Can Technology Startups Hire Talented Early Employees? Ability, Preferences, and Employee First Job Choice

Abstract

Early stage technology startups rely critically on talented scientists and engineers to commercialize new technologies. And yet these startups compete with established technology firms to hire the best workers. Theories of ability sorting predict that high-ability workers will choose jobs in established firms that offer greater complementary assets and higher pay, leaving low-ability workers to take lower paying and riskier jobs in startups. We propose an alternative view in which heterogeneity in both worker ability and preferences enable startups to hire talented workers who have a taste for a startup work environment even at lower pay. Using a longitudinal survey that follows 2,394 science and engineering PhDs from graduate school into their first industrial employment, we overcome common empirical challenges by observing ability and stated preferences prior to entry into the labor market. We find that both ability and career preferences strongly predict startup employment with high-ability workers who prefer startup employment being the most likely to work in a startup. We show that this partly reflects dual selection effects whereby worker preferences result in a large pool of startup job applicants and startups make job offers to the most talented workers. Additional analyses confirm that startup employees earn approximately 17% lower pay. This gap is greatest for high-ability workers and persists over workers’ early careers, suggesting that they accept a negative compensating differential in exchange for the nonpecuniary benefits of startup employment. Data on job attributes and stated reasons for job choice further support this interpretation. This paper was accepted by Toby Stuart, entrepreneurship and innovation. Funding: This work was supported by the National Science Foundation SciSIP Award [Grant 1262270] and the Ewing Marion Kauffman Foundation (Junior Faculty Fellowship). Supplemental Material: The online appendix and data are available at https://doi.org/10.1287/mnsc.2023.4868 .

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 48 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.