How to design predictable scheduling laws that not only benefit workers but also firms’ bottom line?

Over the past few decades, much of the media and policy debate around labor issues have focused on low wages. Labor issues related to work schedules have received far less attention. In fact, 17% of the U.S. labor force works on unpredictable or unstable schedules with short advance notice (Golden 2015). They are disproportionately concentrated in lower paid occupations in the retail and service sectors. According to a national survey on retail jobs, 87% of retail workers report hour variations in the past month with the average variation equivalent to 48% of their usual work hours, 50% report a week or less advance notice, and 44% say that their employer decides their work hours without their input (Lambert et al. 2014). The prevalence and the rapid growth of unpredictable and unstable schedules has resulted in many social issues, including difficulties arranging childcare and threats to households’ economic security (Henly and Lambert 2014).

The economic trade-off of predictable schedules and the ongoing policy debate

Unpredictable and unstable schedules are so prevalent in service businesses, because labor accounts for a significant part of the operating cost of service businesses, especially in retail, food, and hospitality services. Having just enough (but not too many) workers on hand is essential to balancing customer service and profitability. As firms try to strike that balance, many—especially those in the service and retail sectors—practice “just-in-time” (JIT) scheduling, which entails managers scheduling their employees “on the fly” based on immediate workplace needs. By using just-in-time scheduling, service firms mitigate the uncertainty they often face in customer demand and employee no-shows. This helps them reduce the labor hours needed and thus labor cost (Terwiesch and Cachon 2012). While JIT scheduling can be effective in reducing firms’ labor costs, it also leads to highly unpredictable and fluctuating schedules for workers, which negatively impact their quality of life, especially among low-income workers. In short, firms have been using JIT scheduling to transfer business risks to their employees.

Recent local and state policies aim to reduce this practice. Since 2014, one state (Oregon) and multiple cities (e.g., Chicago, Los Angeles, New York, Philadelphia, San Francisco, Seattle, and Emeryville, California), have passed various forms of “predictive scheduling laws,” sometimes also referred to as “fair workweek laws.” In general, they require employers to post work schedules in advance and provide additional pay for any last-minute schedule changes. Some versions of such laws, (e.g., the ones in New York City, Seattle and Emeryville, California), also require employers to offer part-time workers the chance to increase their hours before adding new staff (Wolfe et al 2018).

Service firms, especially those in the retail, food, and hospitability industries, argue that such requirements remove the staffing flexibility they need to operate their businesses effectively, which may lead to bankruptcy and eventually loss of jobs. Indeed, such policies have received strong resistance from employers in the service and retail sectors and are still pending or have failed to pass in many cities and states across the U.S. States including Arkansas, Georgia, Iowa, and Tennessee even prohibit jurisdictions within the states from passing predictable scheduling laws. Predictable scheduling laws also differ in the level of advance notice they require firms to inform their workers about their schedules. For example, the city of New York requires 72 hours advance notice (for its retail workers) and the state of Oregon initially required one-week advance notice but later increased to 14 days, while most other cities require 14 days advance notice.

Is JIT scheduling really that beneficial to service firms?

In light of this debate, Masoud Kamalahmadi (University of Miami), Yong-Pin Zhou (University of Washington) and I conducted a study to answer whether and to what extent the flexibility created through just-in-time scheduling benefits the firm and how policy makers can better design predictable scheduling laws (Kamalahmadi et al. 2021). On the one hand, it is clear that just-in-time scheduling helps firms reduce their labor cost as explained earlier. On the other hand, the potential impact of just-in-time scheduling on the workers’ productivity, and thus the firm’s revenue, was not well understood. It was the goal of our study to seek objective evidence that can shed light on this important issue.

Data and Analysis

To conduct the study, we collected a large and granular dataset from a full-service national casual dining restaurant chain in the United States. The restaurant setting is an ideal context for our study for several reasons: (1) the restaurant industry is large and contributes nearly $1 trillion annually to the U.S. economy (National Restaurant Association 2023); (2) service in restaurants is labor intensive—the industry is the second-largest private sector employer in the United States, currently employing around 15 million workers (about 10% of the U.S. workforce) (National Restaurant Association 2023); (3) the restaurant industry is one of the largest employers of part-time workers who are often subject to unpredictable work schedules because of the practice of just-in-time scheduling (Lambert et al. 2014).

We used transaction-level data (approximately 1.4 million transactions in total) generated at the restaurant chain’s 25 Pacific Northwest stores from January 2016 through September 2016. The dataset also included detailed work schedule information on the waitstaff (henceforth referred to as “servers”), which reports their scheduled and actual work shifts. During our observation period, the restaurants informed their servers about their scheduled shifts one week in advance. However, a server’s actual shift could differ from her scheduled shift, as managers may adjust the schedules closer to the day of service to adjust to staffing and customer demand mismatch.

We focused on two commonly used just-in-time scheduling practices in industry: “short-notice scheduling” and “real-time scheduling” (Lambert 2008). Short-notice scheduling refers to the practice of asking a server to work on a new shift beyond her originally scheduled shifts shortly prior to the day of service (mostly two days in our data), while real-time scheduling refers to the practice of asking a server on the day of service to stay longer beyond her current scheduled shift. In addition, for ease of reference, we use “regular schedules” to refer to the original shifts that are posted one week in advance. Among the three types of schedules, real-time schedules were the most unpredictable for the servers as they had no advance notice at all, while regular schedules were the most predictable as they had the longest advance notice. We studied how working on schedules with different predictability levels impacts servers’ productivity, measured by sales per check and meal duration. At the time of our study, there was no law requiring premium pay for just-in-time schedules. Thus, the per-hour pay was the same across regular and just-in-time schedules in our data.

We estimated the causal impact of just-in-time schedules on sales at the check level. Specifically, for any given check, we regressed the check size in dollar amount on the schedule types of the server who handled the check. To make sure the difference in check size was due to the servers’ schedule types and not other factors, we adjusted for several factors: the server’s fixed effect (which controls for a server’s intrinsic sales ability); tenure (defined as the number of days the server worked at the restaurant); the number of tables the focal server handled during the focal check; whether the check was opened during the last hour of the server’s shift; and their fatigue level at the time (defined as the cumulative number of transactions the corresponding server had served that day before the start of the focal check). Besides servers’ characteristics, we also controlled for the party size, timing (hour, day of the week, week of the year), and location of the focal check.

Despite all these control variables, there were still potential unobservable factors that may have biased our estimations. For example, one may expect customers’ characteristics to impact sales. However, customers are assigned to servers in a round-robin fashion in our data and thus customers’ characteristics should have been statistically similar across servers with different schedule types. Thus, while customers’ characteristics are unobservable, they likely did not bias our results. Another potential missing variable is managers’ demand forecast. Managers may have been more likely to use short-notice or real-time schedules when the actual sales were higher than expected. As a result, the standard ordinary least squares (OLS) estimator may have overestimated the true impact of just-in-time scheduling on sales. To address the endogeneity issue due to unobservable confounding factors, such as sales forecasts, we used an instrumental variables approach. The use of instrumental variables is a well-known econometric technique to help address the types of issues we highlight here (Angrist and Pischke 2009).

Theory: the mechanisms through which just-in-time scheduling impact sales at restaurants

There are several potential negative and positive effects of just-in-time scheduling on sales.

Negative Effects. Just-in-time scheduling creates unpredictable schedules for servers that may hurt their sales performance for a number of reasons. First, unpredictable schedules make it difficult for workers to coordinate work with life, particularly in terms of childcare, continuous education, and workers’ overall health (Henly and Lambert 2005, Dicksen et al. 2018). Such work-life conflicts could prevent workers from being fully present and meeting the demands of their work (Henly and Lambert 2014). Second, workers may sense a lack of control due to the schedule unpredictability, which may lead to negative emotions that undermine performance (Zeytinoglu et al. 2004, Takahashi et al. 2011). Lastly, workers may also feel a sense of inequity and injustice at work due to schedule unpredictability (Spector and Fox 2002, Krischer et al. 2010, Avgoustaki and Bessa 2019). Workers with such perceptions may retaliate by engaging in counterproductive behavior such as being less attentive to the customers and making less effort to promote sales (Adams 1963). We refer to the combined effects of the above mechanisms as the “unpredictability effect”.

Positive Effects. Just-in-time scheduling could also have a positive impact on servers’ sales performance. U.S. laws do not guarantee a minimum number of working hours to workers (Lambert 2008). In a recent survey of retail and food service workers in Seattle, 33% of respondents would like to work more hours, and 10% suggested that they could barely pay their bills with their current working hours. Servers may view just-in-time schedules as “gifts” which provide them with opportunities to work more hours and make more money (Akerlof 1982). To this end, they may be more motivated and make more effort during the just-in-time schedules (Spector and Fox 2002, Avgoustaki and Bessa 2019), which could lead to higher sales. We refer to this positive effect as the “income opportunity effect”.

Short-notice schedules and real-time schedules differed in the length of advance notice (unpredictability level) and the length of the schedule (income opportunity level): the short-notice schedule shifts were 6 hours long on average while the real-time schedule shifts were 2.5 hours long. Therefore, the size of the negative and positive effects and thus the net effect could vary by schedule type.


We found that the check size from short-notice schedules was not statistically different than the check size of regular schedules. This may be due to the fact that the short notice of these schedules, even though not ideal, provided some time for servers to adjust their plans. Moreover, the unpredictability effect was balanced by the sizable income opportunity effect due to the long length of these short-notice schedules. In contrast, we found that the check size from real-time schedules was about 4.4% smaller than the check size of regular schedules. The real-time schedules’ lack of advance notice induced a strong unpredictability effect, and the short length of these schedules did not provide an income opportunity large enough to offset the inconvenience associated with their unpredictability. We also found that the smaller check size during real-time scheduled shifts was due to the servers’ reduction of up-selling (servers recommending more expensive items) and cross-selling (servers recommending additional items). Meanwhile, servers’ schedule types did not appear to impact customers’ meal duration.

We further explored the potential heterogeneous impact of just-in-time schedules among servers with different intrinsic sales abilities and across different days of the week. Short-notice schedules actually increased the check size by 4.3% on average for servers with high intrinsic sales ability but reduced the check size by 1.9% for servers with low intrinsic sales ability. Moreover, real-time schedules reduced the size of the checks handled by servers with low intrinsic sales ability two times more than servers with high intrinsic sales ability. We also found that real-time schedules hurt server productivity significantly more during weekends than weekdays.

Besides adding servers by short-notice and real-time schedules, managers can also adjust to real-time demand by canceling existing schedules. We conducted analyses to investigate whether canceling a server’s schedule had any immediate spillover effect on the server’s productivity in the hours before and after the canceled hours and on the productivity of other servers who worked during the canceled hours. We found no such immediate spillover effects. While our study focused on the immediate impacts of just-in-time scheduling on worker productivity, another recent study suggested that scheduling volatility often caused by just-in-time scheduling could lead to higher worker turnover and thus reduce worker productivity in the long term (Bergman et al 2023).

Management and Policy Implications

Our empirical results indicate that firms must strike a delicate balance between the pros and cons of utilizing just-in-time scheduling: While just-in-time scheduling provides scheduling flexibility which reduces labor costs, it may hurt server productivity and thus the firms’ revenues. To inform restaurants on how to use just-in-time scheduling to optimize profitability, we propose an analytical scheduling model that incorporates both the flexibility created through just-in-time scheduling (adding new schedules or canceling the existing ones) and the productivity impact. Given the 4.4% productivity loss during the real-time schedules, managers should consider deploying more regular and short-notice schedules and not wait to use real-time schedules. Such a shift in scheduling pattern not only leads to more predictable work schedules for the servers, but it can also improve the restaurant’s expected profit by up to 1% (any profit improvement can be financially consequential for an industry with an average net profit margin of 3-5%). In short, our study highlights how real-time scheduling may hurt a firm’s bottom line, which provides another reason why firms should shift towards more predictable schedules such as regular and short-notice schedules besides the recently introduced predictable scheduling laws.

Our results also have important implications for the design of predictable scheduling laws. The uncertain nature of the retail and service sectors in terms of customer demand and employee no-shows does call for a certain level of flexibility in scheduling to better match consumer demand with labor supply. Our study suggests that short-notice schedules do not harm the productivity of the workers, and could even increase worker productivity among the most competent workers. This indicates that, despite the added stress of uncertainty from short notice, workers may perceive short-notice scheduling positively overall because of the opportunity to make additional money. To this end, policy makers may consider relaxing the advance notice requirement from two weeks to a few days. In fact, that is what the predictable schedule law in New York City does, by requiring only 72 hours advance notice for retail workers.

As part of the predictable scheduling laws, besides requiring moderate advance notice (i.e., a few days), policy makers may also consider providing incentives for companies to adopt digital scheduling platforms which allow workers to see the schedules, to voluntarily sign up for unfilled schedules, or trade shifts with coworkers, in real time. This will give workers better control of their schedules and adjust schedules quickly if their plans change, while also providing transparency and flexibility the employer needs to run their business effectively. In fact, multiple companies, including Gap and Walmart, have already deployed such scheduling platforms to manage the schedules of their large workforce. A recent experiment at Gap provides some evidence that such scheduling platforms (when executed properly) could be a promising remedy for the controversial just-in-time scheduling practice particularly prevalent in the retail and food service industries (Kesavan et al 2022).

Thanks to the increasing volume of data and sophistication of AI-powered modern scheduling software, the practice of just-in-time scheduling is unlikely to go away anytime soon. Given the severe social consequences of such scheduling practices, it is no surprise that some states and localities have passed laws to limit the practice which, in turn, have received great resistance from the business community. Our study demonstrates that by better understanding the impact of worker schedules on worker behavior and productivity with field data generated by relevant companies, we can design scheduling laws that protect the workers while also benefiting firms’ bottom line.



Adams, J. S. 1963. “Towards an understanding of inequity.” The Journal of Abnormal and Social Psychology 67, no. 5: 422.

Akerlof, G. A. 1982. “Labor contracts as partial gift exchange.” The Quarterly Journal of Economics 97, no. 4: 543-569.

Angrist, J. D., and J. S. Pischke. 2009. Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton University Press.

Avgoustaki, A., and I. Bessa. 2019. “Examining the link between flexible working arrangement bundles and employee work effort.” Human Resource Management 58, no. 4: 431-449.

Bergman, A., G. David, and H. Song. 2023. “‘I Quit’: Schedule Volatility as a Driver of Voluntary Employee Turnover.” Manufacturing & Service Operations Management.

Dickson, A., L. Golden, and R. Bruno. 2018. “Scheduling stability: The landscape of work schedules and potential gains from fairer workweeks in Illinois and Chicago.” Project for Middle Class Renewal, University of Illinois at Urbana-Champaign.

Golden, L. 2015. “Irregular work scheduling and its consequences.” Economic Policy Institute Briefing Paper, no. 394.

Henly, J. R., and S. Lambert. 2005. “Nonstandard work and child-care needs of low-income parents.” In Work, Family, Health, and Well-Being, edited by S. Bianchi, L. Caspar, and R. Kind, 469–488. Lawrence Erlbaum Associates. 

Henly, J. R., and S. J. Lambert. 2014. “Unpredictable work timing in retail jobs: Implications for employee work–life conflict.” ILR Review 67, no. 3: 986-1016.

Kamalahmadi, M., Q. Yu, and Y. P. Zhou. 2021. “Call to duty: Just-in-time scheduling in a restaurant chain.” Management Science 67, no. 11: 6751-6781.

Kesavan, S., S. Lambert, J. Williams, and P. Pendem. (2022). Doing Well by Doing Good: Improving Retail Store Performance with Responsible Scheduling Practices at the Gap, Inc. Management Science68(11), 7818-7836.

Krischer, M. M., L. M. Penney, and E. M. Hunter. 2010. “Can counterproductive work behaviors be productive? CWB as emotion-focused coping.” Journal of Occupational Health Psychology 15, no. 2: 154.

Lambert, S. J. 2008. “Passing the buck: Labor flexibility practices that transfer risk onto hourly workers.” Human Relations 61, no. 9: 1203-1227.

Lambert, S. J., P. J. Fugiel, and J. R. Henly. 2014. “Precarious Work Schedules among Early-Career Employees in the US: A National Snapshot.” University of Chicago.

National Restaurant Association. 2023. “A New Normal Takes Hold; Restaurants Focused on Paths to Continued Growth in 2023.”

Spector, P. E., and S. Fox. 2002. “An emotion-centered model of voluntary work behavior: Some parallels between counterproductive work behavior and organizational citizenship behavior.” Human Resource Management Review 12, no. 2: 269-292.

Takahashi, M., K. Iwasaki, T. Sasaki, T. Kubo, I. Mori, and Y. Otsuka. 2011. “Worktime control-dependent reductions in fatigue, sleep problems, and depression.” Applied Ergonomics 42, no. 2: 244-250.

Terwiesch, C., and G. Cachon. 2006. Matching Supply with Demand: An Introduction to Operations Management. McGraw-Hill Education, New York.

Wolfe, J., J. Jones, and D. Cooper. 2018. Fair workweek ‘laws help more than 1.8 million workers. Economy Policy Institute.

Zeytinoglu, I. U., W. Lillevik, B. Seaton, and J. Moruz. 2004. “Part-time and casual work in retail trade: Stress and other factors affecting the workplace.” Relations Industrielles 59, no. 3: 516-544.