Search the Site

Episode Transcript

JOHNSON: My name is Katie Johnson, and I’m a data scientist. 

Johnson is 31 years old, and lives in London. She grew up near Bristol, went to university in Birmingham, and since then has held a series of increasingly impressive jobs at a series of companies. These were all what are known as I.C. jobs, I.C. standing for individual contributor. Which means what? 

JOHNSON: It is someone who makes as opposed to managing people who make. 

Johnson loved being an I.C.; she loved analyzing data, and she was really good at her job. But after a while, she thought it might be nice to become a boss. 

JOHNSON: I wanted to manage more and more people.

DUBNER: And you wanted to manage more people, because why? You were just power-hungry like the rest of us? 

JOHNSON: I think there’s a couple of reasons. So the first is that I wanted to start getting more autonomy over what I was working on. I would be working on stuff in my I.C. role, and I’d think, “This isn’t the most important thing.” And I thought that if I became the leader of the team, then I would get to pick what I worked on. 

Okay, that seems sensible. 

JOHNSON: The other reason was to have more impact at the companies I was working at. So you could describe this as “having a seat at the table.” 

Also sensible. 

JOHNSON: I guess the final reason is that we all kind of — not everyone, I guess, but I was included in this — have a concept that being more successful means being more senior. And so in order to not necessarily show others, but definitely myself, that I had achieved and become successful, I needed to keep moving upwards within a company.

Johnson’s father, in his own career, had seen things differently. 

JOHNSON: So my dad has been a network engineer. He recently retired, but he’s been that for his whole career, and he had absolutely no aspiration to become the manager. He’s like, “Why would I want to do that?”

But Katie Johnson did want to become a manager. And several firms were willing to make her one. She took the most appealing offer, at a software firm that helps companies acquire new customers.

JOHNSON: And I was sent on some management training and had to do what can only be described as a very long personality test. And the idea was to tell me what I was good at being good at. 

And what was she particularly good at? 

JOHNSON: Critical thinking, attention to detail, courage — all these internal, thinky-type characteristics. 

You can see why Katie Johnson would seem to be a great boss. Her new job title was Head of Data and Analytics; she had roughly ten people reporting to her. The promotion came with more money, more prestige, more leverage to set the agenda. Also, however, more meetings. 

JOHNSON: Oh, so many meetings. Like, compared to being a data scientist — I’d maybe have a half-hour meeting in the morning, and then I’d just be free to do coding and thinking and making stuff. But I was in meetings — I think Tuesdays, I used to be in meetings for like seven hours.

DUBNER: No offense, but did you not see that coming?

JOHNSON: No, I really didn’t. I thought it would just be like my normal data-scientist job with a few one-to-ones on the side. That was okay because it’s quite interesting; you’re talking about the work, you get into quite depth from problems with my team. It’s more the meetings, like an hour’s coffee with someone to try and set up a better working relationship with their team. Times that by like five or 10 other teams — it’s just draining. 

Keep in mind, this was happening during the pandemic shutdown — so all these meetings were virtual. And as drained as Johnson was from all those meetings, she was getting good reviews as a manager. 

JOHNSON: Yes. People would tell me what a great job I was doing. I was coming across well.

But she found that being a boss made her miserable.

JOHNSON: I would finish my day in my study, walk into the living room, put a blanket over my head, and cry because I was in so much pain at how bored I was.

In retrospect, Katie Johnson had plainly erred in wanting to become a boss. But she’d also felt that management was the only sensible way to advance her career. And if you look at how most firms and institutions around the world operate, you’d have to agree with her. The question is: does this standard operating procedure produce good bosses or bad bosses? Or even horrible ones? The horrible boss is a familiar caricature. We all know the stereotypes: the screamer, the sadist; the idea-stealer, the passive-aggressivist. These are some of our most enduring characters in film. You remember Blake from Glengarry Glen Ross, played by Alec Baldwin?

BLAKE: Put that coffee down. Coffee’s for closers only. Do you call yourself a salesman, you son of a b****?

Or in the film Office Space — when Peter is trying to escape the office on Friday afternoon, and he gets snagged by the boss?

Bill LUMBERGH: Hello, Peter. What’s happening? I’m gonna need you to go ahead and come in tomorrow. So if you could be here around nine — that would be great, mm-kay? Oh, oh, and I almost forgot — I’m also gonna need you to go ahead and come in on Sunday, too. 

Then there’s Miranda Priestly, played by Meryl Streep, in The Devil Wears Prada. 

Miranda PRIESTLY: You have no style or sense of fashion.

Andy SACHS: Well, um, I think that depends on what your —

PRIESTLY: No, no — that wasn’t a question.

The “horrible boss” motif is so attractive that the director Seth Gordon made a film called Horrible Bosses

Bobby PELLITT: Yeah, we’ve gotta trim some of the fat around here.

Kurt BUCKMAN: Trim the — what do you mean by “trim the fat”?

PELLITT: I want you to fire the fat people.

Truly horrible bosses do occasionally turn up in real life — especially in Hollywood itself. The producer Scott Rudin, for instance, has been accused of years’ worth of alleged abuses — like smashing an assistant’s hand with a computer monitor. But even in Hollywood these days, and especially in more normal industries, this sort of grotesquery is harder to get away with. Bosses who are outright monsters are more likely to lose their jobs. But how much attention are we paying to the more common type of bad boss — someone who’s simply incompetent, or overstretched, or even just miserable being a boss, like Katie Johnson was? Do we even know how many “bad bosses” are out there? The more you dig, the more you learn that the science of boss behavior is not very scientific. One Gallup poll shows that roughly 50 percent of American employees have, at some point in their career, left a job because of a bad boss. But an employee might have 10 or 20 bosses over a career, so maybe that number isn’t so bad. A survey of European employees found that only 13 percent rated their current boss as bad. So maybe the Hollywood caricature is way off. Still, considering that nearly all of us will at some point in our lives have a boss or be one, we thought there might be some boss questions worth asking. And so, today on Freakonomics Radio: when a boss is a bad boss, have you ever wondered why? 

Steve TADELIS: There’s no reason to believe that a great salesperson will be a great manager.

And yet this kind of promotion happens all the time. Why is that? 

Kelly SHUE: So there are two ways to motivate people: we can pay them a whole lot more or we can give them an opportunity for promotion.

Today on the show: why good employees become bad bosses, and whether that will ever change. Spoiler alert: probably not. 

*      *      *

One of the reasons I became a writer, years ago, is because I didn’t particularly like having a boss. Like Katie Johnson, I prefer to set my own agenda, my own pace. I also really like working alone. Also like Katie Johnson, I am not particularly fond of meetings, so I wouldn’t be a very good boss either. Fortunately, at Freakonomics Radio, there are a couple other people who do all the bossy stuff, leaving me pretty much free to do this — what we’re doing right now. Asking questions, trying to find answers. So here’s a question I’ve always been curious about: how important are bosses anyway? I don’t mean C.E.O.s, the ultimate boss; if you’re interested in that, we once did a series called “The Secret Life of C.E.O.s.” Today, we are just talking about your standard-issue middle manager — do they really matter?

TADELIS: Yes. Broadly speaking, managers matter. Bosses matter for outcomes.

That is Steve Tadelis. He’s an economics professor at U.C. Berkeley’s Haas School of Business — a training ground for future bosses. Management is not something that Tadelis himself aspires to. 

TADELIS: Tell me how close you are to administration so I know how far away to be from you.

But he has spent time, while on sabbatical, working as a boss at some well-known firms. 

TADELIS: When I was at eBay and Amazon, I managed teams, and I enjoyed it very much.

DUBNER: How do you assess yourself as a manager in that realm?

TADELIS: I’m blushing so — 

DUBNER: Because you’re the best ever?

TADELIS: No, but I’m pretty good, so I’m feeling a little uncomfortable.

DUBNER: Your positive self-assessment is based on direct feedback or just a general warm-glow feeling?

TADELIS: At eBay and Amazon, the feedback was actually formal through surveys.

Surveys, that is, with questions like: “On a scale of one to five, how much do you agree with the following statement: my boss generates a positive attitude in the team.” Or: “my boss is someone I can trust.” Or: “my boss provides continuous coaching and guidance on how I can improve my performance.” These surveys led Steve Tadelis to ask his own, bigger questions about bosses. For instance: 

TADELIS: Does it really matter? Do these measures of manager skills or characteristics — do they really have any value for the firm? Is there some way in which managers who score higher on these surveys are actually contributing more? 

These are eternal questions in the field known as personnel economics. You could ask the same questions about any manager: the head coach of a football team; the chairperson of your homeowners’ association; the President of the United States. But as I mentioned earlier, the academic literature on the impact of bosses is not particularly advanced. You can see why, if you think about it. There are so many variables in the relationship between a boss and their employees that it can be hard to pinpoint the effects of the boss. This is why most research focuses on one single quantifiable metric: productivity.

TADELIS: For example, there is a paper by the late, wonderful economist Eddie Lazear, Kathryn Shaw, and Chris Stanton, where they show that there is variation in output of employees based on the managers that are in charge of them.

That paper, from 2015, analyzed data from a single firm that the researchers were not allowed to identify; but it appears to be something like a call center. The analysis looked at what happened when a worker moved from what the researchers identified as an average boss to a high-quality boss.  Such a move, they found, increased productivity by as much as 50 percent — so if this were a call center, and a given worker handled 100 calls per shift under an average boss, an excellent boss could boost that to 150 calls. So at least in this type of setting, a “good” boss is doing something right, but the data couldn’t say what. Steve Tadelis wanted to learn more. So he teamed up with Mitchell Hoffman, an economist at the University of Toronto’s Rotman School of Management, to write a research paper.

TADELIS: I had access to interesting data and people in this company — that will have to be unnamed, because when it comes to personnel data, companies are very hesitant. 

Tadelis would only say that this firm did “high-tech, knowledge-based” work. Maybe, given his history, you might picture a firm like an eBay or an Amazon. In any case, he is looking at a very different type of work than the earlier research with its narrow measure of productivity. 

TADELIS: What we’re doing is opening the hood up a little bit. 

And what sort of data did they have access to? 

TADELIS: We have data that allows us to measure the impact of a particular manager skill that we’re calling people-management skills as opposed to just, “do managers matter?”

“People management skills” meaning the sort of things you find on those employee-feedback surveys: how well the manager coaches and communicates, how trustworthy they are. So that’s the boss data. On the employee side, Tadelis and Hoffman had a lot of concrete data: subjective performance scores as well as how often a given employee was promoted or given a raise, the number of patents they filed, and whether they stayed at the firm or left. 

TADELIS: In these high-tech, knowledge-based companies, retention is a very, very important focus because getting these high-skilled workers is not easy. And there’s a lot of competition.

There’s a lot of competition especially now, with more Americans quitting their jobs — for reasons related to the pandemic and otherwise. Although we should say: this paper used data from before the pandemic.

TADELIS: And when you lose an employee, especially an employee that’s very valuable, then it could take months to replace them.

So Tadelis and Hoffman set about to sort through all this data to look for any causal relationships between the rating of a given manager and the various outcomes of the employees working under them. What’d they find? For the most part, it was a big bag of nothing.

TADELIS: We didn’t find that the ratings of the managers seem to impact the subjective performance of their employees, their income, their promotions, or patent applications in a meaningful way.

That’s right. On all those employee outcomes — performance, earnings, patents — it just didn’t seem to matter whether the manager was highly rated or poorly rated. But: there was one other outcome to look at: employee retention. 

TADELIS: Bingo. 

Tadelis and Hoffman looked at employees at this one firm who moved from a manager with a poor rating to one with a high rating. 

TADELIS: That’s associated with an attrition drop of about 60 percent. 

That is huge. And within that huge effect was an important nuance.

TADELIS: What we see then is that managers help retain better employees more than worse employees, which shows that the impact of being a better manager is strongest where it matters the most.

So a good boss seems capable of keeping the best employees happy, and presumably productive. Conversely, a bad boss might drive away the best employees. The Tadelis-Hoffman paper was published in 2021 in the Journal of Political Economy, one of the best econ journals. So, okay, the economics literature on bosses and management just got a little bit deeper. But remember, employee retention was the only outcome where it seemed to matter whether a boss was good or bad. And if you ask Steve Tadelis a more fundamental question — like “What does a good boss actually do, to instill this loyalty?”

TADELIS: This is where I have to take a step back and say that there are certain things that may be outside the scope of what economists should be dealing with. 

DUBNER: If you were to make a list of things that you would like to measure were it possible given the data, what would some of those things be? 

TADELIS: Really good question. Something that’s very hard to measure that I believe is important is compassion. I guess if this is going to be on the radio, I might lose my economist card. 

Steve Tadelis is not the only economist who has been frustrated by the lack of evidence for what makes a good boss good. Maybe compassion is as important as he suspects — but we just don’t have any large-scale, empirical evidence yet. The Stanford economist Nicholas Bloom has been studying leadership and management for years — and yet:

Nicholas BLOOM: No one could really give us a straight answer on what defined a good or bad leader. You look at the data, and there’s 10 different recipes for success. Maybe they each work for a particular case study, but I’ve still, 20 years later, struggled to find anything that’s the secret recipe beyond saying, “Sure, there are some people better than others.” But it’s damn hard to tell what it is.

This has not stopped leadership gurus from promoting their pet theories. As Bloom puts it, there is “a ton of B.S. around this, from airport bookstore pulp fiction.” And here’s another reason to question the literature on management and bosses: as we’ve been hearing, most of the boss data comes from employee surveys. Have you ever taken a survey that rates your manager? If so, were you told it was anonymous? Did you believe it was anonymous? Were your answers objective? Or, did you maybe think: “Well, my boss thinks I’m good at my job, so I’m going to say they’re good at theirs.” Or vice versa: “I don’t think my boss likes me, so I’m sure not going to give them a good rating.” As we’ve said before on this show, survey data can be the lowest form of data. Here, again, is Steve Tadelis:

TADELIS: I’m sure you know that economists are very wary about using surveys, and economists believe in what we call a revealed-preference approach, meaning how you behave is telling me a lot more about you than what you say about yourself.

Just how big is the gap between what people say and how they behave? Over the years, I have heard many economists give many examples of this gap. Steve Tadelis’s example is my all-time favorite: 

TADELIS: There is a lot of discussion about privacy and privacy regulation these days. And you hear a lot of people saying how their privacy is important to them. And then you turn to them and say, “Here’s a Snickers bar. Could I have your mother’s maiden name?” And they say yes. So it’s a little bit confusing when you tell me that you really care about privacy, and then you just scroll down on every app you download and click yes, yes, yes. That doesn’t tell me that you really care about privacy. So the same is true for many other types of behavior. 

So let’s keep in mind that much of what we’ve been told in the past about good bosses — and bad bosses — is not exactly evidence-based. Researchers like Tadelis and Hoffman and Bloom have been chipping away at the black box of boss behavior, but we’ve got a long way to go. This means we need to keep looking for good data and asking good questions. So, coming up after the break: who becomes a boss, and why? If so many people think the boss-selection process is stupid, why do firms keep doing it? And whatever happened to Katie Johnson? 

JOHNSON: I’d get to the end of the day and the last thing I want to do is talk to someone else. 

I’m Stephen Dubner; this is Freakonomics Radio; and remember: you can get our series “The Secret Life of C.E.O.s” — and all our past episodes — on any podcast app. 

*      *      *

Have you ever thought about where a boss comes from? What I mean is: why a given employee will rise from the ranks to become a manager? Here’s someone who’s been thinking about that, a lot. 

SHUE: My name is Kelly Shue. I am a professor of finance at the Yale School of Management. 

Kelly Shue, along with Alan Benson and Danielle Li, published a paper in The Quarterly Journal of Economics — another top journal — called “Promotions and the Peter Principle.” 

SHUE: The Peter Principle is a very funny and popular management book written by Laurence J. Peter, and his book offers an explanation for why we might see incompetent bosses everywhere.

“Incompetent bosses everywhere?” Okay, I’m listening. What is this explanation? Let’s go back to Laurence J. Peter himself. This is from a 1973 documentary.

Laurence PETER: The Peter principle states very simply that in any hierarchy, an employee tends to rise to his level of incompetence.

DUBNER: I’m sorry — as many times as I’ve heard that phrase, I still laugh at it, just because it sounds like it’s going to be not irreverent. And then it turns immediately irreverent, which makes me chuckle.

SHUE: Exactly. I think it’s a funny idea, but it also rings true. And it’s funny in a kind of unpleasant way, because it reminds people how much they dislike their bosses.

Peter was a Canadian education scholar. He used his daily observations to form a theory about job promotions.

PETER: I saw that very often the competent individual was promoted to something he couldn’t do. I saw a competent mechanic, where I used to take my car. He was terrific. He was very responsible, very precise, knew exactly what he was doing. So they made him foreman. Now he’s no longer fixing cars, and he’s trying to manage other mechanics. And he’s very incompetent.

The more Peter looked around, the more he saw people who were good at their jobs routinely stumbling into bigger jobs they weren’t good at.

PETER: In any organization where competence is essentially eligibility for promotion and incompetence is a bar to promotion, people will rise to their level of incompetence and tend to stay there.

The book he wrote, with Raymond Hull, was called The Peter Principle: Why Things Always Go Wrong. It wound up selling millions of copies. The book was meant to satirize corporate strategy. Nevertheless, a variety of big firms tried to recruit Peter to become their management guru. He declined, saying that he didn’t wish to rise to his own level of incompetence. Kelly Shue again.

SHUE: His idea is that firms and organizations tend to promote people based upon their performance so far. What that means is a worker who is good at her job will be quickly promoted to a new job role, which might require a different set of skills. If she is good at that new role, she’s going to be promoted again — until she reaches a position where she is actually a bad match for that new job role. And then she will no longer be promoted. 

On the one hand, it would seem to make perfect sense that you promote someone who’s good at their job; you don’t want to promote the bad workers. On the other hand, managing is not the same as doing. 

TADELIS: There’s no reason to believe that a great salesperson who knows how to negotiate deals will be a great manager. 

That, again, is the Berkeley economist Steve Tadelis.

TADELIS: I look here and my company, Berkeley. Great researchers often make for lousy department chairs. Great engineers often make for lousy engineering managers. 

But here’s the thing about the Peter Principle. Even though the theory had been around for half a century, no one had ever checked, with real data from real companies, whether Laurence Peter was right. A few observations — about a car mechanic, or an academic researcher-turned-department-chair — those do not constitute empirical proof, especially in the realm of management and all that “airport bookstore pulp fiction.” This is where Kelly Shue and her co-authors come in. They wanted to see if the Peter Principle actually exists and if so, what should be done about it. First step: get hold of some data. 

SHUE: We got our data from a company that offers sales performance management software and services. 

Shue can’t tell us the name of the company, but picture something like Salesforce.

SHUE: A typical client of our data provider is a firm that employs business-to-business sales workers. And that client firm would input the sales numbers and the whole organizational structure into a software program. And what we’re doing is we’re studying the data that these client firms uploaded into the software program.

DUBNER: How many firms and how many workers? 

SHUE: We see data for about 40,000 business-to-business sales workers at over 130 different U.S.-based firms.

DUBNER: And how many of those were in managerial roles?

SHUE: Five thousand managers and about 1,500 promotion events.

DUBNER: So in terms of empirical studies in your realm, this is considered a pretty large and robust data set, or would you have liked it to be even bigger than that? 

SHUE: I would always prefer a bigger dataset, but for this type of question, a very large and comprehensive data set.

DUBNER: So these are sales workers and sales managers. What makes sales a good business function to study? 

SHUE: One is: it is important to study sales workers because almost 10 percent of the U.S. labor force are somehow involved in the sales function. The other benefit is that we have a very good measure of their performance. So we can test: are the stronger performers more likely to be promoted? 

DUBNER: So that makes a lot of sense from your perspective — as the scholar. From my perspective, as someone who’s not in sales, I would think, “Well, your findings may not translate very well, that in a field like journalism or in health care, or in many other fields, the measurables aren’t nearly as measurable as they are in sales.” So how generalizable do you think your findings are? 

SHUE: I believe it’s likely to apply to other settings where the skills required to succeed at one level differ from skills required to succeed in the next level. So some examples are science, manufacturing, academia, entrepreneurship.

DUBNER: Can you think of industries or sectors where this problem wouldn’t apply?

SHUE: It’s actually hard for me to think of a setting in which this problem wouldn’t apply at all. I’ve also seen it in the context of government structures. A good example is actually the ancient Chinese imperial examination system. It’s famous for being a meritocracy even thousands of years ago. So you would take a test and the top scorers on the test would become administrators within the government bureaucracy. But their problem was they would make the test based upon familiarity with classical poetry. And the people who were best at that test would then become tax collectors, which is a different skill set. 

DUBNER: But ancient Chinese poetry was an incredibly rich and diverse body of literature, yes? So I could imagine how a mastery or even a deep appreciation of that could theoretically apply across a number of skills? 

SHUE: Theoretically, yes. 

DUBNER: You sound unconvinced.

SHUE: And to be fair, I do not have the historical data to prove that being the best at classical poetry means you are not the best at tax collection. 

DUBNER: Since you don’t have that data, let’s look at, say, modern U.S. politics. How would you assess the relationship between a person who’s electable and a person who will govern well? 

SHUE: That is a very good point. So, someone who is electable might be very charismatic, very good at public speaking. Whereas the actual function once someone has been elected might involve being good at deal-making, back-office politics, or understanding the actual details of the policies that they’re passing. 

DUBNER: Do you know anything about that question empirically? 

SHUE: I’m drawing a blank, but you really did raise a very good research idea. Maybe I will look into this. We’ve been thinking about settings where this type of problem might apply for a long time, but somehow I’d never thought about the government or elected-official example you just raised. But it seems spot on for having potential as a problem. 

Okay, before I hijacked this conversation with Kelly Shue to talk about politics and ancient Chinese poetry, we were talking about her research paper that tried to identify the Peter Principle in the wild. As Shue told us, she had performance data on roughly 40,000 sales workers at around 130 companies. The next step was to confirm that companies indeed use an employee’s job performance as a trigger for promotion. The answer: yes!

SHUE: We find that doubling in worker sales corresponds to a 30 percent increase in their probability of being promoted. Another way to look at it is — if someone is the top sales worker within their team of five or six people, then that top sales worker has about triple the probability of being promoted relative to the average sales worker.

DUBNER: Now, is that alone evidence of the Peter Principle.

SHUE: No. Just to promote based upon past performance isn’t necessarily a Peter Principle problem, because it could be that the best salespeople really are the best managers of salespeople. And in that case, you want to promote the best salespeople. 

DUBNER: Okay. So the next step, I guess, is seeing whether the best salespeople indeed do become the best managers. How do you do that? 

SHUE: So first, we’re going to measure the quality of each manager. Managers in our data are no longer directly involved in sales. Their job as a manager is to coordinate and facilitate the sales of their subordinates. 

DUBNER: And presumably those subordinates are people they worked with side-by-side and maybe competed against just the week before they were promoted? Is that the case often? 

SHUE: We actually see, for the most part, people when they’re promoted, they’re rotated to a different team, possibly because the firm overall is exactly afraid of those internal team dynamics that you just described. So we don’t want to call someone a good manager just because her team sells a lot. Because we’re worried that maybe she was lucky, and she was assigned to great sales workers and those sales workers could have been great regardless of her managerial input. To get around that problem, we’re going to measure manager quality as the manager’s value-added to her subordinate sales. If my subordinates sell more when they work under me than when they worked under other managers, then I would be considered a high-quality manager. 

So here’s the key question Kelly Shue is asking: does being a good salesperson make you a good manager of other salespeople? Here’s what she found:

SHUE: A manager with double the pre-promotion sales as another manager leads to about a 6 percent decline in subordinate sales.

DUBNER: Oh my goodness. 

SHUE: Yes. What we find is that among promoted managers, those with low sales prior to their promotion, they are actually better at managing their subordinates.

Let me say that again: “Oh my goodness.” When these firms select people to be managers based on their current job performance, they are actively making themselves worse off. In other words: the Peter Principle is as real as Laurence Peter said it was. And, I’m editorializing here, it would also seem to be incredibly stupid.

SHUE: If the firm’s only goal were to have the best possible managers, then the firm could, by putting more weight on collaboration experience and less weight on sales numbers — the firm could promote better managers and raise overall firm sales numbers by about 30 percent.

That’s assuming that “collaboration experience” is in fact more important for a manager than just high sales numbers. Still, a 30 percent increase in revenue simply by killing off the Peter Principle? That would seem to be a no-brainer. So does this mean that modern firms simply aren’t aware of the age-old Peter Principle? 

SHUE: Most firms are aware of the Peter Principle problem, and it’s a problem that they purposely choose to live with. Some evidence we have indicating that: in situations when the firm is trying to select a new manager who is going to be in charge of a very large team — so that’s a situation in which manager quality matters a lot — in those situations, firms put less weight on a worker’s sales numbers, probably because they know you’re going to end up with a bad manager. 

So Shue is arguing that firms know they will get worse managers by simply promoting people who’ve been good at their previous jobs, rather than people who might actually be good managers. And yet, for the most part, they continue to do it — even though it hurts their profits. Why would they do that? Economists are always telling us that companies are, by definition, profit-maximizing machines. Knowingly promoting a bad manager does not sound very profit-maximizing. So are companies just making a mistake? 

SHUE: A firm having a Peter Principle problem doesn’t necessarily mean that the firm doesn’t understand what it’s doing or is making a mistake. 

So what is going on? 

SHUE: What we believe is happening is the firm is doing its best to motivate workers. And they face a trade-off. 

Okay, this is where it gets really interesting.

SHUE: Promoting based upon past performance is very motivating to workers. So it’s a very strong incentive system. We can also work out that it’s in some ways cheaper than offering really strong pay for performance. So there are two ways to motivate people: we can pay them a whole lot more, or we can give them an opportunity for promotion, which they might value a whole lot because that’s something that they can put on their resume, and it increases their status in society. You don’t want to brag about your pay on your resume.

DUBNER: I mean, the minute you say that, it makes me think, “Wait, maybe we should make it more acceptable for people to brag about their pay.” Because wouldn’t that be more efficient in the end and encourage less promotion of people who are going to be bad managers?

SHUE: That’s a fantastic idea. I don’t know of research testing that directly, but I do know in other cultures there’s differences in it being more socially acceptable to talk about your compensation. 

DUBNER: I mean, I was kind of half-joking, but it would be interesting if there was some metric or a badge you get saying, “I’m really good at what I do, and I’m so good that I’ve been rewarded a lot of raises, and plainly, I’m very valuable to the firm. And I could be a manager if I wanted, but I’m better than that.” That’s an incredibly ham-handed, naive way of putting it, but is there any mechanism in managerial science for that kind of delineation between success on a financial level and success on a managerial level? 

SHUE: There have been some interesting attempts in that direction. So I’ve heard of many technology-focused firms, especially those in Silicon Valley — they face this problem that they have a pool of very talented and skilled engineers, and those engineers may not be the best managers of engineers. Many of those firms offer something called a dual-career track, where someone can rise in the ranks of being an engineer — basically having a higher and higher title. So you can start as engineer, then distinguished engineer, then lifetime-distinguished engineer. And that’s a way for the firm to recognize someone’s contributions in a public way without moving them over to management. 

The Berkeley economist Steve Tadelis has also noticed this movement.

TADELIS: In companies like eBay, Google, Amazon, Facebook there’s the term of I.C. or independent contributor, and you will have people who are at the level of V.P. not managing a single person because they are just gods in their realm of engineering, or coding, or architecture, and so on. By distinguishing between I.C.s and the so-called management talent, the firm is saying, “Look, we are going to promote people in ways that reward them for what they’re great at. You’re not a great manager — you’re not going to get incentivized by becoming a manager.”

DUBNER: Has that model trickled out at all of that high-tech realm? 

TADELIS: One area where I have seen it is in consulting companies, where you have this kind of deep technical talent — think of Ph.Ds., etc. — that will remain and be very heavily rewarded for the work they do, and they will not manage people.

The fact that Kelly Shue and Steve Tadelis can identify a handful of cases where career success is not tied to a promotion into management — well, those are exceptions that prove the rule. As Shue found in her research, the Peter Principle is alive and well, as absurd as that may seem. It is yet another confirmation that management science, as lovely a phrase as that may seem, is not yet very scientific. Most firms stick with what they’ve always done: when an employee is good at what they do, you turn them into a manager to oversee other people who do what they used to do. Even if they’re not cut out to be a manager. Like our friend Katie Johnson, the English data scientist we met earlier.

JOHNSON: I didn’t see that there was another path whereby I could be director level, but not have direct reports. I just didn’t ever see that. 

Looking back, there were some clues that Johnson wasn’t quite manager material. You remember, during management training, that she took that personality test, and she told us the areas where she got high scores?

JOHNSON: Critical thinking, attention to detail, courage, all these internal, thinky-type characteristics. 

Well, those weren’t the only results of this test. 

JOHNSON: Things that I can do, but I struggle with, was compassion, empathy, relationship-building. I saw this output, and I was like, “Why didn’t anyone do this to me before I got this job?” Because this just screams, “great data scientist, not so great manager.”

But it was too late. She’d already been made a manager. As you’ll recall, it wasn’t going well. 

JOHNSON: I would finish my day in my study, walk into the living room, put a blanket over my head, and cry.

DUBNER: So let’s say we’re talking a scale of zero to 10. Where would you put your median satisfaction when you were an I.C., or a maker? 

JOHNSON: When I was a maker, I’d put myself as an eight and a half. I actually loved what I did. I absolutely loved it — the only reason I would even deduct one-point-five points is because there were some frustrations, as I mentioned, about not being heard and not being autonomous. 

DUBNER: And then where would you put it, zero to 10, when you’d become a full-on manager? 

JOHNSON: I would say I put myself more at like a four or five — a six would be a great day.

DUBNER: Okay, that’s your personal satisfaction. I do see, however, on LinkedIn a review from your manager. He writes, “Katie is a rounded and passionate data leader with all the qualities required to inspire, manage and lead a team. Plus, she’s got brilliant I.C. skills to boot,” and he notes that you are “a real unicorn in the data analytics field.” So that sounds like you were the best manager ever. 

JOHNSON: Yeah, it’s really nice isn’t it? 

DUBNER: It’s really nice. Did he write that before or after you decided to quit? 

JOHNSON: He wrote that after. 

You heard that right. Katie Johnson quit that management job. She quit being a boss entirely. She went back to working as a data scientist, at a different firm.

JOHNSON: I don’t know if you can ever be successful at something you don’t like. I want to do something that I love, and I’m really passionate about, because that’s the only way — maybe other people are different, but I have to love it. I have to be like, on a Sunday night, “I can’t wait to start my work tomorrow and get back to what I was doing.” And I was never, ever, ever going to have that in my management job. 

DUBNER: So before you ever became a manager, as a maker, you said your average satisfaction or happiness was around eight and a half. When you became a manager, it dropped to, let’s call it, five, six on a great day. What is it now? 

JOHNSON: I’d say it’s a nine and a half now. I’m super happy. 

DUBNER: Are you getting paid less now as a data scientist than you were as a manager? 

JOHNSON: I’m getting paid more. 

DUBNER: How did that happen?

JOHNSON: I think there are more individual contributor roles now that pay good money. I think that this technical-specialist route is becoming more prominent and more rewarded — and people do realize that there are going to be a lot of people who don’t want to become the manager. And how do you motivate them?

DUBNER: I believe you looked at the Peter Principle paper, is that right? 

JOHNSON: Yeah, I did.

DUBNER: The way the Peter Principle is usually described is to me almost comical — it’s that people rise to the level of their incompetence, which I find is a bit cruel-sounding, because one could also say that people rise to their ceiling of competence, right? And then maybe they’re not as good at that; it’s not like they suddenly turn into idiots. But I am curious, just your thoughts on the notion of promotion into management as a reward for being good at what you’ve been doing all along. 

JOHNSON: For me, this is where the idea of splitting out those levels of seniority — so maybe you don’t become the manager, but you become a technical expert and you are paid and rewarded for that — is something that helps with the incentives. What I would say on that, though, is often we have this dual-career track of, “Okay, you can be a manager, or you can be a technical specialist.” But even though you might get a quote-unquote promotion and be paid more, the technical specialists still might be excluded from high-level conversations. So being a manager just has this connotation of seniority that a technical specialist doesn’t necessarily, and you still might be overlooked in terms of just the respect. And I think that is motivating more than just, “Hey, here’s a promotion, here’s a new job title.” I think people want that autonomy and that having a seat at the table, people caring what you think — it has to come with that. 

DUBNER: I would think that many people who are promoted from some sort of maker to some sort of manager — that it would be hard to step back, if for no other reason than it seems like a loss of status, yes? 

JOHNSON: It definitely feels like a loss of status. I guess for me, I’m lucky that I don’t care what people think as much as other people, or at all. 

DUBNER: I’m sure that was identified in your personality test as well. 

JOHNSON: Yeah, complete rogue — doesn’t care what others think. People judge you, which I find interesting because I don’t know anyone who likes their job as much as I do. So for people to look upon me and feel sorry for me in a sense that I have chosen to go backwards in terms of career hierarchy, it’s kind of telling in terms of what we value out of a career. 

DUBNER: And you can tell them that if you hadn’t done this, you wouldn’t be on Freakonomics Radio

JOHNSON: Exactly. I got what I wanted.

DUBNER: Was this the plan all along?

JOHNSON: Yes, it’s a big, long game.

Thanks to Katie Johnson for sharing her bossdom-and-back story; and to Kelly Shue, Steve Tadelis, Nick Bloom — and all their collaborators for trying to make this thing we call “management science” a bit more scientific. 

*      *      *

Freakonomics Radio is produced by Stitcher and Dubner Productions. This episode was produced by Ryan Kelley; we had help this week from Jeremy Johnston and Jared Hohlt. Our staff also includes Alison CraiglowGreg Rippin, Gabriel Roth, Zack Lapinski, Rebecca Lee Douglas, Morgan Levey, Julie Kanfer, Mary Diduch, Jasmin Klinger, Eleanor Osborne, Emma Tyrrell, Lyric Bowditch, Jacob Clemente, and Alina Kulman. 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 PodcastsSpotifyStitcher, or wherever you get your podcasts.

Read full Transcript

Sources

  • Katie Johnson, senior data scientist at Wagestream.
  • Steve Tadelis, professor of economics at the University of California, Berkeley Haas School of Business.
  • Kelly Shue, professor of finance at the Yale University School of Management.
  • Nick Bloom, professor of economics at Stanford University.

Resources

Extras

Episode Video

Comments