Although the 2020 Democratic primary race is barely more than a month old, Sen. Elizabeth Warren (D-Mass.) has already worked to define her role: She is running as a policy wonk.
Warren laid the groundwork for her campaign with detailed proposals to tax wealthy individuals and impose significant new regulations on large corporations. Yesterday, CNN published a profile of Warren describing her as a "wonky professor." Warren herself has embraced this perception, saying last month that her strategy to win over Democratic primary voters would be to "nerd out." As The New York Times reported last month, "Ms. Warren's passion for policy minutiae has become her way of standing out in an increasingly crowded Democratic field." At the moment, this is the central message of her campaign: She cares about the details, and she will get them right.
Warren's penchant for wonkery, however, has been vastly overstated. Although she is probably more familiar with the mechanics of economic policy that many of her 2020 rivals, she is also prone to relying on dubious, and arguably dishonest, methodology in order to support the progressive policies she favors.
Just yesterday, for example, Warren released a proposal calling for a vast new program to federally fund child care. The program would make childcare free for families earning up to about $50,000 a year and would subsidize care for families earning more. The program would be expensive; Warren puts the cost at about $700 billion over the course of a decade. She says would pay for the program using revenues from her wealth tax, pointing to estimates from UC Berkeley economists that the tax would raise $2.75 trillion over the same time—more than enough to offset the cost of the program.
There are reasons to suspect that a wealth tax wouldn't raise nearly as much revenue as Warren estimates, which is why a majority of the OECD countries that have tried such taxes have done away with them.
But there is a much bigger problem with Warren's figures: As The Washington Examiner's Philip Klein wrote yesterday, she's using estimates that rely on two different sets of opposing assumptions. Warren's cost estimate relies on dynamic scoring, which builds growth effects into the model. Basically, it assumes that widely subsidizing child care would boost the economy by allowing more people to work, either in an enlarged child care sector or in other types of jobs. That assumption is built into the cost estimate, giving it credit for any economic benefits it might provide.
The revenue estimate for Warren's wealth tax, however, relies on what's known as a static analysis—it counts no growth effects into its assumptions, presumably because a wealth tax, by taxing the sort of people who are likely to make large, economy-building investments, would have a negative impact on growth. In other words, it ignores any negative impacts. As Klein writes, "Warren is relying on two different methods of analysis, one of which makes her spending seem less costly, and one of which makes her tax plan seem like it would raise more money." Either this is a deliberate attempt to mislead, or it is an oversight that just so happens to be extremely convenient.
These are admittedly wonky details. But that's what Warren says she's focused on, and she is not getting them right. Nor is this the first time that she has launched major policy initiatives based on dubious evidence.
More than a decade ago, Warren, then a Harvard professor, co-authored a scholarly paper finding that more than 40 percent of bankruptcies were the result of medical bills. This research garnered reams of media coverage and made a substantial impact on public policy debates. The paper contributed to Warren's reputation as an economic policy superstar, and, eventually, an influential U.S. senator with a national profile and ambitions to higher office. It is not a stretch to say that this research helped put her on the path that eventually led to her current presidential campaign.
That research, however, was based on dubious methodology that plausibly appears to have been designed to research a specific conclusion. It relied on survey data from about 1,800 Americans who had gone bankrupt, and then counted how many of them had medical debt over $1,000, claimed medical bills had caused the bankruptcy, missed weeks of work due to illness, or mortgaged a home as a result of medical bills.
The problem was that the survey relied on a limited selection of people who had already gone bankrupt rather than the public as a whole. The research wasn't really measuring what most people who read about it probably believed it was measuring; namely, how many people were driven into bankruptcy by medical bills. Instead, it measured how many already bankrupt people had a moderate amount of medical debt, or said they had experienced some sort of medical shock to their household. It was an unusual choice of metric that ended up vastly inflating the public perception of the role medical bills play in causing bankruptcy.
The research and its methods were widely criticized at the time by both academics and journalists (my wife, Megan McArdle, was among the critics). As health economist Craig Garthwaite told The New York Times last year in a look back at the bankruptcy research, "There are no reputable economists who I deal with who believe the number in the paper or the methods in the paper are appropriate in trying to get at the true underlying question."
The team behind the research adamantly defended the work, and even published a follow up raising the percentage of medical bankruptcies to 62 percent, leading to headlines declaring that medical bills caused a majority of bankruptcies. Warren herself tended to stay out of the fray, but continued to work with the same co-authors for years, implicitly standing by the work.
Last year, however, a different team of researchers attempted to answer the same question by examining a much larger trove of data based on credit reports of people who had been hospitalized in the state of California. The study, titled "Myth and Measurement: The Case of Medical Bankruptcies," criticized previous research for "[assuming] that whenever a person who reports having substantial medical bills experiences a bankruptcy, the bankruptcy was caused by the medical debt." It concluded that large medical expenses only caused about 4 percent of bankruptcies, a result that might represent a slight undercount but is almost certainly closer to reality than what Warren and her co-authors concluded.
Warren's political fame and policy wonk persona are thus founded in large part on poorly designed research that led to a wrong but attention-generating conclusion. Although she is more conversant in the language of economic policy than most of her fellow legislators, she is best understood not as a straight-shooting academic but as an advocate who cherry picks convenient data and conclusions to support a predetermined narrative. We have a word for someone like that, but it's not "wonk." It's "politician."