Marius Guenzel discusses his research on CEO Stress, Aging, and Death, a study co-authored with Ulrike Malmendier. He explains that high-pressure leadership roles come with real and measurable health costs. The study used machine learning to track facial aging and showed that heavy job demands, especially during crises like the Great Recession, can accelerate aging by roughly a full year over a decade. Guenzel emphasizes how emotionally difficult responsibilities, such as layoffs, are key drivers of stress, and encourages future leaders and researchers to prioritize balance, empathy, and the human side of work.
Adriana Macias, O’Donnell Center for Behavioral Economics: Can you give us a brief introduction to the research paper on CEO, Aging, and Death? Tell us a little about it.
Marius Guenzel, The Wharton School, University of Pennsylvania:
Absolutely. What we’re interested in is understanding how the CEO role is influenced not just by the fame, money, and the success story behind it, but also by how job demands and the stress that comes with that job affects them as people, as human beings. The one-sentence summary of the paper is in the title: CEO Stress, Aging, and Death. Meaning, CEOs have very stressful jobs and the more stress they experience, the worse their health outcomes, both in the aging process and, ultimately, in longevity. We find that when a CEO has higher job demands, they age faster and die younger, which is a stark outcome.
Adriana: Definitely. Can you tell me a little more about what got you interested in this topic? Was there a particular CEO story or event that led you to this research?
Marius:
Sometimes we hear in the media stories where a CEO has a heart attack and maybe even dies in office. These are the salient examples that stick with us, and they’re reported in the news because CEOs are responsible for entire companies and workforces. We were interested in asking whether these are just outlier stories or whether there’s a more systematic pattern, maybe not necessarily people dying in office of a heart attack, but a systematic health pattern.
Beyond that initial hook, this emphasizes the human aspect of a job and of economic research more broadly. Often, economists are interested in consumption or GDP growth, which are important but less personal. Even in behavioral economics, some of the outcomes can feel intangible, like framing effects or defaults on people’s behavior, for instance. This question of the CEO felt very human, and I was excited to work on it.
Adriana: The human aspect of research is important, as we hope that research is applied eventually. One thing that stood out to me was the “no pay trade-off” and the stress effects of that on CEOs. What does that say about the market and human well-being as a whole?
Marius:
That was something we were very interested in. Once we establish that there are health costs to being a CEO (costs of success), the next question is whether that’s priced. Are CEOs aware of these costs, and do they demand higher pay when the job is more demanding and stressful? We find that it’s hard to measure, but with appropriate measurement caveats, we find no evidence of compensation for a more stressful job. That’s striking. It suggests either people aren’t aware of the health consequences of their jobs, or they don’t care, or maybe some of both.
Adriana: You also used machine learning in the analysis. Can you tell me more about that aspect?
Marius:
Definitely. If you want to study health outcomes, including in the CEO context, measurement is the key challenge. People have looked at mortality effects before because we have good data, but there’s a familiar image in people’s minds: Think of U.S. presidents or any president, and how they looked when they entered office versus when they left office. That “before/after” look, like Obama’s hair getting grayer, for example; this inspired us to measure aging effects through facial images.You can do this with a handful of photos. Humans can rate perceived age, but for thousands of images you need a scalable approach, and that’s where machine learning comes into the picture. With advances over the last decade or so, the machine can really learn to associate facial features with perceived age. We applied machine learning to estimate how old a person looks and how fast they aged over time using multiple images across a CEO’s tenure.
Adriana: Did you follow individuals during periods of less stress versus more stress and how did that show up?
Marius:
We do follow individuals even after they step down as CEO, because the health effects we study are visible in the long run, not just “dying in office” events. What we didn’t look at is subsequent periods of less stress at the individual level. That’s hard to identify because you’d need multiple exogenous events per person. We focused on a clear event: the Great Recession, when job demands were heightened, and there’s a lot going on in the firm. Further tests when stress subsides, that would be much harder to identify at a person-to-person level. It’s interesting and important to learn how to manage stress, but difficult to measure empirically.
Adriana: That makes sense. It makes me wonder: if you’re tracking how CEOs age, what happens when they have periods of less stress? Could machine learning help understand that, maybe even to help manage stress?
Marius:
There’s a lot of research we hope to inspire when it comes to managing stress. Or, what are the specific mechanisms through which stress materializes? So, for example, stress can lead you to stop going to the gym, be less healthy, even have your mental health affected, and so on. But how can we measure not just stress and the long-run health or longevity consequences, but also the mechanism consequences in terms of diet, exercise, social interactions, smoking, drinking, etc.? That could help us identify when stressful moments subside (e.g., when you return to the gym, when you start eating healthy again). That would be interesting to study, including using machine learning techniques; for example, dieting, tracking body composition from images, etc. But that’s for future work.
Adriana: If you ever do that research, let me know, I’d be curious. Can you tell us, from this research, what surprised you the most?
Marius:
One thing that certainly surprised us was that the effects we found were quite large. It’s not like it’s going to cut life-expectancy short by a day. To give some perspective and some concrete numbers, we find that if your firm and industry were hit hard during the Great Recession, there’s a marked increase in job demands, you age about an extra year over the following five to ten years. Aging an extra year within a decade is a substantial effect. Starting from the baseline assumption that CEOs have brilliant careers and make a lot of money, finding there are not just some health effects, but substantial ones, was surprising. That also opens the question: once we’ve established that, who are the people who still want to become a CEO? Who makes it to the top of the firm? What’s the CEO selection look like? Who wants to opt out-maybe because it’s too stressful for them or their families? We didn’t study that directly, but it would be interesting for future work as well.
Adriana: I wondered about that, too. It’s definitely something to examine.
Marius:
Absolutely. We’ve been inspired by real-life documentaries from different CEOs across countries, where some CEOs realize, often in hindsight, that although they made it to the top, their marriages didn’t last, or they missed their kid’s first day of school, or things that felt okay in the moment because they rationalized, “I’ll have many years to do xyz,” but later became missed opportunities. People who can realize this ex ante, they might be happier, even if that means not becoming CEO.
Adriana: The research certainly gives perspective. How do you think the CEO job can be designed to be less stressful?
Marius:
One idea is to have more of a team structure at the top splitting responsibilities, so it’s not 120-hour weeks for one person. Or, delegate responsibility with clearer support systems. It would also help to define what “job demands” actually mean and what specifically makes the role stressful, as well as how we can define the CEO role so it becomes less stressful. That’s another way to tackle the costs of success, rather than only focusing on the benefits.
Adriana: Could you speak to the stress level at the top and the ripple effects across the organization? We’re seeing it with the CEO, but what does it mean for the rest of the organization?
Marius:
One thing I want to make sure I get across: Even though we look at the CEO’s health costs, I’m not saying we should feel sorry for CEOs. They’re highly paid, high-status individuals. There are many other jobs both within organizations and outside of large firms that may have even higher stress outcomes. For example, surgeons, air-traffic controllers, or even food delivery workers dealing with constant hassles. Across the population, there are professions that come with the highest stress level. We’re not claiming CEOs have the most stress, but we are saying that when CEOs, even when they have all the resources they need, when they have more stress relative to other CEOs who have less stress, the health consequences are enormous, and that’s the real takeaway.
Adriana: Absolutely, we clearly see that in the research. So, where does this research go next?
Marius:
One point we are working on right now and that I alluded to briefly earlier is that we’re working on identifying which job aspects are most stressful for CEOs. We’ve established that industry crises, like financial crises, certainly increase stress levels and are measurable through these health outcomes. But what is it about the crisis? Is it that there’s a lot of pressure from shareholders? From banks? From other stakeholders? Communities? What is it?
One answer inspired by the CEO interviews that we established in a follow-up paper focuses on when CEOs have to lay off employees. Again, it emphasizes this human aspect–not just the human cost of the job in terms of health consequences, but also the reason for the cost is the decision to let someone go and to fire someone, not because of poor performance but because the crisis necessitates layoffs, and that really seems to anecdotally stick with CEOs. It takes a toll on them. That’s something we empirically found when we look at the long-run health outcomes triggered by a crisis. These effects are stronger when, during a crisis, the firm had to make layoffs. It’s really this employee focus that seems to have a large effect on CEOs.
Adriana: Do you think this has changed your personal view on leadership?
Marius:
I’d say that the common way to describe a leadership position is something to pursue, and emphasizing the positive aspects is important. Those may still outweigh the negatives, but for me it clarified that leadership doesn’t come without costs. Often, our broader discourse can be black-and-white and this is certainly gray. You might get rich and have impact, but you should also consider the downsides and that is often missing from the discussion.
Adriana: In an ideal world, what would the perfect profile of a CEO look like?
Marius:
If I were to choose a CEO, I would want them to be empathetic with the workforce. I think there are potential benefits that can help attract better talent and improve morale as a result. So, there are positive effects if a CEO truly cares about employees. I’m not saying we should hire people who feel nothing when layoffs happen. Rather, we should design institutions that reduce the pain for both workers and leadership during hard times. That way, we still select the right people into top positions while alleviating the burdens of the role.
Adriana: Can you speak to the gender differences?
Marius:
Certainly. Empirically, whenever you work with CEOs, you end up with a sample skewed to white and male (at least for now, maybe that will change in the future). That’s why we can’t separate effects by gender or race/ethnicity in this paper. It does raise the questions: Is there a component in terms of the selection for the people who want to make it to the top are overconfident about the health consequences that might be stronger in men than women? We know men tend to be more overconfident in general; perhaps that extends to their health potential. This is speculative, so it could help us give some more color on the selection. Are there more consequences in women than men by different races or ethnicity? Maybe once we have a more diverse sample in the future, we can study these differences directly.
Adriana: Any advice for younger researchers who share a passion and broader vision in economics? And any advice for future CEOs?
Marius:
For younger researchers across different fields in economics and finance, I always think it’s important to study things you truly care about and are passionate about. If we hadn’t cared about manager health, we might not have thought to use AI tools on facial aging. Pick a field that you are truly passionate about and that you can identify something that’s missing from that field.
Another aspect that is also important to take into account, especially in light of our research on health consequences of being a CEO, but I think it extends to many professions, is that our academic job is still a job. Don’t let your identity hinge entirely on publications, etc. Keep balance, hobbies, social interactions. Keep the distinction between work and life, and that’s also my advice for CEOs. It’s easy to identify completely with the company, and make your whole life about success, and forget the private side of life that matters just as much, if not more.