The Sentinel #15: Dynamic Pricing and Social Norms
Transparency, Strangers and Ratings.
New subscribers: Welcome to The Sentinel.
This is a newsletter in which we think about the models of the world. This newsletter won’t be programmatic — some posts will be personal and experiential; some data-oriented and dispassionate. Sometimes, the textual wanderings are loosely designed ruminations and provocations. Over multiple posts, they will weave a theme — from threads of eclectic interests — a mosaic that reflects our complex world.
Read these posts at your leisure and help me grapple with the depth of our problems with our personal illuminations.
Early Subscribers,
Thank you for the permission you offer to share these thoughts in your inbox. The newsletter is “free”, but I value your time. I find it takes time to read and reflect on these slow, non-reactionary thoughts.
If you like this newsletter, please share it with someone who would also enjoy reading these notes. Just click below.
Some days it seems that no one likes dynamic pricing.
Recently, in March 2023, the NY State AG office proposed a series of rules ostensibly “to protect consumers and small businesses”.
New York’s price gouging law bans companies throughout the supply chain from taking advantage of the market disruption to increase their profits for vital and necessary goods and services.
The proposed New York rules establish a formula for determining the baseline price of goods priced by dynamic pricing algorithms: the median price for the same good at the same time of day in the same area across the week before the market disruption.
It applies to changes in pricing during abnormal market disruptions “caused by extreme weather events, military action, energy disruptions, strikes, or national or local emergencies, or another event that leads to a declared state of emergency.” or states of emergency related to such events as declared by the governor.
—
Are they correct? Dynamic pricing is often misunderstood, and the effectiveness of such policies highly depends on the norms of society.
Life is dynamic. We are constantly adjusting our actions in reaction to the vagaries of the world. Our actions (hopefully, not our values) are etched in shifting sands. Even the rulings and proposals are dynamic! So, what should we allow or disallow? It is going to be super hard to talk about all of the complex stakeholders in one post.
So I begin with, the simplest real-world example of the reasonable success of how dynamic pricing was implemented in public goods provision.
—
A recent publication by Feldman, Li, and Tsai (2022)1, “Welfare Implications of Congestion Pricing: Evidence from SF Park” ($ link, unfortunately) looks at the pilot dynamic pricing program implemented by the San Francisco Municipal Transportation Agency (SFMTA) at 6000 parking spots between the years 2011-13. Pnina Feldman at Boston University and Jun Li at the University of Michigan are alumni of the Wharton doctoral program.
1. The success of Dynamic Pricing depends on Norms and Operational Transparency.
Based on much of the research in Operations, I argue that the success of dynamic pricing implementation hinges on concerns about fairness and equity as much as on generating economic value. Two factors that influence the success of a dynamic pricing policy are evolving societal norms and effective communication of the policy.
As a famous meme goes, in the nineties, it was not ok to get in cars with strangers. In 2022, we not only summon strangers through an app, with the trustworthiness of ratings, but we also care about how much that stranger rates us.
So, society has to be accepting of the policy and the policy implementation has to have operational transparency. These are the two crutches that a good Revenue Management policy depends on. Ride-share companies such as Uber and Lyft, have changed the societal norms on whether it is safe and ok to get into a stranger’s car (with their permission), but also what’s acceptable in surge or variable pricing.
But, how to be operationally transparent?2 To be fully transparent is challenging in both conceptualization and execution. How to signal that a 3X surge price is fair? Sometimes, the argument is straightforward. For example, everyone expects to have a hard time parking on Friday night. What about a rainy Wednesday night? What about dropoffs near a hospital? It is in such cases, operational transparency helps with restoring customer goodwill.
—
2. Dynamic Pricing may not always improve Utilization
I would estimate that city parking spaces during peak hours are already highly utilized on average (compared to large sprawls in suburbs). For a highly utilized asset, dynamic pricing is effectively another method of rationing the asset. Dynamic pricing of meters may only shift customers who use the parking spots (while generating revenues, of course!). However, dynamic pricing doesn’t solve the congestion issue, per se.
To visualize this issue, think of a doctor’s office which is highly utilized. She sees a large panel of patients with the same illness (for simplicity and comparison to congestion policies). Patients have long waits since the doctor just doesn’t have enough hours to see all patients.
The equity argument would suggest seeing as many patients in the time available. This is still rationing because many won’t be served. So how should we decide who gets the service? Perhaps first come first served. Researchers call this policy “FCFS discipline” — an abbreviation of First Come, First Served discipline — in Queuing Theory. A policy that implements a discipline like FCFS still rations.
Dynamic pricing argument would imply that the physician is able to see patients with a higher willingness to pay (WTP), maybe seeing each of them over a longer interaction time. Such a policy means that the doctor sees fewer people for longer visits, and/or sees people who paid earlier. (Some of us are aghast at the idea of pricing services and how unfair it is. We should note that’s what is nominally the effect of using insurance policies for screening patients).
The first-come-first-served process seems like a “fair” policy. But, such a policy has a lot of inefficiencies. If patients have different ailments, the FCFS policy will be very poor. In fact, even if the tasks are comparable, there can still be fairness issues with FCFS. For example, would it be acceptable if early job applicants have a better chance of getting a job than applicants who came in later (assuming everyone applied before the application deadline)?
In both cases, the doctor’s utilization could remain nearly the same, it is only the composition of the patient population that has changed.
The harder question that is missed in the arguments about dynamic pricing is “Which composition is socially acceptable?”.
—
3. Dynamic Pricing has to be “simple” to improve Consumer Welfare.
First, the paper shows, as I argued above, that the primary concern is not to redistribute the composition of demand, but a better utilization. This may not occur in areas with moderate utilization if drivers drive out of (or don't drive to) those regions seeking better prices.
Then, there is a cost of complexity (related to the transparency argument). If the pricing structure gets very complex, congestion pricing may actually lead to more search traffic, not less. Therefore, a simpler policy may achieve higher welfare than a complex one.
Why?
Feldman and others find that an increased level of price dispersion causes more people to search for spots with better pricing. This search occurs when there are empty spots, that have high prices. In fact, rationally speaking, drivers are more likely to think more spots may be available if they see empty spots. Such searches for lower pricing spots might even exceed the congestion created due to the search for better spots under static pricing.
This is a cool paper that demonstrates the complex effects of dynamic pricing efforts on the consumer.
Back to NY State, I am sure that the state has some interest in managing and restricting unfair business practices, but it is unclear whether a blanket regulatory ban on dynamic pricing would be effective. This is an issue of operational complexity and deserves a careful analysis of goods, supply chains, and social norms in play when it comes to the product.
It’s New Year’s Day in my hometown. So, to those readers who are celebrating - Happy New Year! Happy Vishu! Puthandu Vazhthukkal!
Pnina Feldman, Jun Li, Hsin-Tien Tsai (2021) Welfare Implications of Congestion Pricing: Evidence from SFpark. Manufacturing & Service Operations Management 24(2):1091-1109.
Operational Transparency is an important area in Operations research literature. In future posts, I will talk about some transparency research, including that of Ryan Buell at Harvard.
Regarding surge pricing in Uber like operations, we tried out an open ended exercise in class. We defined lock-in prices as options to be purchased by both passengers and drivers (guaranteed upper bound for passengers and guaranteed lower bounds for drivers, with a small non-refundable premium) and tried to think through whether this gives more information on the local supply demand gap that could occur and also the willingness to pay/willingness to supply at some price through these options. In principle, this should help in setting appropriate prices and also advance information for increasing supply and adjusting demand, wherever possible. Could not think through all of it, but it was interesting, and one more way of being 'transparent' about surge pricing.
Loved the article. I just finished teaching the revenue management module in my Healthcare and Service Operations class, so I shared this with my class.
On the social norms front, I always find it interesting that they vary so much by culture. Separate pricing for lunch menus was such a weird thing to me when I first came to the US.