The COVID-19 pandemic resulted in unprecedented disruptions and a reorganisation of the global economy. The impact of the pandemic on productivity is complicated. A growing literature seeks to understand productivity patterns during and following the pandemic (Bloom et al. 2021, Pionnier and Luu 2023, Dao and Platzer 2024). One industry particularly exposed to the disruptions of COVID was the restaurant industry, the second largest industry in the US in terms of employment. Initially this was due to its reliance on face-to-face interactions (Pető and Koren 2020), but it also became one of the industries most affected by labour shortages (Soldani et al. 2022).
It may come as a surprise, then, that US restaurant labour productivity has surged to a level some 15% higher than it was prior to COVID. This is even more startling given that productivity in the restaurant industry had been stagnant for decades. Figure 1 shows this flat trend in restaurant productivity followed by a surge that started during the pandemic and persisted even as overall economic conditions seemed to return to normal.
Figure 1 Annualised real sales per Employee (1992=100), food services and drinking places
Notes: Figure shows an index (1992 = 100) of annualised monthly real sales per employee for the US Food Services and Drinking Places industry (NAICS 722). Nominal seasonally adjusted sales are from the Census Monthly Retail Trade Survey report. Real sales obtained by deflating by CPI series for food away from home. Seasonally adjusted employment from the Bureau of Labor Statistics.
To understand this phenomenon, our recent paper (Goolsbee et al. 2025) utilises mobile phone data tracking customer visits and credit and debit card spending at over 100,000 limited-service restaurants across the US to examine potential explanations for the productivity surge. Analysing limited-service restaurants, which include fast food outlets and coffee shops, can provide insights into the entire restaurant industry. Limited-service restaurants comprise about 45% of industry employment and sales, with productivity patterns that mirror the overall industry.
Ruling out potential explanations
We rule out three potential explanations for the productivity surge: COVID-related demand fluctuations, economies of scale, and shifts in restaurant market power.
First, any explanation tying productivity to the collapse of restaurant demand during COVID would not explain the sustained rise in productivity because that collapse was temporary. In fact, real restaurant sales are currently about 20% higher than in the years prior to the pandemic.
Second, scale economies within restaurants cannot explain the productivity change. If COVID-related failures meant that resurgent demand led surviving restaurants to become larger and benefit from increasing returns to scale in their production processes, then labour productivity (although not total factor productivity) would increase. However, restaurants did not become larger. Average employment per restaurant in the sample dipped at the start of the pandemic and recovered after, but did not return to its pre-pandemic average until early 2022 – well after the productivity surge in early 2021. More comprehensive data for all restaurant types do not show an increase in employment per establishment.
Third, a shift in market power does not seem to be at work. Our preferred measure of labour productivity is real sales per employee. However, if price deflators do not perfectly capture market-power-based markup changes, some of the price increases could ‘spill’ into implied real spending, making labour productivity seem overstated. As a robustness check, we create an alternate measure of labour productivity: customer visits per employee. This is a proxy for real output that does not depend on the accuracy of the price series. We observe similar patterns in both the spending- and visits-based productivity measures and can therefore rule out a market power explanation.
The rise of short visits
Having dismissed these candidate explanations, we note a persistent change to the restaurant business during this period: customers are spending less time in restaurants when they visit. The share of customers spending ten minutes or less in restaurants grew significantly during the pandemic and has not receded. Figure 2 shows this evolution by plotting the shares of visits coming from a number of customer dwell time lengths. Average dwell times fell, and most of the reduction came from the rise in the share of the visits lasting less than ten minutes. This share had been steady before the pandemic at about half (recall that we examine limited-service restaurants, where sit-down meals are not the typical customer experience).
Figure 2 Average customer visit dwell time shares
Notes: Figure shows average share of panel restaurants’ customer visits by dwell time category from SafeGraph’s Monthly Patterns data. Shaded area indicates period after processing changes in SafeGraph’s Monthly Patterns, beginning in May 2022.
We believe that growth in take-out or delivery is the primary driver of the jump in short visits rather than faster service times for eat-in customers. If the increase in visits under ten minutes reflects decreased service times among all eat-in customers, one might expect that at least some of the growth would come out of 11 to 20-minute visits. This is not what happened in the data. The share of visits of between 11 and 20 minutes did not fall when the share of visits under ten minutes jumped. In fact, it increased slightly. We also found in other data sources that the usage of delivery apps surged at the start of the pandemic and has remained elevated.
With take-out, the customer orders on their phone and then comes into the restaurant to pick the food up without eating it there. With delivery, the customer orders food to be delivered to their home either via a food app or directly from the restaurant itself. It is important to recognise that either of these things connotes a substitution of home production for restaurant labour. The customer cleans up after themselves and washes their own dishes, for example. And delivery services substitute for the customer travelling. But from the restaurant’s perspective, they still represent a new stream of demand. If the restaurant can prepare meals for consumption for such quick-turn customers in addition to their regular customers with the same labour force, that would show up in the data as a clear, legitimate increase in their productivity.
Regression results show a strong correlation across restaurant levels between declines in customer visit times and higher productivity. We use the regression coefficients to calculate a total implied productivity growth of 11% due to shorter customer visit times. This is similar in magnitude to the observed change in limited-service restaurant industry aggregates, suggesting take-out and delivery can explain much of the observed aggregate increase in productivity. Similar patterns hold if we examine the relationship between productivity and visit times within specific restaurant chains, including McDonald’s, Chick-fil-A, Taco Bell, Wendy’s, and Burger King.
Conclusion
Restaurants experienced a persistent surge in productivity coming out of COVID that contrasted sharply with the multi-decade productivity stagnation that preceded it. Micro data on consumers suggest that this surge was strongly correlated with the rise in take-out and delivery customers staying only a short time in the restaurant. This striking result of a one-time surge in the level of productivity is particularly interesting in that restaurants adjusted their processes to meet the new ‘technology’ of customer demand that arose during COVID. This experience calls for future research into other industries that may have experienced similar bursts of service-sector productivity.
References
Bloom, N, P Bunn, P Mizen, P Smietanka and G Thwaites (2021), “The Impact of COVID-19 on Productivity”, VoxEU.org, 18 January.
Dao, M and J Platzer (2024), “Post-pandemic Productivity Dynamics in the United States”, IMF Working Paper 24/124.
Goolsbee, A, C Syverson, R Goldgof and J Tatarka (2025), “The Curious Surge of Productivity in US Restaurants”, NBER Working Paper No. 33555.
Pető, R and M Koren (2020), “It’s retail stores and restaurants, not farms and fisheries, that suffer most from social distancing”, VoxEU.org, 29 April.
Pionnier, P and N Luu (2023), “Aggregate labour productivity growth during the pandemic: The role of industry reallocations”, VoxEU.org, 7 March.
Soldani, E, O Causa, N Luu and M Abendschein (2022), “The post-COVID rise in labour shortages across OECD countries”, VoxEU.org, 28 November.
