Since the COVID-19 pandemic, work from home (WFH) has become a fixture of the labour market in many countries. For instance, Aksoy et al. (2024) report that across 34 countries, a third of full-time workers engaged in hybrid or fully remote work arrangements in 2023. This widespread adoption has sparked considerable interest from policymakers seeking to understand the implications of WFH for productivity, inequality, employment dynamics, and city structure (e.g. Eurofound 2024).

The implications of work from home on wages are particularly interesting. A growing literature finds that workers who work from home earn on average higher wages than workers who do not. An important part of this premium reflects occupational and educational differences, since higher-paying occupations and better-educated workers are more likely to work from home. However, recent evidence shows that an economically meaningful wage premium persists even among workers with the same occupation, education, and other observable characteristics (Rossi-Hansberg et al. 2023, Pabilonia and Vernon 2025).

At first glance, this finding is puzzling. A large body of research shows that workers typically value WFH as an amenity and are willing to accept sizable wage cuts in exchange for it (Mas and Pallais 2017). One potential explanation is that working from home raises worker productivity, prompting firms to pay higher wages (Pabilonia and Vernon 2025). Yet this raises a further question: if productivity gains are substantial, why do not all workers in teleworkable jobs work from home?

An alternative explanation for the observed work-from-home wage premium is selection: workers with higher unobserved productivity or greater bargaining power may be more likely both to secure higher wages and more flexible WFH arrangements. In recent work (Li et al. 2026), we provide evidence that selection is indeed the main driver of the work-from-home wage premium in France.

Data and methodology

Our analysis combines three administrative data sources. First, the French Labour Force Survey provides information on workers’ WFH status, self-reported earnings, hours worked, and key individual characteristics such as education, occupation, gender, and age. From this, we calculate hourly wages as earnings divided by hours worked. Second, we link these data to the universe of firm registry records, which allow us to observe employer characteristics including sales, workforce size, and firm age. This linkage is important, as a large literature documents substantial and persistent wage differences across firms (Card et al. 2013, Song et al. 2018). Finally, we match workers to social security records, which provide detailed information on their earnings and employer histories. Together, these data allow us to study wage differences associated with work from home while accounting for both worker and firm heterogeneity.

Work from home and wages

We begin by documenting the relationship between WFH and worker wages in the post-pandemic period, from mid-2022 – after all COVID-related restrictions had been lifted – through the end of 2024. In this sample, we first document a sizeable raw wage gap between workers who work from home and those who do not. As shown in the first bar in Figure 1, on average, the hourly wages of WFH workers are 35% higher than those of on-site workers.

Much of this gap, however, reflects occupation and education differences. In France (as in other developed economies), remote work is far more common in higher-paid occupations and among more educated workers. Once we look within occupation, education, and location, the wage premium associated with WFH shrinks substantially, to about 12% (second bar in Figure 1). Controlling further for other observable worker characteristics – such as age, gender, and job tenure – reduces the wage premium to 6.6% (third bar in Figure 1). While lower, this is still a significant number. This result survives numerous robustness checks including using alternative measures of hourly wages and WFH status.

Figure 1 WFH hourly wage premium (%)

What could be driving this substantial difference in wages? One possible interpretation is that the difference is driven by employers. Indeed, it might well be that higher-paying firms are also more likely to offer more flexible work arrangements. However, we find that this explanation is not borne out by the data: the premium remains even after controlling for firm characteristics, or when comparing workers within the same firm. The fourth bar in Figure 1 shows that adding controls for firm age, size, and revenue productivity only brings the premium down from 6.6% to 6.4%. This suggests that firm-level pay policies alone cannot account for the work-from-home wage premium.

Next, we explore whether the premium could be driven by selection. To do so, we leverage the social security data to extract worker’s wages before COVID-19. This reveals a striking pattern. Workers who work from home after the pandemic were already earning higher wages before the COVID-19 pandemic. Even after controlling for a rich set of worker and firm characteristics, pre-pandemic hourly wages – measured as the average wage in 2018–19 – can account for virtually the entire remaining wage premium. The last bar of Figure 1 shows that adding the pre-COVID wage control brings the premium down to 1.1% and is not statistically significant. 

In other words, the workers who are working from home post-pandemic were already paid higher wages before WFH became widespread. We also find that WFH is not associated with higher post-pandemic wage growth, as one might have expected if this type of work led to significant productivity gains. Overall, this evidence points to selection as a central driver of the WFH wage premium: workers who are more productive, or have better negotiation skills, are able to get both higher hourly wages and the right to work from home more often.

Our findings have several important implications. First, they suggest that conventional wage measures may understate inequality in a world with widespread WFH, as higher-paid workers are also more likely to receive the amenity of being able to work from home. Second, they place a bound on potential productivity gains from working from home (or, alternatively, suggest that employers have not shared these productivity gains with their workers). Third, by highlighting the role of selection, they imply that changes in WFH policies – such as return-to-office mandates – can affect the allocation of talent across firms and, potentially, aggregate productivity.

References

Aksoy, C G, J M Barrero, N Bloom, S Davis, M Dolls and P Zarate (2024), “How and why work-from-home rates differ across countries and people”, VoxEU.org, 1 May.

Aksoy, C G, N Bloom, S Davis, V Marino and C Özgüzel (2025), “Fully remote work expands recruitment and boosts productivity”, VoxEU.org, 1 June.

Card, D, J Heining and P Kline (2013), “Workplace Heterogeneity and the Rise of West German Wage Inequality”, Quarterly Journal of Economics 128(3).

Eurofound (2022), The rise in telework: Impact on working conditions and regulations, Publications Office of the European Union.

Li, H, J Sauvagnat and T Schmitz (2026), “The Work-from-Home Wage Premium”, CEPR Discussion Paper 20996.

Mas, A and A Pallais (2017), “Valuing Alternative Work Arrangements”, American Economic Review 107(12).

Monte, F, C Porcher and E Rossi-Hansberg (2023), “Work from home in cities: A tale of coordination and multiple equilibria”, VoxEU.org, 7 September.

Pabilonia, S W and V Vernon (2025), “Remote work, wages, and hours worked in the United States”, Journal of Population Economics 38(18).

Song, J, D Price, F Guvenen, N Bloom and T von Wachter (2018), “Firming Up Inequality”, Quarterly Journal of Economics 134(1): p1-50.