Will Future Negotiations Really Be More Equal? How Equal First Offers Lead To Unequal Pay Outcomes In Gig-Economy Platform Negotiations

Abstract: Negotiations have always been central to social and economic interactions, but online labor-market platforms have transformed job negotiations by requiring workers (not employers) to make first offers for their pay rates, exposing all workers to “first-offer effects.” Proponents claim that such policies should help to decrease or eliminate pay gaps among workers with similar résumés, but this claim implicitly assumes that first-offer effects are not systematically unequal for different groups of workers. We find the opposite: analyzing negotiation data from over 10,000 platform workers, we show that first-offer effects are weaker for non-White workers (more specifically, those likely to be perceived as non-White). To identify plausibly causal effects of first offers, we leverage quasi-experimental variation in the anchoring effect, using the fact that people react disproportionately strongly when a price’s “left digit” is different even when the actual change in price is very small (e.g., when a price crosses just above a round number, such as being USD 40.01/hour rather than USD 39.99/hour). We find that, even after statistically adjusting for workers’ résumé characteristics and their first-offer amounts, making first offers just above (vs. just below) “left-digit” thresholds (e.g., USD 40.01/hour vs. USD 39.99/hour) triggers discontinuous pay boosts – but these boosts are 26% weaker for non-White workers. Importantly, whether first offers fall just above (vs. just below) a “left-digit” threshold appears to be quasi-random (uncorrelated with workers’ résumé characteristics or ethnicity), strengthening causal inference. Examining downstream implications, we show the résumé-adjusted pay gap for non-White (vs. White) workers is larger (not smaller) in platform negotiations (where workers make first offers) compared to benchmarks from traditional negotiations (where employers usually make first offers), and this pay gap becomes nonsignificant if we control for first-offer amounts and their interaction with ethnicity. Our results have important implications for platform design and policy.

Keywords: negotiations; field data; ethnicity; gig economy; quasi-experiment