Search the Site

Episode Transcript

Hey there, it’s Stephen Dubner. Before we get to today’s episode, I want to ask for your help, for a special series we’re just starting to make. It’s about mentorship. And this is where you come in: we’re looking for some good stories. It could be about a mentorship in business, or academia, or in sports; it could be a spiritual mentor, or someone who helped you become a better parent, or spouse. Or maybe you’re the mentor. Or: maybe you have a mentor who doesn’t even know they’re your mentor. No relationship is too small, or too weird, if it matters to you. Send us an email with some of the particulars; we’re at We look forward to reading your stories, and interviewing some of you for this series. Thanks in advance. And now: today’s episode.

*      *      *

Here’s a phrase you have probably heard before: “The road to hell is paved with good intentions.” The sentiment goes back at least to the Bible; but the way it’s used today likely began with the 18th century English writer Samuel Johnson. Since then, versions of the phrase have appeared in the works of Charlotte Brontë and Lord Byron; Soren Kierkegaard and Karl Marx; Ozzy Osbourne and Madonna.

MADONNA: The road to hell is paved with good intentions, yeah.

“Yeah” … but how would an economist think about it?

Josh ANGRIST: I would say economics is fundamentally about trade-offs. And there are always trade-offs.

Today, on Freakonomics Radio, three stories about good intentions gone bad in the workplace.  

ANGRIST: If anything, it made them worse off by reducing their employment rates.

Marina GERTSBERG: I find that junior female academics start fewer new projects.

But: can economists help turn good intentions into good outcomes?

Zoe CULLEN: I can’t emphasize enough that with slight adjustments you can get your cake and eat it too. 

Let’s eat some cake. 

*      *      *

DUBNER: When you look back, Josh, at the body of research that you’ve done, some of which brought you a Nobel Prize — congratulations —  how do you assess its importance or leverage in terms of influencing public policy? And what I really want to know is, is that a goal? I suspect you’ll say not. But maybe if you’re being 110 percent honest, you might say, “Well, a little bit, sometimes.”

ANGRIST: Well, I like to influence public policy. And I’m happy when I influence public policy, but that is not what I get up in the morning and set out to do. I’m an academic, and what I set out to do is high-quality scholarship. I like to get things published in top journals. That’s how I measure my influence. Now, ultimately, a lot of the work I do does affect public policy, or at least it becomes part of the discussion, and that’s gratifying. 

This is Josh Angrist.

ANGRIST: And I’m a professor in the Department of Economics at M.I.T.

He did win a Nobel, along with David Card and Guido Imbens, for their, quote, “methodological contributions to the analysis of causal relationships.” That’s a fancy way of saying that these three economists have found reliable ways to measure if a given factor (let’s call it x) is the actual cause of a given outcome (let’s call it y). Or does something other than x cause y? Or does x perhaps cause z, which may be the opposite of y? If you are a policymaker hoping that x policy will cause y but it causes z instead — well, good intentions, but not-so-good outcome. This gap, between policy intention and policy outcome, is something that Josh Angrist is particularly interested in. Consider U.S. tax policy:

ANGRIST: We don’t tax the rich at 100 percent. But sometimes we’ve been taxing poor people at 100 percent, or even more, because there’s what’s called a cliff, where when you cross a threshold, you lose an entire benefit. Well, the incentives there are very poor for employment. You don’t want to change jobs and you don’t want to move. So all of a sudden, there’s a lot of trade-offs.  

Here’s the thing: making good social policy is hard, for a lot of reasons. First, it’s difficult to know for sure what works, whether x will actually cause y. Second, policymaking is part of politics, and politics is messy, with all kinds of compromises to be made along the way. But perhaps the trickiest thing is that the people who are targeted by a given policy may react in a way the policymakers had not anticipated — in a way that may cause the policy to, essentially, backfire, at least to some degree. This has come to be known as “the law of unintended consequences.” It’s not really a law but it is a powerful and usually unwelcome force. Josh Angrist has been thinking about unintended consequences for pretty much his entire career. His first big research finding, along with the economist Daron Acemoglu, had to do with the A.D.A., or the Americans with Disabilities Act. It was signed into law in 1990 by President George Bush, the first.

ANGRIST: The Americans with Disabilities Act expanded civil rights protection, meaning you could not fire or refuse to hire or pay less on the basis of a disability. 

DUBNER: And what disabilities were included in the first version of the A.D.A.? 

ANGRIST: Well, it’s not concrete That was one of the things that had to get figured out. And there was a lot of litigation about what could be counted. It’s tricky because maybe the employer doesn’t know you’re disabled. So there’s always a lot of litigation in the U.S. When a new policy comes in, the courts kind of decide. And ultimately the courts gave a fairly broad interpretation. So it covers a wide range of physical disabilities, including some that might not be obvious to an employer, like back pain. The other thing is — and this was relatively novel — the A.D.A. requires employers to accommodate disabled workers. It’s not clear what that means. The law says it has to be reasonable. So, for example, if you’re a construction worker building skyscrapers and you’re in a wheelchair, I don’t have to accommodate that you can work on the 110th floor. But if you work at M.I.T., I do have to accommodate that you can get into your space and do your work. 

And here is how President Bush put it at the time:

George H. W. BUSH: With today’s signing of the landmark Americans for Disabilities Act, every man, woman, and child with a disability can now pass through once-closed doors into a bright new era of equality, independence, and freedom.

That sounds pretty great, doesn’t it? At least from the employee side. It might make things more complicated and expensive from the employer’s side but — hey, there are tradeoffs everywhere, right? For a society intent on providing good employment opportunities for everyone, the A.D.A. seemed to say that it was worthwhile to ask employers to make these accommodations. Josh Angrist, meanwhile — he got to wondering: would there be some unintended consequences of this law with such obviously good intentions? So, he went looking for some data. And he found it, in what’s called the Current Population Survey, which is administered every month by the Census Bureau.

ANGRIST: That’s where the unemployment rate comes from. You know, every month you hear the unemployment rate on the radio? That’s coming from a survey of 60,000 households. And there’s a bunch of questions there: Did you work? Were you looking for work? But there’s actually a lot more there. There’s employment, there’s earnings, there’s schooling. And as it turns out, there’s a question about disability: “Do you have a disability that limits work?”  

For people who answered that question “yes,” Angrist wanted to know whether the A.D.A. was helping. But: he’d need some kind of control variable, a way to compare workers affected by the new law with similar workers who weren’t affected. Luckily, the A.D.A. provided one: companies with fewer than 25 employees were exempted from most of the law’s provisions. This gave Angrist and Acemoglu a nice little natural experiment, as economists like to call it. So: they measured, they sorted, they analyzed — and they wrote up their findings for the Journal of Political Economy, one of the top journals in their field. What was the headline result?

ANGRIST: The surprising unintended consequence is that the A.D.A. does not seem to have helped disabled workers. If anything, it made them worse off by reducing their employment rates and their annual earnings. 

DUBNER: And can you explain why their employment would have been reduced? What’s actually happening at a firm? 

ANGRIST: Well, employers, to the extent that they can tell who’s disabled, they just don’t want to get into it. Because they don’t know what the costs of accommodation are going to be. The cost of accommodation could be very mild. It could just be a matter of letting somebody work on the ground floor, say. But they could be very high. And they’re sort of unknown. Once you go down that road and you’re in the business of accommodating, you potentially are on the hook for something big. Now mostly that’s not going to happen, but it is a problem. 

DUBNER: Now, wait a minute. Wasn’t part of the Americans with Disabilities Act a provision that a firm wouldn’t discriminate against hiring a disabled worker? And therefore, by choosing not to hire a disabled worker because you didn’t want to deal with the consequences, you are violating the A.D.A.? 

ANGRIST: But it’s much harder to make a case on the hiring front than on the discharge. How do you make a case that I’m not hiring? You don’t have any data on who I interviewed. And, it’s much harder to prove that, because I can always say, “Well, that guy wasn’t qualified.”

DUBNER: We don’t know if Josh Angrist was our first choice for this interview, for instance. 

ANGRIST: Totally. Probably not. So some workers, you might decide, even if you don’t know they’re disabled, you can kind of predict this person is going to be trouble. And they’re going to sue me. And that becomes much easier once they’re in the workplace. And the thing is, you don’t have to win for this to be a problem.  Mostly you won’t win. It’ll settle. But it’s a hassle. And it’s a cost.

DUBNER: When I first read this paper years ago, it was one of those mind-blowing but obvious-in-retrospect findings. Like, if you hadn’t spelled it out, I wouldn’t have thought of it on my own. But once you spell it out, you see, yes, that’s the way humans behave. 

ANGRIST: Well, it wasn’t obvious at the time, and it certainly was controversial. 

DUBNER: You must have upset some people? 

ANGRIST: Yeah, yeah, there were people that weren’t happy. Some disabled groups that were proud of the A.D.A. There were some economists that didn’t accept the finding and went and did their own work on it. But the finding mostly held up, in my view. We also did some cross-state analysis. Some states have more litigation than others. And we saw that that was a good predictor of where relative employment of disabled workers is going to fall. 

The A.D.A. has been amended since its original passage; it also had its scope narrowed by several Supreme Court cases, while some states have passed their own laws to protect workers with disabilities. That said, such workers are still far less likely to be employed than workers without a disability. Economists, meanwhile, continue to study which policies are best at actually helping workers with disabilities; that, after all, is the economist’s job: to analyze the costs along with the benefits.

CULLEN: I mean, my job is to just point out the trade-offs.  

That is Zoe Cullen. She, too, is an economist, at the Harvard Business School. Her students are in training to create and manage organizations.

CULLEN: I like the study of organizations. I think that’s where my topics, which would maybe traditionally be more on the labor end of economics, become managerially relevant. 

For instance: how do firms and managers set salaries for their employees? Do all employees with the same job title and experience get paid the same? And: how much do employees know about their colleagues’ salaries? For employees, this is an important question if you’re looking for a job or if you’re hoping for a promotion.

CULLEN: You know, you find out that you’re up for promotion and you’re going to negotiate this new contract. And the first thing you wonder is, well, what are they willing to pay me? And if you could only just talk to the people who recently were promoted, recently negotiated their contracts, maybe you could hold out for a better deal.

Most firms are not very explicit about the salaries they pay. Sometimes they’ll give a range. And there are sites like Glassdoor, that compile data from former and current employees. But this kind of data isn’t complete, or even all that reliable: it’s posted anonymously; it includes a lot of lower-paid and often disgruntled employees; the data can also be outdated, and fail to include total compensation, like benefits. From the other direction, some companies have been found to solicit fake positive feedback on Glassdoor. So, if you look at the whole picture — the available data around salaries is often imperfect information. Most of the theories put forth by labor economists, meanwhile, assume something closer to perfect information.

CULLEN: Yeah, so for example, if an employer has perfect information about market prices, they’re going to indeed know everything they need to know about market prices to make their decision. When we introduce incomplete information, that’s typically a model where we have to be more explicit about exactly what’s outside the information set. So in the case of market wages, the employer might have private information about exactly what they think a candidate is worth and what they would be willing to pay, but they only have either a signal or know the distribution of pay that their competitors are drawing from. 

An employer does have the obvious advantage of at least knowing what they pay all of their employees. Employees, meanwhile, have much less information. This gives the employer some real leverage, and some firms exploit that leverage when they can. It happened to Josh Angrist when he landed his current job at M.I.T.

ANGRIST: I came to M.I.T. as a full-time faculty member in 1996. I was happy as a clam to be at M.I.T., thinking I’ve really done well. And my former thesis advisor, Orley Ashenfelter, who’s a very famous labor economist, came to my office to chat, and Orley said, “So, what do you make?” And I said, “I make $85,000. I can’t believe I make $85,000 a year.” And he said, “Oh my God.” He said, “You can’t work for that. That’s not what tenured labor economists make. If that gets out, that’s going to be very bad. So you need to go and get a raise.” 

DUBNER: Orley was teaching at Princeton at this time?  


DUBNER: He’s inciting you to riot against M.I.T. basically? 

ANGRIST: Exactly. I didn’t know what my peers make — as somebody has written, you’re more likely to know about your colleagues’ sex lives than their salary. And I now know, having served on some committees, that the variance within departments can be huge. So I was absolutely on the low end. 

DUBNER: So what did you do?

ANGRIST: I went and I said, “I’m going to leave.” And I had to generate — you know, what academics have to do is generate offers. If you want to threaten to leave, it’s like baseball; you have to say, I’m going to get myself traded to the Yankees. 

DUBNER: Could you get an offer from Princeton, for instance? 

ANGRIST: For instance. And then it has to be credible, so you have to be prepared to take it if they don’t pony up. 

DUBNER: So you did that.

ANGRIST: I did that and eventually I got a raise. 

DUBNER: What’d you get a raise to?

ANGRIST: I got a very large raise. I think I got up to around 150, which shows you how I was very underpaid.

DUBNER: What cut of that raise did you give to Orley?

ANGRIST: I still owe him. The thing is, I owe Orley for so many things, that wouldn’t begin to pay it.

DUBNER: What does that say to you? The very fact that that gap exists, between the 85 and the 150, and all you had to do was basically make it transparent. What does that tell you about the way firms operate? M.I.T. is just another firm. 

ANGRIST: Well, some labor markets are more efficient than others. So, we live in a labor market where — meaning professors —  there’s a lot of variance in pay, and there’s also a lot of variance in productivity. We’re much closer to people in the performing arts or sports — every actor or entertainer, every M.L.B., N.B.A., N.F.L., you name it, they’re all their own market, right? They all have their own package of attributes. And it is kind of hard to know what should they be paid.

So what would happen if firms decided to be more transparent about what all their employees were paid? Wouldn’t that be a net benefit for employees?

CULLEN: That’s what I think people think. 

Coming up: the law of unintended consequences strikes again.

*      *      *

You may have noticed that there is a growing trend in what are called pay transparency laws, which require firms to give more complete information about what employees should expect to earn. A number of countries have passed such laws recently, and it’s catching on in the U.S. too, in places including California and New York. The Harvard Business School economist Zoe Cullen has been studying the impact of these laws. What are the desired outcomes of such policies, and are those outcomes achieved? Are these policies fair? Cullen says that pay-transparency policies come in three categories. The first establishes what she calls horizontal transparency.

CULLEN: It’s just a simple way of saying, I have a peer, what is the pay transparency between these two peers? 

The second category is called cross-firm transparency.

CULLEN: This is where the visibility is now between firms. So you’d be learning about what you would get if you moved. 

DUBNER: This is, let’s say, the Yale School of Management comes after Zoe Cullen from Harvard Business School, correct? Has that happened, by the way, just for the record?

CULLEN: I’m not allowed to publicly talk about that, am I? It’s not only transparency that a worker might have about different firm salaries, but it’s also actually what a firm might perceive their competitors to be paying. 

The third category is what Cullen calls vertical transparency.

CULLEN: Vertical is this understudied, very important mid-point. There are very few instances where public policy is really focused on vertical transparency, but this is happening to some extent by accident, which is finding out not just what your peers are earning, but what your senior management is earning, what you would earn if you were to go up the corporate ladder. 

DUBNER: Hmm. Why would you say public policy hasn’t yet cared much about vertical? Because I would think, given all the conversation about pay inequality and income inequality over the past many years now, that that would be actually a focus even more than horizontal, perhaps.

CULLEN: My sense is that most of us think firms are taking care of this in in the way that they are doing in the economic textbooks. In an economic textbook, you might say, well, the firm has to internalize the career incentives of the employee, and they should incentivize them to both stay with the firm and see that they are growing into bigger positions in order to keep them loyal to the firm. So the idea would be that it’s the onus of the firm to essentially be very clear about the steep salary progression.  

DUBNER: Do younger employees tend to talk about pay transparency, and want to have more pay transparency, than older workers? 

CULLEN: Well, I’ve heard millennials do chat more about pay, but let me just say this: we also see a strong pattern of secrecy as you become richer. And age and wealth are highly collinear. Tom Nicholas and I, my colleague at H.B.S., looked at who withholds their information about income from the U.S. Census. And there, the patterns seem pretty strong that it is about the money. 

DUBNER: Now, you must have some psychologist friends, put on your psychologist hat for a moment and try to explain why that would be. 

CULLEN: The term that comes up most often when you just have an open text box for “Why don’t you want to share this information?” is around fear of resentment. So that word, “resentment,” I take quite seriously. You might also think there’s some strategic element to this that is harder to articulate. So, you know, in the context of a workplace, you might have a sense that if it got out that you just got a big raise, perhaps other people would be vying for the same pie. 

“Vying for the same pie”: meaning: if your salary goes up, there may not be as much pie left for me. So what happens once salaries are made public? That’s the question Cullen set out to answer. She recently published a paper that surveyed all the relevant research on pay-transparency policies, including her own research. This was for the Journal of Economic Perspectives. Most economics journals want their contributors to describe the research but remain agnostic about policy recommendations; but the J.E.P. is different.

CULLEN: They like us to take a stance. Hence, the title “Is Pay Transparency Good?” So, less of a literature review and much more about where I stand on this topic. 

DUBNER: And the implication there would seem to be that more information is good, full stop. Is that the case? 

CULLEN: That’s what I think people think.

DUBNER: And you’re here to tell us … ? 

CULLEN: Just that it’s, you know, the typical economist, not-so-simple view. 

Cullen’s own research used data from a variety of sources. Some came from TaskRabbit, the online platform where gig workers submit public bids for different jobs. Some data came from ADP, the huge payroll company that manages worker paychecks and benefits across the country. She also relied on Census data that tracked wages and employment histories for more than 5 million people. This allowed Cullen to compare workers in states that have pay-transparency regulations with those that don’t. Here’s what she found: at companies in states that have pay-transparency laws, wages became — as she puts it — “linked together.”

CULLEN: When one person negotiates for their salary, typically it will be in an environment that’s highly private. Once you start to introduce transparency as a way of pinning the employer to pay people the same, suddenly one person’s negotiation affects someone else’s.

Okay, so that makes sense: when salary information is private, there might be a lot of variance; once it’s public, wages tend to converge. And that’s exactly what these laws are trying to accomplish: to prevent firms from rewarding or punishing individual employees based on some kind of bias or favoritism. But that wasn’t the only way these laws affected firms.

CULLEN: You can just see how it increases their incentives to bargain aggressively. Because if they save a dollar with one worker, they save that dollar with everybody. Ultimately you start to see what I called compressed pay.  

In one paper, Cullen finds that in states with transparency laws that protect workers’ rights to discuss their compensation, there was, overall, a 2 percent decline in wages.

CULLEN: Essentially, the set of pay transparency policies that have been most popular, are also having this unintended consequence of linking bargaining practices across workers in lowering average wages. 

DUBNER: So does your finding, Zoe, that pay transparency can lead to overall lower pay — does it strike you as an unintended consequence of the policies? 

CULLEN: I don’t think policymakers talked about that component of it. 

DUBNER: So, that’s a yes?

CULLEN: Yeah, I think I didn’t know it myself. And I found it very surprising.

DUBNER: Can you give us a sense of the impact of pay transparency on any observable pay gap between women and men? I assume that is a major goal of pay transparency.

CULLEN: I think it’s pretty uniformly great at achieving this goal. In all these evaluations of country-wide policies, we’ve seen positive effects on gender gaps. Pay transparency leads to more equal pay. 

DUBNER: Is it too reductive to say, then, that pay transparency is on average, significantly better for women than men? 

CULLEN: No. I think that it’s a challenging statement to make, in part because — well, I think the presumption that you have in the back of your mind is that women are at the lower end of the pay scale. And so this is a question of, do the people at the bottom necessarily do better under pay by transparency? When we talk about societal level transparency, I don’t think it’s clear. 

DUBNER: So it sounds like you might consider this a sort of growing pain of pay equity. Would that be accurate?

CULLEN: Yes. And I can’t emphasize enough that, with slight adjustments to what these policies are doing, in fact, you can get pay growth and upward pressure on pay through transparencies. It’s as though you can get your cake and eat it too, with just thinking a little more broadly about the intended goal.

DUBNER: Can I have those slight adjustments, please?

CULLEN: Yes. Don’t focus — so I say “don’t,” I’m using the imperative with you.

DUBNER: That’s okay. That means you’re comfortable. I’m happy about that.

CULLEN: Okay. Thank you. Stephen, I think that’s an invitation. 

DUBNER: It is an invitation. I hereby formally invite you to use the imperative. 

CULLEN: So, don’t focus so much on trying to get one employee to compare themselves to another employee. We want the comparisons to be one firm realizing that in order to compete for talent, they need to raise wages. Firms have to understand what the competing wages look like. And employees have to figure out where to send those applications. The idea here is really important. You have a setting where employers don’t typically have full information about market pay. Workers don’t have full information about market pay. With simple information tools, you can see that employees respond by submitting applications not only outside of their own occupation, but to higher-paying firms within their occupation. When you start to increase what they know about the rest of the market, it’s exactly the people who are being underpaid that are positively surprised.

There will likely be more pay transparency laws in the coming years. The Biden administration just announced it plans to require federal contractors to provide job applicants with expected salary ranges. I know there are quite a few Freakonomics Radio listeners in the White House. If you want to give Zoe Cullen a call to help design that policy, we’d happily pass on her number. I do wonder if she’ll use the imperative with you! Coming up: one more instance of good intentions and an unintended consequence:

GERTSBERG: It’s definitely a frustrating finding, because it shouldn’t be a trade-off.

*      *      *

Over the past few decades, a lot of policies have been designed to make workplaces more equitable for women. But, as we’ve been hearing, not all policies achieve their desired result. Consider the research done by the economists Peter Blair and Ben Posmanick. They found that family-leave policies, which give employees time off for childbirth and other family issues, have had the unintended consequence of increasing the average pay gap between women and men. Why? Because women are more likely to take advantage of these policies, and men end up earning more on average. So, as we’ve been hearing throughout this episode: good intentions do not guarantee good outcomes. Consider the recent work of Marina Gertsberg. She was born in Ukraine, grew up in Germany, got her Ph.D. in the Netherlands, worked for a time in New York, and now teaches in Australia.

GERTSBERG: I’m an assistant professor at the University of Melbourne, in the finance department.

Before getting her Ph.D., Gertsberg had a variety of internships and jobs in a variety of industries — banking, management consulting, the art industry.

GERTSBERG: I worked my whole life in industries that are male-dominated. And I have to say that I felt fairly comfortable as a woman. I didn’t notice that much discrimination or anything like that.

More recently, Gertsberg got to wondering about discrimination in academia — specifically, in academic research. She decided to focus on research collaborations. This is an especially important area for younger academics.

GERTSBERG: Working with a senior person, that has a lot of advantages. The senior person has a network. Promoting the paper is easier. Potentially the name of the senior academic also helps with conference submissions. Also in the publication process, the name of the senior person will help. Of course, the knowledge of the senior person is also very helpful. They usually have more experience with how to frame the paper. What is a convincing methodology? What is an interesting question? Collaborations are really important for productivity in academia.

But collaborations in academia are different from collaborations in many professional settings. In most places, a junior person is assigned to a senior person, or gets attached to a project that a senior person is running.

GERTSBERG: Part of our profession is that our collaborations emerge organically. We decide ourselves who we work with. So if you don’t want to work with a person, you just won’t work on a project with them. No one can force you. What is also a really important aspect is that the lines between the professional and the personal oftentimes blur. So, we create ideas outside of the office. We work outside of the office. We discuss ideas in informal settings such as in cafes, over dinner. And these are contexts where ambiguous situations could arise.

“Ambiguous situations” like: is this collaboration purely professional, or is there perhaps a romantic or sexual component? Given the realities and history of the male-female dynamic, Gertsberg thought about cases in which a junior female academic collaborated with a senior male academic. And she started to put together a research project. Much like the Josh Angrist research we heard about earlier, that had a fulcrum event — in that case, the passage of the A.D.A., the Americans with Disabilities Act — Gertsberg’s research would also have a fulcrum, a before and an after. In this case, the before and after was the #MeToo movement.

GERTSBERG: So, people view the event date of the #MeToo movement as October 15th, 2017, when Alyssa Milano tweeted that she was sexually harassed, and encouraged other women to come forward as well.

The #MeToo movement was meant to expose men who had sexually harassed women, and to prevent future harassment. Those were the intended consequences. Gertsberg wondered if there might be an unintended consequence as well.

GERSTBERG Women and men equally started to express the perception that, yeah, men may be taking up precautionary action in interacting with women, starting to be more careful because they are concerned they would be accused of sexual harassment. And yeah, this is the time when the idea was actually born. 

“The idea” being a research paper, which she would eventually call “The Unintended Consequences of #MeToo: Evidence from Research Collaborations.”

GERTSBERG: At the time, I also pitched it to a senior academic, to just get a sense whether this was a good idea. The senior academic told me that I should rather not work on this at stage in my career.

DUBNER: Why not?

GERTSBERG: Well, because potentially it would be too controversial.

DUBNER: Was this a male or female academic?

GERTSBERG: It was a male academic.

DUBNER: And yet you ignored this person’s advice, plainly?

GERTSBERG: No, I didn’t ignore it. I didn’t work on it for some time. Because that was during my Ph.D. And I had other things on my plate as well, so I decided to not work on it. 

But Gertsberg couldn’t get the idea out of her head. Once she got settled into her first academic job, in Australia, she put together a hypothesis.

GERTSBERG The interesting thing about the #MeToo movement is that its purpose is to increase protection for women from sexual harassment. So technically, women should feel very comfortable, or more comfortable to work with men after the movement. On the other hand, there’s a lot of anecdotal evidence and also survey evidence that men are concerned about sexual harassment accusations after the #MeToo movement. So it is unclear what effect the movement would have on collaborations between women and men. So it is a two-sided hypothesis.

To test her hypothesis, Gertsberg began collecting data on junior female economists, women who had recently gotten their Ph.D.s and were hired into university economics departments on a tenure track shortly before the apex of #MeToo. There were not all that many women in her sample — fewer than 100. She gathered data on the research papers they were publishing, and who they were collaborating with. She wanted to know what happened to collaborations between those junior female academics and their male colleagues. And what’d she find?

GERTSBERG: I find that junior female academics start fewer new projects after #MeToo. And that is mainly due to fewer collaborations with male coauthors. Before the #MeToo movement, on average, junior women would start 1.6 new projects per year. And after the #MeToo movement, women start on average 0.9 new projects per year. So, the magnitude is 0.7 projects fewer after the #MeToo movement, which is about 44 percent.

DUBNER: Oh my goodness.

GERTSBERG: And 60 percent of that decline are due to fewer collaborations with male coauthors.

DUBNER: I mean, that is a massive drop. How do you think about the size of the harm to the career of a junior female academic based on that number?

GERTSBERG: We already know that women have less output than men. And that partially explains worse career outcomes for women in academia, such as lower tenure rates. So, having this productive output is crucial. And if that declines, that could widen the gender gap between women and men in academia. Of course, it is also important to then look at the outcomes. For example, tenure rates of women after #MeToo or the publication outcomes, and that is also something that I’m tracking. One challenge is that the publication process is very, very long and for a lot of the projects that started after #MeToo, there are no outcomes yet.

DUBNER: What can you tell us about the degree to which junior women sought out senior women to collaborate with?

GERTSBERG: After #MeToo, women do not increase collaborations with any types of women.

DUBNER: And why do you think that is?

GERTSBERG: It could be, for example, that women may need more time to adjust and find new collaborators among women. It could also be the case that there are simply not enough women for women to substitute with. So if you think about the numbers, even among junior female academics, they are about 30 percent female. So even substituting with new junior woman is fairly difficult with those numbers.

DUBNER: So you’re saying that — among senior female researchers, there’s only like 16 percent?


DUBNER: So you’re saying those 16 percent are probably oversubscribed with junior collaborators already, and there just isn’t enough availability for junior women to make new collaborations with them, yes?

GERTSBERG: Yeah, that could be one explanation.

For what it’s worth, Marina Gertsberg had no collaborators on her paper about this unintended consequence of the #MeToo movement. It’s worth noting that her sample size was small, and her timeframe relatively tight, so we should be cautious in giving her findings too much weight. But a group of researchers at the University of Cambridge seem to have corroborated Gertsberg’s results, using different methods. And if you go outside the academic world, you see a similar effect. A recent Pew poll shows that nearly 50 percent of men say it’s harder for them to know how to interact with women at work. And here’s how a headline from Bloomberg News put it: “Wall Street Rule for the #MeToo Era: Avoid Women at All Cost.” Here’s Marina Gertsberg again, talking about her own research.

GERTSBERG: It’s definitely a frustrating finding, because it shouldn’t be a trade-off. Women shouldn’t have to choose between a safe workplace  — not being sexually harassed — and their career outcomes or productivity. On the other hand, I think my finding suggests that we can do something about that. As some time passes, there will be new equilibrium, and men and women know how to interact with each other. But it is also important to define what the expectations are for behavior. So that men don’t think, “Oh, I just have to say the right thing, and I’m going to get fired. And so it justifies my behavior by not working with women.”  

And what kind of feedback has Gertsberg gotten since publishing these controversial findings?

GERTSBERG: I posted the first version of the paper in August 2022 on Twitter, and I received a lot of reactions from Twitter. I believe this is also when Josh Angrist became aware of the paper. And I received an email from him saying that the paper was interesting, and that it would create a lot of controversy. That was really a high point of my career at the time.

And here, again, is Josh Angrist, from M.I.T.

ANGRIST: I mean, it has a little bit of an A.D.A.-like flavor. So, you tried to protect a group — in this case women, mostly young women. And maybe what you did is you made people think, “You know, what’s in this for me? I might just get in trouble. So, better for me to stay away.” So it’s a lot like the employer who’s worried about being sued by not accommodating or discharging a disabled worker and then having to deal with litigation.

DUBNER: I have to say, it’s so fascinating to hear you talk about these constructs in a way that is, you know, quite rational and compelling and so on. It’s also, however, Josh, if you don’t mind me saying so, a little bit depressing. Because there are all these well-intended policies and people writing policies trying to help other people, and we find that — not in a, anywhere near the majority of cases, or at least I gather not, but in not a tiny fraction — there is a backfire effect, or this unintended consequence. So, how does one as a right-thinking human who wants the best for people and wants employers and employees to be happy and well-compensated and so on, how do you think your way around the big issue of all these unintended consequences, that promote worse outcomes instead of better outcomes?

ANGRIST: Well, first and foremost, I want to draw your attention to the trade-offs. Personally, I guess that’s why I’m an economist. You know, I don’t find the possibility of trade-offs depressing. I find the possibility of trade-offs interesting. That’s what I study. I recognize that it’s not a perfect world. And that policy design is always about trade-offs. And there’s still things that are worth doing in spite of the trade-offs. But I want you to look at that in a clear way, and be aware of that.

Consider us aware. Thanks to Josh Angrist, Marina Gertsberg, and Zoe Cullen for their excellent teaching today. I learned a lot, hope you did too.

*      *      *

Freakonomics Radio is produced by Stitcher and Renbud Radio. This episode was produced by Zack Lapinski. Our staff also includes Alina Kulman, Augusta Chapman, Eleanor Osborne, Elsa Hernandez, Gabriel Roth, Greg Rippin, Jasmin Klinger, Jeremy Johnston, Julie Kanfer, Lyric Bowditch, Morgan Levey, Neal Carruth, Rebecca Lee Douglas, and Sarah Lilley. Our theme song is “Mr. Fortune,” by the Hitchhikers; the rest of the music this week was composed by Luis Guerra. You can follow Freakonomics Radio on Apple PodcastsSpotify, or wherever you get your podcasts.

Read full Transcript


  • Joshua Angrist, professor of economics at the Massachusetts Institute of Technology.
  • Zoe Cullen, professor of business administration at Harvard Business School.
  • Marina Gertsberg, assistant professor of finance at the University of Melbourne.


Episode Video