I have a new working paper with Daniel Hellerstein and Nathaniel Higgins (both with USDA) on the use of quotas in procurement auctions.
This is a new area for us. Our motivation came from thinking about the Conservation Reserve Program and a rapidly emerging literature on "payments for ecosystem services." Basically, the government or environmental interests or the carbon market or whatever wants to buy a lot of something--say carbon sequestration services, water quality benefits, wildlife habitat, etc.--from a large and extremely heterogeneous pool of sellers.
One issue surrounding the heterogeneity of sellers is that we need to put the environmental services provided across varied landscapes, locations, and situations on an equal footing, which requires some method of valuing the environmental benefits. That's a hard thing to do, but it's not new. And I think people are already doing about as good a job as could be expected. Or at least I don't think I've got anything to contribute in this area.
Aside from valuation, it seems to me the biggest challenges everyone has been worrying about essentially come down to price discrimination: The buyers of environmental services want to pay different prices to different sellers according to their opportunity costs for providing those services. I can see a lot of practical reasons for wanting to do this, even if doing so involves a little bit of inefficiency. I wrote about this a bit here.
So, how can one go about price discriminating if the buyer knows costs differ across sellers but they don't know how much they differ, or perhaps even who has high costs and who has low costs? Well, a simple thing to do is to just have a procurement auction and put a modest quota or limit on the share of offers accepted by any observationally similar group of sellers. This causes sellers within low-cost groups to compete with each other much more aggressively. And it causes all sellers to generally compete more aggressively because they realize sellers within low-cost groups are competing more aggressively.
It turns out that solving these kinds of auction theoretically is quite a bit of work. But if costs do in fact vary a lot across groups, quotas can save the buyer a lot of money. If groups are actually quite similar, quotas have no real benefit, but no real cost either.
We also ran some experiments and found somewhat greater savings from quota in the laboratory than in theory, and less of an efficiency loss verses standard pay-as-offered auctions.
I think there are many potential applications besides CRP or PES programs, so the paper is pitched more generally. I also figured out a simple but powerful new technique for solving Bayesian Nash equilibria in asymmetric auctions, but that would only be of interest to a more limited audience.
How much might this kind of auction save the Federal government if they used it for CRP? I don't think that question is strictly answerable given the available data. But I think it's highly plausible that it could eventually save hundreds of millions of dollars per year while simultaneously improving environmental outcomes. Part of this is because the way they currently go about price discriminating looks hugely inefficient (see here). It would be a lot simpler to implement than the current program, too.
Will they do it? I'm not going to hold my breath. But I'm going to shamelessly sell the idea, because I think it would implement exactly what they seem to be trying to achieve in a way that's simpler, would possibly be perceived as fairer, and is almost surely more efficient. If I could get them to do this, and it actually worked, I'd have concrete evidence that I earned my Wheaties. Other applications would just be icing on the cake.