Scientific rewards for biomedical specialization are large and persistent.

BACKGROUND: While specialization plays an essential role in how scientific research is pursued, we understand little about its effects on a researcher's impact and career. In particular, the extent to which one specializes within their chosen fields likely has complex relationships with productivity, career stage, and eventual impact. Here, we develop a novel and fine-grained approach for measuring a researcher's level of specialization at each point in their career and apply it to the publication data of almost 30,000 established biomedical researchers to measure the effect that specialization has on the impact of a researcher's publications. RESULTS: Using a within-researcher, panel-based econometric framework, we arrive at several important results. First, there are significant scientific rewards for specialization-25% more citations per standard deviation increase in specialization. Second, these benefits are much higher early in a researcher's career-as large as 75% per standard deviation increase in specialization. Third, rewards are higher for researchers who publish few papers relative to their peers. Finally, we find that, all else equal, researchers who make large changes in their research direction see generally increased impact. CONCLUSIONS: The extent to which one specializes, particularly at the early stages of a biomedical research career, appears to play a significant role in determining the citation-based impact of their publications. When this measure of impact is, implicitly or explicitly, an input into decision-making processes within the scientific system (for example, for job opportunities, promotions, or invited talks), these findings lead to some important implications for the system-level organization of scientific research and the incentives that exist therein. We propose several mechanisms within modern scientific systems that likely lead to the scientific rewards we observe and discuss them within the broader context of reward structures in biomedicine and science more generally.

Buyers' role in innovation procurement: Evidence from US military R&D contracts.

This study provides the first quantification of buyers' role in the outcome of R&D procurement contracts. We combine together four data sources on US federal R&D contracts, follow-on patented inventions, federal public workforce characteristics, and perception of their work environment. By exploiting the observability of deaths of federal employees, we find that managers' death events negatively affect innovation outcomes: a 1% increase in the share of relevant public officer deaths causes a decline of 32.3% of patents per contract, 20.5% patent citations per contract, and 34.3% patent claims per contract. These effects are driven by the deaths occurring in the 6 months before the contract is awarded, thereby indicating the relevance of the design and award stage relative to ex post contract monitoring. Lower levels of self-reported within-office cooperation also negatively impact R&D outcomes.

Imputation of missing information in worldwide patent data.

We present a general method for imputing missing information in the Worldwide Patent Statistical Database (PATSTAT) and make the resulting datasets publicly available. The PATSTAT database is the de facto standard for academic research using patent data. Complete information on patents is essential to obtain an accurate picture of technological activities across countries and over time. However, the coverage of the database is far from complete. Our data imputation method exploits detailed institutional knowledge about the international patent system, and we codify it in a SQL algorithm. We provide two datasets related to the imputation of missing country codes and missing technology classification. We also release the algorithm that can be easily adapted to impute other pieces of information that are missing in PATSTAT.

Geocoding of worldwide patent data.

The dataset provides geographic coordinates for inventor and applicant locations in 18.8 million patent documents spanning over more than 30 years. The geocoded data are further allocated to the corresponding countries, regions and cities. When the address information was missing in the original patent document, we imputed it by using information from subsequent filings in the patent family. The resulting database can be used to study patenting activity at a fine-grained geographic level without creating bias towards the traditional, established patent offices.

The procurement of innovation by the U.S. government.

The U.S. government invests more than $50 billion per year in R&D procurement but we know little about the outcomes of these investments. We have traced all the patents arising from government funding since the year 2000. About 1.5 percent of all R&D procurement contracts have led to at least one patent for a total of about 13,000 patents. However, contracts connected to patents account for 36 per cent of overall contract value. The gestation lag from the signing date of the contract to the patent filing is on average 33 months and does not depend on the type of R&D performed. Patents that are produced faster also seem to be more valuable. We find strong decreasing returns to contract size. Conditional on generating at least one patent, a 1-percent increase in the size of an R&D contract is associated with 0.12 percent more patents.