In recent years, the emergence of platforms has surely changed our world. Platforms have also aggressively sold the hiring of on-demand labor as the notion of “flexibility”: Flexibility not only as a benefit for firms but also as a benefit for workers. Here are a couple of ads from Uber and TaskRabbit. Task Rabbit calls the on-demand work as “work. life. balanced.” The idea has also spread to “traditional” non-platform service providers such as Amazon, which has begun the Amazon Flex program. I previously wrote about Amazon’s Last-mile delivery complications with Amazon Flex (Go it Alone on the Last Mile?) on the blog.
A recent New York Times article, by Noam Scheiber, focuses on how steady hours can help both workers and the profits, based on a research study at the apparel retail firm GAP.
Here is the link to what I see as a very impactful paper “Stable Scheduling increases Productivity and Sales” by Professors Williams, Lambert, Kesavan, and others. (Saravanan Kesavan is an Operations colleague at UNC who has been examining such scheduling issues in retail, through empirical analysis).
While brick and mortar retail is facing the biggest challenge in years, retail stores as a channel are not going away. The key to running a successful brick and mortar store lies in executing a careful trade-off between high efficiency and high customer experience. On-demand and flexible scheduling increase efficiency (by using labor only when they are needed), but decreases customer satisfaction (by exacerbating poor worker retention, affecting motivation, etc.). This finding is a nice confirmation of the story of the importance of stable scheduling of labor.
The structure of work and labor has been a pre-occupation since the time of Adam Smith. Through work-flow empirics, we have been constantly looking at how to improve efficiency by cutting non-value adding activities. Improving efficiency is more complicated than we have thought.
—
Findings of the Paper: Overall, this paper nicely confirms the importance of stable schedules in retail. The paper is not about “fixed schedule” vs. “on-demand scheduling”. Gap introduced two actions in all their stores: Freezing schedules two weeks in advance, and eliminating “on-demand” calls (i.e., tentative schedules that could be canceled hours before).
Then, during the 10-month long pilot study in 28 stores, GAP introduced the following 5 scheduling rules in 19 stores (the rest of the stores were used as controls): 1. Workers could swap schedules. 2. Managers scheduled consistent schedules, 3. Managers aimed to improve consistency every week, 4. Minimum (guaranteed) work hours for the core team (Part-time Plus) and 5. Targeted temp hiring during anticipated peaks.
The authors record many improvements (consistency and predictability of schedules, etc.). But, two things are of note: Worker input improved, and the overall sales improved, with good ROI.
—
Here is the main point: 100% Efficiency is physically unattainable, but is also undesirable.
1. The unattainability of Efficiency: Demand is Volatile and Supply is Sticky.
Retail stores are living through an era of instability. However, even in good times, the demand always varied depending on product styles, prices, location, and a variety of other features.
If the goal was to run the store with maximal labor efficiency possible, then the only way is to hire temporary labor to the minute, and/or pay for labor only when they work. Of course, such temporary workers would have minimal to no training. All retail jobs require some level of decision-making and judgment, which can only be learned through consistent time input.
Secondly, the supply of labor for such temporary retail jobs is very low. Platforms like Uber, TaskRabbit work with flex labor because of transportable underutilized assets. Agents drive around with assets (cars) that are low on utilization, get the job done (jobs typically involve some degree of transportation). Unlike such jobs, retail job labor supply has a participation constraint (fixed costs, such cost of driving to the store and back, and working out of the car) that can only be made up with few consecutive hours of work.
Hence labor needs to find some guaranteed hours of work at the retail store (that can make up for fixed costs). In principle, this is the motivating factor behind some support for efforts such as UBI (Universal Basic income). So, the best labor mix in retail is going to be some “fixed” proportion of regular labor and some temporary labor for spikes in demand.
2. The undesirability of Efficiency: Science of Sleep and the Nature of Work
Having a fixed amount of labor with guaranteed hours and income seems to be inefficient, especially when the demand is low. One often-suggested solution is to improve the utilization of fixed, regular labor. I argue that such a drive for efficiency is also undesirable because it ignores the structure of work.
The best analogy to think about the structure of work is the nature of sleep. We spend between 25% to 35% of our lives “switching off” of our bodies and brain cells (partially) and going into a state which does not directly aid our overall efficiency. More sleeping hours hurt efficiency (of course). But, as mounting research into sleep science has demonstrated that been showing that sleeping less doesn’t help either. Lack of sleep increases lethargy, errors, and decreases motivation.
Using breaks like sleep, bodies need to recharge. This is also our bane of existence that exposes us to the threat of automation.
Retail workers need breaks (scheduled, unscheduled) and low impact rests. As the authors found out in the study, such an effort to guarantee work hours, improved their input into the store operations, and also enhanced store revenues.
The article cites my department colleague, and Wharton Retail Operations expert Marshall Fisher, who says that the study provides compelling evidence that,
if you treat people decently, you get better results,
I cannot agree more with that assessment.