefits and cross-country interdependence and prove supermodularity of the model to solve this otherwise NP-hard problem. I find the effects of manufacturing reshoring policies are nonlinear, contingent upon firm heterogeneity, and they accumulate dynamically.
Staggered Rollout for Innovation Adoption
Ricardo Fonseca
I consider a mechanism design approach to innovation adoption and show how it is optimal for the principal to induce artificial scarcity to speed it up. Take-up of a new product generates information about its value for others, so agents want to free-ride before irreversibly adopting it themselves. This causes a time-delay externality that a principal seeking to achieve an adoption target as quickly as possible (for example, a government trying to reach herd immunity through vaccination while agents are uncertain of their personal vaccination benefits, not internalizing the positive externality of reaching the adoption target) seeks to avoid. Scarcity speeds up learning because it limits free-riding. I show that the possibility of imposing supply restrictions is always beneficial compared to free supply. I also show that optimal supply plans are simple in that there is a batched supply release with fewer batches than agents' value types. I fully characterize such optimal plans for settings with up to three types and show that (non-optimal) supply plans may be Pareto improving.
Spillovers and the Direction of Innovation: An Application to the Clean Energy Transition
Eric Donald
This paper argues that cross-technology knowledge spillovers are critical for understanding policy’s role in the transition to clean technology. I develop an endogenous growth model with clean and dirty technologies and a network of cross-technology spillovers. I derive formulas for the size and speed of technological transition, following a policy reform, which show that greater spillovers across technologies induce a faster transition but at the expense of a smaller long-run impact of policy. Such spillovers also prevent the lock-in of dirty technology. The economy’s spillover structure can be summarized by a sufficient statistic matrix, which I estimate using patent citation data. Applying my model to US transportation and electricity generation, I find that cross-technology spillovers are mid-sized: they prevent lock-in but imply a slow transition with a high long-run impact of policy. I conclude by examining how cross-technology spillovers affect optimal clean innovation subsidies. Contrary to conventional wisdom, I find that optimal clean innovation subsidies are small, and this holds quantitatively even in the absence of optimal carbon prices. This is because innovation policy should reward a technology’s centrality in the spillover network: i.e. the extent to which it enables future innovation. In my calibration, clean technologies have low centrality, resulting in small subsidies.
Technology Adoption, Learning by Doing, and Reallocation
T. Jake Smith
As Schumpeter observed, the process of technological change involves creative destruction of the existing economic structure. In this paper, I theoretically and empirically investigate creative destruction in agricultural production due to a major new technology, genetically engineered (GE) crop varieties. Specifically, I study how consolidation of farmland into larger farms owned by fewer farmers was affected by the introduction and adoption of GE corn seed. I hypothesize that successfully using GE seeds requires adopting farmers to learn about the new technology. I develop a model to show that this learning by doing leads to higher adoption rates for farmers with more land, which in turn allows these larger farmers to profitably buy land from smaller farmers. Using detailed data on the adoption of GE corn and U.S. Census of Agriculture data, I present novel empirical evidence consistent with the predictions of the learning-by-doing model. I then use a difference-in-difference approach, guided by the theoretical model, to estimate the impact that the GE seed introduction had on farmers exiting the market. I estimate that the introduction of GE seeds led to an exit of about 18,300 corn- growing operations, or 4% of the corn-growing operations existing in 1997. This amount of exit represents 13% of the net exit of nearly 150,000 corn-growing operations between 1997 and 2017.
The Effect of Funding Delays on the Research Workforce: Evidence From Tax Records
Wei Yang Tham, with Joseph Staudt, Elisabeth Ruth Perlman, Stephanie Cheng
We study how an interruption in the flow of research funding from a major NIH grant — the R01 — affects the career outcomes of research personnel. Using comprehensive earnings and tax records linked to university transaction data along with a difference-in-differences design, we find that an interruption of more than 30 days has a substantial effect on job placement for personnel who work in a lab supported by a single R01, including a 2.5 pp increase in the probability of not working in the US. Half of those induced into nonemployment in the US are absent from the 2020 Decennial Census, suggesting that these personnel have permanently left the US. Among personnel who continue to work in the US, we also find that interrupted personnel earn 20% less than their continuously-funded peers. Trainees (postdocs and graduate students) experience the largest increase in the probability of no longer being in the US. Non-faculty and non-trainee personnel (such as staff and undergraduates) have the largest earnings impact
Batman Forever? The role of trademarks for reuse in the US comics industry
Franziska Kaiser, with Alexander Cuntz, and Christian Peukert
We study how trademarks affect reuse of creative works in the comics industry. As a creative industry, the comics industry systematically relies on copyrights. But trademark protection can also be exploited to generate income from the reuse of comic characters or to strategically exclude others from reuse. Our unique data set combines US trademark records of comic characters with information on reuse in print media and franchise products from 1990 to 2017. We find that, on average, additional trademark protection is associated with a reduction in reuse in printed comic books of about 19%. We highlight three mechanisms: first, the negative relationship between trademarking and reuse has been especially pronounced since the early 2000s, when the arrival of digital technologies lowered the costs of entry, promotion, and distribution. Second, our results are driven by less reuse by third parties, not trademark holders. Third, reuse is higher when trademark owners license comic characters to third parties. The negative association between trademarking and reuse carries over to franchise products, but it is weaker and tied to the era of digitization, with a 2% decline in reuse in franchise movies and 9% lower reuse in video games.
Race and Science
Gaia Dossi
What are the consequences of the racial gap in science and innovation? I study this question by combining data on US patents, medical research articles, clinical trials, and research grants with the racial distribution of last names in the US population. Using last names as a proxy for race, I find that the racial composition of scientists affects the direction, as well as the rate, of medical research and innovation. First, I show that Black scientists are more likely to research diseases frequent among Blacks, while White scientists are more likely to research diseases frequent among Whites. Second, across all diseases, Black scientists are three times as likely to design clinical trials with Black participants and twice as likely to publish articles focused on Black individuals. Third, I draw a causal link between race and the direction of research by focusing on diseases more common among Black individuals (e.g., sickle cell anemia) or White individuals (e.g., melanoma) due to evolutionary advantages in their ancestors’ countries of origin. Fourth, I document the causal impact of relative disease incidence on the direction of research by studying an exogenous change in HIV-related mortality among Blacks compared to Whites. I formalize a general equilibrium Roy model with racial frictions and endogenous choice of occupation. Using the data, I quantify the parameters and estimate that removing barriers would increase the overall number of inventors by 1 p.p., a 10% increase from the baseline.
When are Patents Traded and Why: A Dynamic Structural Model of Drug Development and Patent Trading
Jie Fang
Reallocating patents to firms more proficient in their utilization can improve welfare. Moreover, the timing of such trades significantly impacts innovation outcomes. I construct a unique dataset that captures interactions between patent trades and the drug development within the U.S. pharmaceutical sector, and find that 72% of patents are traded before the associated drug hits the market. Drugs involved in patent trades are also more likely to advance to the launch stage compared to those without patent trades. I construct a dynamic structural model for the development process of a new drug, taking into account crucial factors such as trade dynamics, firms’ comparative advantages, transaction costs, and search frictions at various stages of the development process, encompassing discovery, clinical trials, FDA approval, and product launch. The estimation of this model reveals that (i) firms with greater stage-specific experience enjoy reduced development costs at the corresponding stage; (ii) transferring patents to firms with lower development costs enhances the likelihood of a drug advancing to subsequent stages; and (iii) market frictions in patent trading exhibit significant variation across different phases of drug development, with transaction costs reaching their peak prior to FDA registration. Counterfactual analyses show that reducing transaction costs within the patent market at pivotal stages significantly increases the likelihood of drug success and the market value of the drug.
The Effect of Robot Assistance on Skills
Sungwoo Cho
How does working with robots change human capital? To examine how collaboration with robots affects human skills, I exploit a unique setting in which professional baseball leagues provided, and subsequently removed, access to robot assistance for umpires. Umpires demonstrated improved precision and accuracy in ball-strike decisions while using robot assistance, and their performance declined substantially below preassistance levels after it was removed. Both highly skilled and inexperienced umpires exhibited large declines in performance after the removal of robot assistance. Umpires who used robot assistance for longer periods of time faced a steeper decline in accuracy than those who used it for shorter periods. In addition, umpires who worked a full season with robot assistance did not fully return to their initial skill level by the end of the following season. By examining a canceled season during the COVID-19 pandemic, I reject that skill depreciation is solely a result of umpires simply not using their skills. Umpires also experience skill deterioration in determining whether a baserunner is safe, suggesting that the findings are widely applicable to various occupational settings with a similar skill set.
Worker Mobility, Knowledge Diffusion, and Non-Compete Contracts
Jingnan Liu
This paper studies how endogenous worker mobility affects inter-firm knowledge diffusion, innovation, and economic growth. I propose a framework combining endogenous growth and on-the-job search. Firms grow knowledge by in-house innovation and by hiring workers from more productive firms. Knowledge is nonrival, leading to underinvestment in innovation. Non-compete contracts address this underinvestment by allowing innovating firms to enforce buyout payments when they lose workers. However, they discourage diffusion by deterring firm entry. Linking administrative data on patents, firm performance, employment history, and wages from the U.S. Census Bureau, I document that inventors diffuse knowledge across firms and are compensated for knowledge diffusion. Constructing novel micro-level data, I find non-compete contracts are associated with increased innovation expenditure and decreased worker mobility. I calibrate my theoretical model to match the empirical results. Knowledge diffusion, through the channel of worker mobility, accounts for 4% of the TFP growth rate and 8% of welfare. Optimal regulation of non-compete contracts balances the innovation-diffusion tradeoff.
Equilibrium IPR Protections, Innovation and Imitation in A Globalized World
Leo C.H. Lam
What determines the levels of intellectual property rights (IPR) protection in a globalized economy? How does it impact developing and developed countries? To answer these questions, this paper develops a two-country endogenous growth model in which firms can innovate and imitate locally and globally, and governments choose IPR policies strategically. Beyond the traditional static cost and dynamic benefit from IPR protection, this framework spotlights three key dynamic burdens: limiting domestic imitation, restricting global imitation, and discouraging escape-competition innovation. From the empirical analysis on patent assignments and patent litigation data between the US and China, I find (i) the US domestic distribution of technologies is relatively stable, while China’s domestic distribution transformed from the 1990s to the early 2000s and has stabilized since 2005; (ii) the global distribution of technology has transformed from a US-dominant position to become more even; (iii) both local and global imitation are positively correlated with technology gaps. The quantitative analysis suggests three main findings. First, strengthening IPR enforcement leads to an inverted-U-shaped long-run growth rate. Second, governments’ horizon matters. Strong short-run benefits of imitation cause both governments to pick weaker IPR policies when they consider transitional welfare. Lastly, the Nash equilibrium suggests that both countries should enforce lower IPR protections. The over-protection results in significant welfare losses of 6.4% and 7.2% for the US and China, respectively.
Public R&D Spillovers and Productivity Growth
Arnaud Dyèvre
Does the source of Research and Development funding, public or private, matter for aggregate productivity growth? Using a unique firm-level dataset with patent and balance-sheet information covering 70 years (1950-2020), I estimate the impact of the decline in public R&D in the US on long-run productivity growth. I first document three new facts about publicly-funded innovations: they are (i) more reliant on science, (ii) more likely to open new technological fields, and (iii) more likely to generate knowledge spillovers, especially toward smaller firms. I then use two instrumental variable strategies—a historical shift-share IV and a patent examiner leniency instrument—to estimate the impact of the decline in public R&D on the productivity of firms through spillovers. I find that a 1% decline in public R&D spillovers causes a 0.17% decline in productivity growth. Public R&D spillovers are three times as impactful as private R&D spillovers for firm productivity and their impact persists at the sector level. Moreover, smaller firms experience larger productivity gains from public R&D spillovers. I calibrate a model of growth with heterogeneous firms which suggests that the decline in public R&D can explain around a third of the decline in TFP growth in the US from 1950 to 2018, and half of the rise in size inequality between firms over the same period.
Optimal Skill Mixing Under Technological Advancements
Elmer Zongyang Li
Using worker surveys and online job posting data, I document that the U.S. economy has seen a substantial increase in the mixing of skill requirements from 2005-2018, both for incumbent jobs and newly posted vacancies. American workers increasingly work in occupations that demand mixtures of analytical, computer, and interpersonal skills rather than specializing in one of them, even within granular occupations. This change occurred primarily in low- to medium-wage occupations, and workers in occupations that increasingly mix non-routine skills, or those with a broader set of these skills earn a wage premium. To understand the sources of these shifts, I build a directed search model with multi-dimensional skills in which firms optimally choose occupations’ skill intensities before producing with a worker, delivering endogenous specialization in skill demand. Counterfactual analysis shows that the rise in the complementarity of skills in production and in the cost of skills for occupation operation are the main drivers of skill mixing shifts and the corresponding wage and employment dynamics in this period.
STEMming the Gender Gap in the Applied Fields: Where are the Leaks in the Pipeline?
Shasha Wang
In pure-STEM fields, women are no longer a minority, but in applied-STEM fields like computer science and engineering, their representation remains persistently low for nearly half a century. To understand where the leaks are in the pipeline, I develop and estimate a dynamic model spanning from high school to early career to examine four sources of female under-representation: initial skill gaps, preference differences, wage disparities in STEM sectors, and aversion to male-dominated occupations. In the Na- tional Longitudinal Survey of Youth - 1979 Cohort, males show a higher interest in STEM coursework and better STEM skills by 10th grade, primarily in mechanical skills, leading to wider skill disparities. Simulation results show that mechanical skills are more important than math skills in explaining women’s low participation in applied-STEM fields and have contrasting effects on college enrollment and the selection of applied-STEM majors and occupations. Closing gender skill gaps upon exiting high school reduces female under-representation by 67% in applied-stem majors and 31% in applied-STEM occupations. Removing the preference for female-dominated workplaces reduces female under-representation by 29% in applied-STEM majors and 85% in applied-STEM occupations. Equalizing wage offers in STEM sectors has a smaller effect (3% in majors and 10% in occupations). Mandating more high-school STEM courses increases overall STEM participation but doesn’t address the gender gap.
Reluctant to Grow: The Unintended Effects of R&D Tax Credits Targeting Small Firms
Alexandre Lehoux
This paper provides evidence at the firm- and worker-level of how R&D tax credits targeting small innovating firms distort production and earnings. I take advantage of an eligibility change in Canada's largest R&D program in 2004 that allowed firms to increase their production while maintaining eligibility for the generous program. Using matched employer-employee data, I find no impact on R&D spending in the short-term, but significant increases in value-added per worker following the reform. The results are primarily driven by less financially constrained firms, emphasizing the growth distortion effect of the eligibility threshold. Productivity growth results in earnings increasing on average by 2%. Incumbent and lower-earning workers benefited the most, while women saw no increase. Finally, I find no impact on employment, although this is masked by an increase in new hires and departures following the reform. Hiring policy tilts towards recruiting workers coming from higher firm quality, but I see no effect on the individual quality of new recruits.
Technological Change and Unions: An Intergenerational Conflict with Aggregate Impact
Leon Huetsch
Technological progress in the form of automation boosts productivity, but can cause adverse labor market outcomes for transitional generations. I study the role of unions in shaping the outcomes of workers exposed to labor replacement during the automation transition. Using variation across local labor markets in the U.S., I first document that unionization has shifted the incidence of wage and employment declines within routine-manual occupations from older, incumbent to young, incoming cohorts since 1980. Second, unions have accelerated the overall employment decline within these occupations, measured as a greater decline early in the transition, and a subsequent catch-up in less unionized labor markets after 2000. I develop a quantitative model of technological change and unionization which jointly rationalizes the two empirical observations through the interaction of union-imposed firing costs and gradual automation adoption over time. Within automating occupations, unions reduce the welfare cost of automation to older workers along the transition by up to 4% of permanent consumption, lowering their layoff risk and wage decline. The impact is shifted to young workers, raising the welfare costs for cohorts entering during the transition by up to 2%. Incoming workers endogenously respond to automation by entering non-adopting occupations which limits the welfare impact on them. The impact of high unionization spills over into non-adopting occupations as the accelerated reallocation of labor suppresses wages there.
Innovation-Facilitating Networks Create Inequality
Cody Moser, with Paul Smaldino
Theories of innovation often balance contrasting views that either smart people create smart things or smartly constructed institutions create smart things. While population models have shown factors including population size, connectivity and agent behaviour as crucial for innovation, few have taken the individual-central approach seriously by examining the role individuals play within their groups. To explore how network structures influence not only population-level innovation but also performance among individuals, we studied an agent-based model of the Potions Task, a paradigm developed to test how structure affects a group’s ability to solve a difficult exploration task. We explore how size, connectivity and rates of information sharing in a network influence innovation and how these have an impact on the emergence of inequality in terms of agent contributions. We find, in line with prior work, that population size has a positive effect on innovation, but also find that large and small populations perform similarly per capita; that many small groups outperform fewer large groups; that random changes to structure have few effects on innovation in the task; and that the highest performing agents tend to occupy more central positions in the network. Moreover, we show that every network factor which improves innovation leads to a proportional increase in inequality of performance in the network, creating ‘genius effects’ among otherwise ‘dumb’ agents in both idealized and real-world networks.
Embracing the Future or Building on the Past? Growth with New and Old Technologies
Bernardo Ribeiro
Is growth driven by the emergence of new paradigms or mostly through the perfection of existing technologies? And is the allocation of research effort between emerging technologies versus established ones efficient? To study these questions, I propose a new semi-endogenous growth model that incorporates technology vintages and the endogenous evolution of multiple technological paradigms through directed innovation. Despite the fact that technologies continuously emerge, making the state space unbounded, the model is remarkably tractable, allowing me to provide a comprehensive characterization of both the balanced growth equilibrium and the transitional dynamics. From a positive perspective, the model can rationalize two distinct empirical patterns of innovation over time and across technologies. Using two centuries of U.S. patent data, I first document that the age profile of patents has a pronounced hump-shape: the majority of contemporary patents are built upon technologies that are between 70 and 100 years old. Second, this age profile has remained remarkably stable throughout the past century. From a normative standpoint, the theory underscores a misallocation of research effort induced by the tendency among profit-maximizing firms to overinvest in further developing mature technologies. This fundamental inefficiency yields a suboptimally slow development of emerging technologies near the technological frontier. An estimated version of my model implies that transitioning from a laissez-faire equilibrium to the efficient allocation would increase the average growth rate of the economy from an annual 2% to 2.18% over the course of a century. These results shed new light on policy discussions concerning the prioritization of emerging technologies versus established ones. For instance, they provide a rationale for public policy to support investments in cutting-edge technologies, such as quantum computing or metabolic engineering.
Intangible Assets, Knowledge Spillover, and Markup
Yusuf Ozkara
Intangible assets (e.g. software) have unique characteristics compared to physical capital; they are scalable and exhibit spillover effects. This paper develops a structural model to empirically test these features of intangible assets. I introduce intangible capital into the production function as an additional factor input and external knowledge as a productivity shifter. I estimate production functions at the firm level including labor-augmenting, and Hicks-neutral productivity without imposing any parametric functional form. My empirical results indicate a positive and significant impact of intangible capital on a firm’s production. This return to intangibles increases with firm size in all sectors, suggesting that intangible capital exhibits scalability. Moreover, knowledge spillovers increase firm productivity, and the extent of this increase varies depending on firm size, and sector. Large firms and firms in the health sector tend to benefit more from their rival’s knowledge stock. Additionally, I reveal that markups rise with a firm’s intangible intensity, suggesting a potential explanation for the recent rise in market concentration.
Money, Time, and Grant Design
Wei Yang Tham, with Kyle Myers
The design of research grants has been hypothesized to be a useful tool for influencing researchers and their science. To better understand the value of grant design as a policy instrument, we conduct two sets of thought experiments in a nationally representative survey of academic researchers. First, we test whether grants with randomized attributes induce different research strategies. Longer grants increase researchers’ willingness to take risks, but only among tenured professors, suggesting that job security and grant duration are complements. Larger grants increase researchers’ willingness to expand ongoing projects, while smaller grants increase researchers’ focus on starting new projects in new directions, the opposite of what conventional theory would suggest. Both longer and larger grants reduce researchers’ focus on speed, which suggests a significant amount of racing in science is in pursuit of resources. In our second experiment, we find that researchers are relatively unwilling to trade off the amount of funding a grant provides in order to extend the duration of the grant — more money is much more valuable than more time. Our results have implications for organizations that fund science and uncover new characteristics of scientific production functions.
The Effects of the Affordable Care Act on Pharmaceutical Prices, Demand and Innovation
Zhemin Yuan
As the largest health insurance expansion in the U.S. since Medicaid and Medicare, the Patient Protection and Affordable Care Act (ACA) has differential impacts on different age groups. This paper hypothesizes and tests that the ACA demand shock mainly affects medicines targeting conditions that are prevalent in the working age population. Exploiting plausibly exogenous variation across medical conditions in their exposure to the ACA, this paper studies the causal effects of the ACA-driven demand shock on pharmaceutical prices, demand, and innovation. I find that medicines targeting conditions that are prevalent among younger workers were more cheaply priced before the reform because they were less likely to be covered by insurance. The insurance expansion of the ACA has eliminated this price discount as younger workers become less price elastic. I find that the ACA redistributes preclinical research and development (R&D) efforts on conditions that are prevalent in younger workers toward conditions with high prevalence among older workers. Under the assumption that innovation improves the quality of drugs, this redistribution of R&D efforts may improve labor productivity more effectively in the long run, because older workers, compared with their younger counterparts, are more in need of health-enhancing medicines. I find that the post-ACA increase in clinical R&D is driven by the influx of earlier preclinical developments rather than the immediate push of on-the-shelf projects.
Multidimensional Skills in Inventor Teams
Hanxiao Cui
I study the complementarity of multidimensional skills in innovation production and the skill composition of inventor teams. Using patent data linked to inventor social security records and establishment panels in Germany, I construct inventor skills from labor market biographies and uncover a mismatch pattern in inventor teams: There is positive sorting of inventors’ skills even though teams with diverse skills have higher productivity. Quasi-experimental evidence from inventor mobility caused by establishment closures rules out endogeneity or selection bias as the reason behind the discrepancy, pointing instead to search frictions arising from inventor type segregation across labor markets. To rationalize the observed allocation, I build a team formation model in which firms assemble inventor teams for innovation, subject to search costs that are increasing as firms’ chosen team mix deviates from the market composition of inventors. I show that search costs and inventor segregation can reinforce each other in equilibrium, driving the excessive positive sorting in talent allocation. Absent search costs, the share of homogamous teams will decrease by 28.6pp, boosting total innovation by 2.4%.
Return Innovation: The Knowledge Spillovers of the British Migration to the United States
Davide M. Coluccia, with Gaia Dossi
How does innovation diffuse across countries? In this paper, we document that out-migration promotes the diffusion of innovation from the country of destination to the country of origin of migrants. Between 1870 and 1940, nearly four million British immigrants settled in the United States. We construct a novel individual-level dataset linking British immigrants in the US to the UK census, and we digitize the universe of UK patents over 1853-1899. Using a new shift-share instrument for bilateral migration and a triple-differences design, we document that migration ties contribute to technology diffusion from the US to the UK. Through high-dimensional text analysis, we find that emigrants promote technology transfer, but they also nurture the production of original innovation. Physical return migration is an important driver of this “return innovation” effect. However, we find that the interactions between emigrants and their origin communities promote technology diffusion, even absent return migration. Additionally, we show that migration ties propel knowledge flows by fostering cross-border market integration.
Information provision and network externalities: the impact of genomic testing on the dairy industry
Victor Funes-Leal, with Jared Hutchins
We use a differences-in-differences with a matched control group method to estimate the long-term impacts of genomic selection in the American market for dairy cattle genetics. Genomic selection is an application of big data that uses the entire genome of an animal to test for the presence of a set of traits. Unlike pre-existing technologies that require several years of data from a bull's daughters, an animal can be tested as soon as it is born, allowing breeders to identify the “best" animals much faster. Using a data set of all bulls marketed in the US from 2000 to 2020, we find that genomic selection significantly increased genetic gains for all measured traits, particularly milk production, protein, and fat yields, but also increased levels of inbreeding depression, a reduction in the performance of animals whose parents have a high degree of relatedness, as a consequence of genetics companies breeding more animals from established lines to respond to an increased “brand” loyalty towards such lines. Our estimation shows that the increased inbreeding rate of American bulls caused a loss of between 3.6 to 6.7 billion dollars to the entire industry from 2011 to 2019. Solving this externality will require either a mechanism to internalize the harmful effects, such as paying a much higher price for more inbred sires, or a collective action mechanism to select which lines will be bred in the next generation.
Reveal or Conceal? Employer Learning in the Labor Market for Computer Scientists
Alice H. Wu
The efficient allocation of labor relies on the identification of talent. When employee output is not publicly observable, employers have an incentive to take advantage of private information, potentially leading to the misallocation of labor among firms. This paper provides empirical evidence of employer learning and quantifies the impact of learning on job mobility and innovation outputs in the labor market for computer science (CS) Ph.D.’s. CS conference proceedings provide public information on research effort by existing CS workers. Among papers authored by researchers from industry, about one-quarter can be matched to a contemporaneous patent application - an indicator of a more valuable innovation. Yet the fact of the application remains private information at the incumbent employer for 18 months. Consistent with public learning, researchers with a new paper have higher inter-firm mobility rates than do coworkers without a paper. Initially, authors of papers with a matched patent are less likely to move than authors without a patent application. But once the patent application becomes public, their mobility rates cross over. Authors of papers with a matched patent are also 35% more likely to move to a top tech firm. These patterns confirm the predictions of a model in which incumbent firms have initially private information on more productive researchers. Structural estimates of the model suggest that if papers and patents were disclosed simultaneously, high-ability workers would sort more quickly to high-productivity firms. The implied increase in allocative efficiency would increase innovation outputs by about 5%.
Innovation and Technological Mismatch: Experimental Evidence from Improved Crop Seeds
Sergio Puerto
Biases in research and development create a mismatch between the attributes of new agricultural technology and the preferences of farmers. In this paper, I estimate the impact of this mismatch on farmers’ adoption of new drought-resistant seeds. Using a randomized controlled trial in Costa Rica, I recreated counterfactual scenarios for innovators’ seed development decisions by offering some farmers seed matching their preferences and others a seed variety chosen by crop scientists as a blanket recommendation. Results show that mismatch has a significant impact on adoption, with 41% lower uptake among farmers who were offered the recommended new seed. This gap was larger for farms located farther from the research lab where the new seeds were developed, and persisted even in areas with drought exposure. I also find that the new seed varieties were 31% more productive among farmers who adopted their preferred seed. To explain these findings, I propose a model where research constraints limit innovators’ ability to account for farmer heterogeneity. Matching new seeds to farmer preferences relaxes those constraints, and improves productivity by enabling better adaptation to specific farm-level conditions, which are usually private information unknown to innovators. These findings highlight that agricultural innovation is often shaped by innovators’ priorities rather than demand-side signals, especially in developing countries.
Relying on Intermittency: Clean Energy, Storage, and Innovation in a Macro Climate Model
Claudia Gentile
The transition to clean energy technologies is essential to reduce CO2 emissions. One significant challenge associated with renewable energy sources, such as solar and wind, is their intermittency. I study the intermittency problem by introducing a novel micro-founded energy sector with directed technical change in a macro climate model. I show that the aggregate elasticity of substitution between clean and dirty energy crucially depends on the development of storage technologies. If the storage technology is not developed, the economy is trapped in a scenario in which the elasticity of substitution eventually becomes zero. Without policies, the provision of storage technologies is inefficiently low, impeding the transition towards clean, intermittent technologies. In the optimal allocation, the clean energy transition is accelerated with an initial clean energy share increasing from 25% to 70% and a reallocation of all R&D resources away from dirty energy towards clean energy and, in particular, energy storage technologies. The introduction of clean energy subsidies under the US Inflation Reduction Act is successful in increasing the short-run clean energy share, but insufficient to solve the intermittency problem.
Intellectual Mobility Frictions
Jordan Bisset, with Dennis Verhoeven
In order for innovation policy to function optimally, an increase in the demand for invention must be met with a corresponding increase in supply. In this paper we document the presence of a human capital friction which increases the cost of inventor mobility between technological fields, and hampers short- run inventor supply responsiveness. We build a simple model to estimate this friction and find it to be large, even between fields which are relatively close in the intellectual space. The friction varies widely across wider domains of knowledge application – such as Organic Chemistry and Environmental Technology. We estimate the friction reduces the effectiveness of innovation policy by almost 40%. At the aggregate level, our calculations suggest the friction increased the cost of US invention by over $340 Billion between 1990 and 2015.
Consequences of Indian Import Penetration in the US Pharmaceutical Market
Jinhyeon Han
Indian pharmaceutical firms vastly increased generic drug exports to Western countries, and particularly the United States, in response to India amending its patent protection law in 2005. This paper demonstrates that Indian import penetration in the U.S. pharmaceutical industry was stronger among drugs that included active pharmaceutical ingredients which Indian firms had the capability to manufacture before 2005. These ‘treated’ drugs experienced a larger drop in prices and increased competition from Indian firms. However, non-Indian firms responded to the trade shock by exiting ‘treated’ drugs, in that they were more likely to discontinue and less likely to newly register ‘treated’ drugs. This strategic change was stronger for firms with larger shares of drug sales exposed to Indian import penetration. Rather than exploring new pharmaceutical ingredients, non-Indian firms opted to continue selling drugs they already had experience producing. The law change also led to disruptions in foreign supply chains, with ‘treated’ drugs facing more drug shortages. These findings show how competition in the pharmaceutical industry is shaped by supply side shocks and how a trade-driven influx of competition could crowd out incumbent firms.
Markups, Firm Scale, and Distorted Economic Growth
Jean-Felix Brouillette, with Mohamad Adhami, Emma Rockall
We study the consequences of markups for long-run economic growth in a model of firm-driven endogenous technological change. In this framework, differentiated firms engage in monopolistic competition, charge heterogeneous markups, and make forward-looking investments in R&D to improve their process efficiency. Markups distort the scale at which these firms operate and, therefore, affect their incentives to invest in R&D. With dispersion in markups, both the aggregate and cross-firm allocations of such investments are distorted. Using firm-level administrative data from France to discipline our model, we find that correcting the product market distortions induced by markups increases the long-run growth rate of productivity by 1.2 percentage points per year. Nearly 75% of this faster productivity growth can be achieved by simply reallocating R&D resources across firms, revealing that the dispersion in markups, rather than their average level, is more detrimental to economic growth.
Teacher-directed scientific change: The case of the English Scientific Revolution
Julius Koschnick
While economic factors in directed technical and scientific change have been widely studied, the role of teacher-directed scientific change has received less attention. This paper studies teacher- directed scientific change for one of the largest changes in the direction of research, the Scientific Revolution. Specifically, the paper considers the case of the English Scientific Revolution at the English universities of Oxford and Cambridge. It argues that exposure to different teachers shaped students’ direction of research and can partly account for the successful trajectory of English science. For this, the paper introduces a novel dataset on the universe of all 111,242 students at English universities in the seventeenth and early eighteenth century and matches them to their publications. Using machine learning, the paper is able to quantify personal interest in different research topics. To derive causal estimates of teacher-student effects, the paper exploits a natural experiment based on the expulsion of fellows following the English Civil War and uses an instrumental variable design that predicts students’ choice of college based on their home regions. The paper finds strong empirical evidence of teacher-directed change in the English Scientific Revolution. These results illustrate how teacher-directed change can contribute to paradigm change.
Strategic Network Decisions and Knowledge Spillovers: Evidence from R&D Collaborations of U.S. Firms
Kippeum Lee
This paper examines the effect of private R&D investment on productivity, considering R&D collaborations and knowledge spillovers. While existing literature emphasizes R&D’s direct effects on innovation and cost reduction, it often neglects R&D’s role in shaping collaborative networks. Investing in R&D enhances a firm’s learning capacity and augments the firm’s appeal as a collaboration partner. Consequently, the effect of R&D is underestimated without accounting for its role in fostering collaborations. To bridge the gap, I develop a dynamic model of a firm that internalizes its decision on whom to collaborate with and following spillovers. This framework allows R&D to improve productivity and affect the collaboration network, with varying propensities for collaborations across firms. Using the data on firm-to-firm R&D collaborations among U.S. firms from 1980 to 2001, I find the long-term effect of R&D is 16% underestimated if we ignore its subsidiary role in expanding the collaboration network.
The Market Effects of Algorithms
Lindsey Raymond
While there is excitement about the potential of algorithms to optimize individual decision-making, changes in individual behavior will, almost inevitably, impact markets. Yet little is known about these effects. In this paper, I study how the availability of algorithmic prediction changes entry, allocation, and prices in the US residential real estate market, a key driver of household wealth. I identify a market-level natural experiment that generates variation in the cost of using algorithms to value houses: digitization, the transition from physical to digital housing records. I show that digitization leads to entry by investors using algorithms, but does not push out investors using human judgment. Instead, human investors shift towards houses that are difficult to predict algorithmically. Algorithmic investors predominantly purchase minority-owned homes, a segment of the market where humans may be biased. Digitization increases the average sale price of minority-owned homes by 5% or $5,000 and nearly eliminates racial disparities in home prices. Algorithmic investors, via competition, affect the prices humans pay for minority homes, which drives most of the reduction in racial disparities. This decrease in racial inequality underscores the potential of algorithms to mitigate human biases at the market level.
Decline in Entrepreneurship: A Tale of Two Types of Entrepreneurs
Angelica Sanchez-Diaz
Entrepreneurship in the United States has declined in recent decades. Using household survey data, I show that this decline is driven by the falling share of unincorporated self- employment (i.e., sole proprietorships and partnerships), while the share of incorporated self-employment (i.e., S and C corporations) has risen. This pattern is robust across de- mographic characteristics and data sources. To understand these trends, I build a general equilibrium heterogeneous-agent model with occupational choices and two types of self-employment. To investigate the source of the aggregate trends in entrepreneurship, I conduct counterfactual experiments. I evaluate two potential factors observed in the data over the same period: (i) an investment-specific technological change and (ii) a decline in tax progressivity. The results show that the main driver of declining entrepreneurship is tech- nological change, whereas the decline in tax progressivity played a minor role.
Scale-Biased Technical Change and Inequality
Hugo Reichardt
Scale bias in technical change is the degree to which technical change increases the productivity of large relative to small firms. I propose that this dimension of technical change is important for inequality. I first develop a tractable framework with heterogeneous households choosing to work for wages or earn profits as entrepreneurs. Entrepreneurs choose from a set of available production technologies, defined by a fixed and a marginal cost. Large-scale-biased technical change lowers entrepreneurship rates and leads to larger firms on average. With fewer and larger firms, top entrepreneurs are capturing a larger share of the profits which increases top income inequality. Small-scale-biased technical change has the opposite effects. I test the theory by comparing the effects of two technologies that vary in scale bias, but are similar in purpose: steam engines (large-scale-biased) and electric motors (small-scale-biased). Using newly collected data from the United States and the Netherlands, I verify that these two technologies had opposite effects on firm sizes and inequality. Steam engines increased firm sizes, while electric motors decreased them. Steam engines led to increased inequality, electric motors did not. Consistent with scale bias (rather than skill bias), I find that adopting entrepreneurs were the main drivers of inequality increases after steam engine adoption.
The Effect of Inventor Mobility on Network Productivity
Brit Sharoni
When inventors move to new locations, they carry knowledge and expertise, which may be a loss to their previous collaborators. But they might also become a bridge between otherwise disconnected innovation hubs, facilitating information flows and idea diffusion. In this paper, I study the effect of an inventor’s relocation on their previous collaborators’ productivity. A simple patent production model addresses the dual role of relocators as former collaborators and as intermediaries providing access to information. The model helps to guide the empirical analysis and to interpret the results. Empirically, I build a novel dataset combining information about inventors from the USPTO patent data with online professional profiles. Using a matching design, I find sizeable positive effects on the productivity of inventors whose collaborators have relocated. These effects pertain not only to quantity, as gauged by the number of patents, but also to quality, as measured by the number of citations. I show that the core mechanism driving both effects is greater access to novel information networks and information.
Bread Upon the Waters: Corporate Science and the Benefits from Follow-On Public Research
Dror Shvadron
Why do firms produce scientific research and make it available to the public, including their rivals? An important but hitherto ignored benefit is that it can influence the direction of research conducted by external scientists in ways that benefit the focal firm. I show that external scientists often build upon a firm’s publications, producing follow-on findings, which the firm then incorporates into its own future innovations. To account for the unobserved quality of the science involved, I develop a new instrumental variable that relies on the quasi-random assignment of accepted manuscripts to specific issues of scientific journals. Some publications attract more academic attention simply because they appear alongside contributions from prominent authors in the same journal issue. Using data on scientific publications by public firms between 1990 and 2012, I find that follow-on research not only drives firms’ subsequent investments in science but also improves their patenting outcomes. The benefits are more pronounced for technological leaders, firms with complementary assets, and those operating in emerging research fields. In addition to being a valuable input into the firm’s innovations, follow-on research also helps validate the quality of the firm’s internal science, especially when there is greater uncertainty surrounding said science. My findings contribute to the understanding of why firms participate in public science.
Thanks for reading! As always, if you want to chat about innovation, let’s grab a virtual coffee. Send me an email at matt.clancy@openphilanthropy.org and we’ll put something in the calendar.