Professor Katja Seim

The interaction of markets and policy shapes the development of technology, the functioning of firms, and our experiences as consumers. Katja Seim, the Sharon Oster Professor of Economics and Management, investigates the complex forces behind the infrastructure of everyday life with an aim of enabling data-driven improvement to policy tools and making markets function better.

Q: What questions have guided your career?

My work examines the interplay of strategic firm behavior and public policy. Early on that took the form of questions about antitrust, including the overarching issues of market power and pricing as well as narrower topics such as how firms identify immediate competitors and how strategy may shift after a merger.

In 2016, I served a term as the chief economist at the Federal Communications Commission. Since then, many of my projects have come out of issues I encountered there. The research still intersects with firm strategy questions, but it has a different flavor because I’m looking at them from a regulator’s perspective. For example, how can policy induce firms to deliver internet build-out to underserved areas given the high fixed costs and limited number of competitors?

Q: What kind of economist are you? Would you offer a taxonomy of where your work fits in the field?

I’m an industrial organization economist. A key focus of industrial organization economists is the role of market power in market outcomes. That might mean studying efficiency—if we are not producing the right products or the right amount of products, is that because of firms’ market power?

Increasingly, another area of interest for industrial organization economists has been examining equity considerations. If firms are prioritizing the most profitable consumer, they may exclude a wide swath of the population. Which types of consumers are most adversely affected?

With the growth of digital platforms, there’s also been a lot of interest in trying to understand better how they function. Should we be worried about the amount of information that these companies collect? Do consumers understand what they are agreeing to when they engage with the platforms? Is it a good thing or a bad thing that Amazon is so big?

In terms of my approach, I’m entirely an empirical economist. I typically use existing data sets, comparing what we can see in the data with what economic models tell us we might expect to see. That’s a way to assess how well markets function and what we might do to get them to function better.

Q: You mentioned Amazon. It gets a tremendous amount of attention in the press and from the public, yet you’ve found an unusual angle with your work on the company.

Amazon does, rightly, get a lot of attention. Some of what they do might be good for consumers, some of it might be bad. Amazon accounted for about 25% of e-commerce revenue as of 2020. Their growth and dominance of e-commerce have led to concerns about a “winner-take-all” trajectory and the potential for anticompetitive behavior.

My co-authors and I decided to focus on the expansion of Amazon’s logistics capabilities. After having relied on third-party shipping companies for a long time, Amazon grew its own logistics capacities remarkably quickly, to the point where they are now the third-largest shipping company in the world. They handled 20% of package deliveries in the country and offered same-day delivery to approximately 50% of the U.S. population by 2020.

To show the scale of the growth, in 2010 they had 17 fulfillment centers—warehouses where orders are pulled from shelves and put into boxes. They had no specialized locations for sorting packages into delivery zones. And they had just 1 last-mile center for loading packages onto Amazon’s own delivery vehicles. By 2020 they had 234 fulfillment centers, 67 sortation centers, and 388 last-mile centers.

Q: What was the impact of Amazon growing its logistics capacity?

Early on, expanding the logistics network meant Amazon had a physical presence in more states so they had to collect sales tax in those states. That actually reduced demand. But over time, we found sizeable gains for the company and for consumers. Shorter shipping distances lowered cost, increased reliability, and probably also reduced environmental impacts from emissions.

However, we have ongoing work that seems to suggest that when Amazon introduces same-day delivery, there’s a sizable decline in local retail store traffic. It remains to be seen what that means for brick-and-mortar retailers and for cities more broadly. But you can see how that has implications for the amenities of city life.

Q: Amazon is one omnipresent part of life today that you’ve studied. Smartphones are another. Even a decade ago they weren’t particularly reliable as internet devices. Today it’s typically a different story. One of the reasons that came about is because the FCC reallocated spectrum controlled by broadcast television to wireless carriers. Would you explain your work on that?

Reallocating spectrum was a really complicated undertaking. There were technical engineering challenges to solve in order to make more efficient use of the spectrum while allowing broadcasters and wireless carriers to share the airwaves.

There was also an expectation that the proceeds of selling spectrum to wireless carriers would cover what the FCC had to pay out to reacquire the spectrum from television broadcasters. So the FCC spent a lot of time devising a really sophisticated mechanism to make it all work. They decided to use an auction because it’s an effective way to discover what price companies would accept to sell back the spectrum they controlled.

Complicating things, the auction was two sided. One side was to purchase back spectrum that broadcast television stations held. The other side allocated the re-purchased spectrum to the wireless carriers that bid for it. And bids were placed by both sides simultaneously during many rounds of bidding.

Q: This auction truly was groundbreaking. Paul Milgrom and Robert Wilson won a Nobel Prize for designing the simultaneous, multiple-round auction model that was used.

My time at the FCC coincided with the reallocation auction being finalized. It was exciting to be there for that. The FCC reallocation of spectrum is one of the most amazing and sophisticated uses of auction theory ever devised. And it was very successful.

Q: Yet you did work that highlighted shortcomings in the design.

My colleague Michael Sinkinson, our co-author Ulrich Doraszelski, and I did a project on the auction design. The objective wasn’t to try to point out a weakness of the auction; it was to test for realistic behaviors that we might expect and how serious their impact might be in practice.

We got interested because before the auction some private equity companies started buying up television stations. It looked like rather than just trying to flip the stations for profit, they were holding them strategically.

The design of the auction assumed that all of the broadcast television stations would simply bid in—ask for—what they wanted to get paid in the auction. We wanted to know whether there was any way companies that owned multiple licenses would be able to manipulate the outcome by withholding some of their stations from the auction.

Through our modeling, we found that the ownership structure was quite susceptible to manipulation by even just a few players. Depending on the amount of auctioned spectrum, it could raise the cost to the government by between 7% and 21%. Since in the actual auction the government paid $10.1 billion to reacquire the broadcast spectrum, these are significant amounts. We think future auctions can adjust for this.

That said, wireless carriers paid $19.6 billion for the reacquired spectrum. After subtracting costs, the Treasury received $7.3 billion, and we all benefit from the improved use of the spectrum.

Q: You’ve also done work on another use of auctions by the FCC.

Yes. I alluded to it earlier. Many of us take broadband for granted, but the pandemic made clear that many rural areas and lower-income urban areas still need better access and that where broadband is lacking, people are held back. That leaves the question, if the government wants to encourage the build-out, what’s the best way to do it?

The fixed costs are too high for companies to want to build more broadband infrastructure in these areas on their own, so the FCC is willing to subsidize the work. But since there are a limited number of companies in a position to do the work, it’s difficult to get them to be truthful about their actual costs. So the FCC runs an auction as the means to allocate the subsidy dollars.

Our initial analysis showed auctions are quite successful at promoting build out. We’re still doing analysis to understand whether it’s helped people to engage more effectively or whether it’s the most efficient way of allocating the subsidies. To put it slightly differently, there’s a return from the auction, we’re still trying to understand if it’s the best possible return.

These are the kind of complex questions I really enjoy working on. When we come at it from a measurement perspective, the data tell us how well we’re doing putting theoretical insights into practice. I’m trying to understand the facts and then thinking about whether we could do better given those facts.

Q: What impact do you want your work to have?

Rather than any active and directed efforts at shaping things, the impact I’d like to have is as a source of unbiased evaluation. I try to use the best available methods to deliver accurate evidence on questions that matter to policymaking and regulation.

I think oftentimes people have an adverse reaction to the word regulation because it sounds like Big Brother government. That’s not at all what I have in mind. To me, regulation is just the process for identifying a problem with a market. Sometimes, that can be addressed by policy, which can take many forms, but in my view should attempt to do nothing more than necessary to make the market function better.

Ideally policies should be nimble enough to adjust to changes in the environment. For example, 40 years ago, we put rules in place to make sure cable TV was made broadly available. Today, given the ubiquity of options offered by online streaming, we don’t need to limit the prices cable companies can charge. In this instance, regulation can mean a rollback of rules.

Q: Why did you step away from your academic work to take the role at the FCC?

A lot of my research involves ways to improve how policies work. But it does it from a very high level. When the opportunity for a one-year assignment at the FCC came around, I thought it would be a great way to understand how rules and regulations are written: to see who is involved; to see the trade-offs that go into a particular regulation; to experience the process from the initial idea to a draft to actual implemented policy.

That nitty gritty of policymaking is hard to conceptualize or research as an academic. Seeing it firsthand was invaluable. As I said, it has led to a number of projects that approach questions from the perspective of the regulator.

I ended up being there at an unusual time because it was in the transition from the Obama administration to the Trump administration. That made it unexpectedly interesting in other dimensions because I saw lots of regulations come and go because of this change in government. And I saw how people in an agency position themselves under a new administration. Seeing how a transition comes together was fascinating.

Q: Can you point to specific ways your perspective shifted?

Seeing what is involved in moving from idea to policy made me much more modest in what I think economic research can achieve on its own. And it made me much more cognizant of what an economist brings to the table, both the strengths and weaknesses.

As an academic economist, your real strength is the ability to take a step back and say, “The conditions you are describing look quite similar to the circumstances in this other situation and here’s how that worked, but here are other ways it could have turned out given these different parameters.” That ability to offer context and a range of potential outcomes is quite useful.

But for policymaking, you also need a good dose of pragmatism and willingness to take a stand and say, “We are going do x, even if y might indeed happen.” I think academic economists can struggle to continue moving forward as input and trade-offs keep shifting the specifics. It gave me an appreciation of the work required to turn research insights into policy.

Q: How could we make policy better overall?

Again, I’m a very quantitative person. I’d like policymaking and regulation to continue to be more data driven. It’s understandable that it isn’t. When policy is being designed and implemented, there are many competing demands and often a timeline that must be adhered to. Nevertheless, I think we can improve the types of data collections that regulators are able to engage in and then actually use them to come to more factual decisions.