The Incredible Macroeconomic Implications of Uniform Pricing
What does it say about optimal monetary policy?
Large companies often maintain the same prices and wage rates across the entire country, even when local conditions vary. This is both interesting in its own right – it is genuinely strange that firms won’t vary their prices – but it also has extremely important implications for inferring the macroeconomic impacts of regional shocks, as well as implications for measuring trade costs and inequality. Economists can also use these uniform prices as a source of exogenous variation to answer deeper questions.
One of the first papers to exploit uniform pricing as a source of exogenous variation was Capelli and Chauvin’s 1991 test of the efficiency wage hypothesis. To rewind slightly, there are two different theories of involuntary employment in equilibrium. The first is that there are search frictions, and that it takes some time to find the right job. If the cost of finding a job were to decrease, then unemployment would fall. The second is that unemployment is a result of firms having to monitor employees to provide effort on the job. If wages were at the market clearing rate, as under perfect competition, then employees could costlessly find a new job. Unemployment would have no sting, so employers have no way to induce effort. If wages are above the market clearing rate, then workers would be unable to find a job just as good immediately after losing their old job. Thus, involuntary unemployment would remain the same after search frictions totally disappear, and decreases in search frictions would show up as higher wages. (You can see one of my older blog posts for more).
This is obviously very hard to test. How are you to tell what higher wages at a plant are due to? It could be that some plants attract higher productivity workers, such that higher wages are not indicative of higher productivity or effort. Capelli and Chauvin can answer with a dataset from a large automanufacturing corporation which is plainly one of the Big 3. Wages are set by the United Auto Workers, and are the same across all plants, regardless of local labor market conditions. Discipline is initiated by managers, but all dismissals are appealed by the union up to a company wide Appeal Committee. The prediction of the efficiency wage hypothesis is that companies with a higher wage relative to market clearing should have an easier time inducing effort, and that is indeed what they found. The rate of job dismissals was lower when the outside option was worse.
But understanding uniform pricing is useful for more than generating exogenous variation. It also has really big implications for aggregate outcomes. In short, we cannot find aggregate shocks from local shocks. Results like that of Autor-Dorn-Hanson absolutely need not scale up to the country as a whole. The reasoning is fairly simple. Since a regional shock only affects demand to the extent that it affects total demand, agents won’t be able to fully adjust to a purely local shock. Let’s imagine a truly extreme case – there exist an infinite number of regions, and a shock in one region. Because all firms set prices based upon the average demand, and the effect on average demand is 0 owing to it being divided by infinity, we would believe that prices are perfectly inflexible and that monetary policy would be much more stimulative than it actually is.
There are other implications, which are explored by DellaVigna and Gentzkow (2019). They first document that uniform pricing is extensive across US chain stores. Prices within a chain increase by 0.47% for every $10,000 increase in consumer income, while the average price of a chain changes by 4.2% for every $10,000 increase in consumer income. The true change within chain is probably close to zero, and this is entirely due to their data being collected weekly. If prices change within the week, places with more elastic demand sell more units at the lower prices, and they can only observe prices by dividing units sold by revenue. This will mean places with more elastic demand – read, poorer places – having systematically lower prices. When they have yet more detailed data for only one chain store, the true correlation of prices in a chain with local income is a very precisely estimated 0.
There are three big implications. First, prices are too low in rich places relative to the profit-maximizing level. The median chain could raise its annual profits by $16.1 million a year – or 1.3% – by switching to the optimal varied pricing strategy. This naturally increases inequality considerably. Second, we cannot infer national level responses from local responses to shocks. Third, and perhaps most intriguingly to me, this prevents us from estimating trade costs in many settings. The standard way to estimate trade costs is to use how the difference in price varies with the distance from a point of origination to infer the cost of transporting a good over space. Atkin and Donaldson (2015) is the classic example of this in the context of developing nations. If prices are set uniformly by a chain, however, it will tend to reduce the apparent size of trade costs. In the extreme, if prices are set by one firm, then we will apparently have no trade costs whatsoever!
If we’re going to spend all this time on the implications of uniform pricing, it might be worthwhile to comment on why it arises. We have seen how optimally varying prices would appear to gain the median chain $16 million a year in profits – why, then, are companies forgoing this? There are a few plausible explanations, like managers not putting in the time to work out optimal pricing, or marginal costs varying across but I would like to focus on a behavioral one. Consumers consider it an affront and consider it unfair, but only if the currency is the same.
The strongest evidence for this is from Denmark. Denmark has its own currency, but not its own monetary policy. The kronor is pegged exactly to the Euro, and can be exchanged at a fixed ratio. For all intents and purposes, they are the same currency. Thus, if a company sets its prices at the same price in all countries in the Eurozone, it would seem to want to set its price at the same level in Denmark. Yet, as Cavallo, Neiman, and Rigobon (2014) show, prices vary in places outside the Eurozone, even when the currency is essentially the same. It strongly suggests that consumers think in terms of nominal pricing, not real.
The other plausible explanation is managerial inattention. Firms are indeed foregoing profits relative to optimal, but it’s very hard to convince managers to do non-obvious things. The strongest piece of evidence for this is when uniform pricing breaks down — Butters, Sacks, and Seo (2022) show that when there is a clear, well-identified shock to local conditions, firms do indeed change their prices. Specifically, the passage of a statewide tax leads to firms fully passing the shock on in that state, and in that state only. These supply shocks are well-known to everyone, so it makes sense that managers will change prices in response to that, but not due to demand shifters. After all, everyone knows what happens when taxes go up, but even trained economists would have great difficulty detecting demand shifters. This also ties into notions of fairness — changing prices in response to a tax is normal and accepted.
Daruich and Kozlowski have another good paper on the macroeconomic implications of uniform pricing, using data from Argentina. I would first like to take a second to marvel at the extraordinary dataset they have. The Argentinean government, in an effort to push down inflation, began requiring that every retailer provide daily prices on all products in a publicly accessible website. This is a considerable improvement over the Nielsen data which DellaVigna and Gentzkow used, which is weekly. As DellaVigna and Gentzkow note, not being able to observe price changes systematically biases observed prices when stores face different local elasticities of demand. They construct a simple model to illustrate how it can cause a big difference in outcomes, with two cities, which can have either chains or local stores. They estimate that under uniform pricing, the elasticity of consumption in response to a shock is almost a third higher – in other words, people are able to buy a third as a many additional goods when average incomes rise. This is an extremely big change!
Retail prices are set uniformly, but so are wages. Hazell, Patterson, Sarsons, and Taska (2021) show, using extremely detailed data on jobs and wages from Burning Glass, that 35% of firms in the US set the same wages for the same jobs in all locations across the United States. These firms tend to pay higher, rather than lower, wages on average for seemingly identical jobs. They estimate that firms could raise their profits by 4% annually by switching to varied pricing, although they caution that uniform wages may have positive effects on worker morale sufficient to offset this. Voluntary minimum wages, which are conceptually similar, have had limited effects, however. Derenencourt and Weil (2025) show that while it cuts down on separations, there are no spillovers to other firms or jobs. It appears to be a voluntary transfer from employers to employees, in essence, of little macroeconomic relevance.
For all that uniform pricing serves to dampen the flexibility of price, the other way around does matter. On the one hand, prices won’t fully adjust to local shocks – but on the other hand, prices far away will adjust to shocks which didn’t happen to them at all. Ezequiel Garcia Lembergman shows how uniform pricing transmitted shocks during the Great Recession to less affected areas. Places which did better and saw smaller falls in housing prices saw their real wages increase due to the prices of the goods they bought falling.
I have left this to the end, but uniform pricing also says a lot about what menu costs are. (Menu costs are the costs associated with changing a price). There are two basic reasons for menu costs to exist. First, there is a literal cost to changing the price of an item, like changing the labels on items. Then, there are the costs of changing the prices which come from thinking about what the price should be, and trying to calculate the optimal price. There is work on the former – in the 1990s, it was at $105,887 annually per store (0.7% of revenues, 35.2% of net margins, $0.52 per price change) as per Levy, Bergen, Dutta, and Venable (1997) – but it almost impossible to say what the costs of optimizing prices are.
What uniform pricing is evidence for is that most of the cost of price changes is coming from the latter channel, and is invariant to the number of stores in a chain. Imagine a world in which all of the cost of price changes comes from changing labels. If there is a regional shock, then it would be pointlessly expensive to change prices everywhere. If menu costs come from the time spent determining a price, then it would be cheaper to change all prices everywhere. If firms pay a fixed cost to find the optimal price, then the clear implication is that nominal GDP targeting is the optimal policy. See Caratelli and Halperin (2025) for this. Some additional implications are that the costs of inflation should be falling as firms get larger, and if firms do get larger as the economy grows, then economic growth should reduce the harm of inflation. If Michael Peters is to be believed, creative destruction should increase the costs of inflation, and population growth will also increase it.
There is a gun which isn't smoking. If there is a fixed cost to change a price, then larger companies should see more price changes, as the relative cost of changing a price is lower. I have seen no evidence that this is the case. Emi Nakamura (in her slides discussing DellaVigna and Gentzkow) says on slide 17 that there is no evidence that larger stores change their prices more frequently. However, I have not seen this definitely answered positively or negatively, and welcome anyone who knows the area well to comment. In general, research into the exact nature of menu costs is incredibly lacking, despite its enormous importance for government policy. Perhaps the problem is simply intractable.
Hey, as I shared on Twitter, this was a great read. I saved it to hit up in the morning and had a few follow up questions/thoughts:
1. “The prediction of the efficiency wage hypothesis is that companies with a higher wage relative to market clearing should have an easier time inducing effort, and that is indeed what they found. The rate of job dismissals was lower when the outside option was worse.” Isn’t there a confound that the highest compensation jobs will select for the most able, ambitious, and conscientious within a vertical? That leads me to question how much of an effect size can be attributed to the suggested hypothesis.
2. “If prices are set uniformly by a chain, however, it will tend to reduce the apparent size of trade costs. In the extreme, if prices are set by one firm, then we will apparently have no trade costs whatsoever!” Yes, and that sets an upper bound on arbitrage for any aspiring middlemen at ≤0. If you factor in the consideration that the existence of middle men removes price certainty from the mind of the consumer, it’s a small fee to prevent buyer hesitation.
3. “Because all firms set prices based upon the average demand, and the effect on average demand is 0 owing to it being divided by infinity, we would believe that prices are perfectly inflexible and that monetary policy would be much more stimulative than it actually is.. Second, we cannot infer national level responses from local responses to shocks.”
Tempting to play the real analysis game here but I think it’s more useful to focus in on a) considering any national stimulus impacts a ‘region’ of such stores, you have to take the integral over the continuum range defined by the range of stores, yielding a finite, >0 elasticity effect raising prices & b) what you allude to in the closing advantages of whole chains relying on a single price point per good acts like a dampening/insurance effect. I’m mentally thinking of it as a collection of similar assets sharing the same liquidity pool as the spreads between any two of them are fixed. The end result is that vicissitudes, which can hit the store more existentially than consumers (who can shop elsewhere) effectively have their risk of going out of business due to tiny PNL moves vis a vis the correct price hurting them ->0. As the market is infinitely thick, it takes infinitely much to move a price. This would be something macro scale, effecting a region/continuum of nonzero measure like tariffs or taxes or sudden macro demand destruction. This isn’t rigorous but the intuition maps out.
Appreciate your thoughts on any/all of these. Will be deleting this comment after a while because I think I made it obvious what I do, what my background/approach is, and that I’m not formally trained in economics. Genuinely enjoying reading your posts.
"The median chain could raise its annual profits by $16.1 million a year – or 1.3% – by switching to the optimal varied pricing strategy. This naturally increases inequality considerably"
what's the channel by which inequality increases? wouldnt charging higher prices to higher income customers reduce inequality (as a first order effect)?