Andrew Boysen

Let me say a bit more about different shoppers seeing different prices on Instacart for the same products. The framing treats this as alarming evidence that "the notion of a single price is breaking down." The article involves at least three distinct concepts that economists analyze differently: randomized pricing experiments, dynamic pricing, and price discrimination. Let me untangle them: nytimes.com/2025/12/09/busin… Randomized Pricing Experiments (A/B Testing) What Instacart says it's doing is running pricing tests. Some shoppers randomly see $3.99 for eggs; others see $4.79. The purpose is to learn how price-sensitive the market is across different product categories. This is standard A/B testing, the same methodology that virtually every online company uses to improve its products. Netflix tests different thumbnail images. Amazon tests button placements. Retailers test prices. The randomization means the prices are not based on any individual characteristics. You get $4.79 for eggs because a random number generator assigned you to that treatment group, not because an algorithm profiled you. Why would you do that? Well, we draw the demand curve for a monopolist and assume they know that curve and pick the optimal price. But you don't know the demand curve so you want to learn how quantity demanded changes with price. Dynamic Pricing The article discusses "dynamic pricing" where prices adjust over time. As I wrote when Wendy's considered dynamic pricing economicforces.xyz/p/should-…, dynamic pricing means adjusting prices over time. Sometimes this reflects genuine cost differences (workers at lunch rush are more expensive per burger produced). Usually we think about this as being based on supply and demand. I mean everything is, but this is based on the curves moving through the day/week/month/year. Uber's surge pricing is the canonical example: when demand spikes after a concert, prices rise to attract more drivers. Uber, in this example, isn't randomizing across time. They are responding to observed supply and demand factors. Other times, it's an attempt to extract more from consumers willing to pay more. The line between "just supply and demand" and price discrimination isn't always clear. I'll say more on that. Instacart said that prices "never change in real time, including in response to supply and demand." So the Instacart case isn't dynamic pricing in the Uber sense. Price Discrimination Price discrimination means charging different prices for goods that are identical from the seller's perspective, based on consumers' willingness to pay. Senior discounts, student pricing, and coupons are all forms of price discrimination. This is what everyone seems to worry about. It's also not what the study found. But that's not really the narrative of the piece, let alone the Groundwork piece or More Perfect Union. It's all worry about price discrimination. "But there is little doubt that Instacart and other online sellers have the ability to do so." It then notes that "Companies including Delta Air Lines, Amazon and Home Depot have been accused of experimenting with such personalized pricing." A random person speculates "We're going to tailor the price to you".. "We're going to get the maximum amount of money out of you that you're prepared to pay and drain your pocketbook." And the article cautions that frequent price changes "could make it easier for companies to adopt personalized pricing strategies in the future." Then there are a handful of other things topics touched on. There's a paper the Times cites about "algorithmic coercion" is about something specific: when all competitors use algorithms that can instantly match price cuts, firms have less incentive to cut prices in the first place. Algorithms are worth studying. But it's about competitors coordinating on high prices, not about firms extracting personalized prices from individuals. Why Price Discrimination Isn't Always Bad Even if we were talking about price discrimination, and there's no evidence we're not talking about A/B testing, the article's framing assumes any variation in prices harms consumers. This is an empirical question but clearly not always wrong. @_amackay point out the key trade off "many pricing strategies resulted in higher costs for less price-sensitive — and therefore potentially higher-income — consumers, while lower-income consumers paid less." Price discrimination is discriminatory, usually progressive, but never strictly worse for everyone. In my research and writing on price discrimination economicforces.xyz/p/does-pr… briancalbrecht.github.io/alb… I've found that with competition, price discrimination can even maximize consumer surplus. That's a particular model but at least should give some pause. Empirically, its a big mixed bag but papers find real benefits from price discrimination. On the theory side, even without competition, I think there's a good case for wanting more price discrimination economicforces.xyz/p/monopol… But there are qualifications, like the algorithmic coercion paper above. The type of pricing matters for thinking about counterfactuals. Those pesky counterfactuals. The Groundwork piece found the price variation could add up to $1,200 over a year of grocery shopping. That calculation compares a world with randomized testing to a world where everyone gets the lowest tested price. To be kind, that's not a plausible counterfactual. Without A/B testing, Instacart doesn't charge the lowest price in the range, it charges some uniform price, probably near the middle. Under randomization, a chunk of consumers pay more than they would under uniform pricing, a chunk pay less. One question is whether the average price would rise or fall. That's an empirical question but clearly $1,200 isn't plausible. Personalized price discrimination is different. If you ignore the people who gain from price discrimination because the price was dropped for them (seniors who don't go when they don't get the discount), there's at least a story about the algorithm systematically extracts more from people willing to pay more. That's a coherent concern about how the pricing system is just driving up prices. But it's not what the study found!