The Exchanges

Q I'm curious how you think about the application of the research that you found. So on the one hand, rational capital, there's now a human capital index that you created, and you've shown that it's outperformed. On the other hand, you have all these insights That companies could deploy to make them much better companies. How have you thought about where you want to make the most impact?

A So I think about three separate ways. One is we have an index, you can invest, you can give companies who are better to their employees more money. You would also want to make this public. You would make this public so that companies who want to raise more money would say, hey, we're treating our employees better. Give us money in the same way that companies do ESG. Let's double down on that. I also think that companies need to start changing the roles of HR. HR in most company is a procedural process. Legal function. They do the sexual misconduct and the module. They worry about procedural things. I think that HR needs to be an R&D function. They need to continuously try different things and improve how the company is running. I don't think there's one recipe for all companies. You know, everybody wants to be appreciated. People want to feel that they can make honest mistakes. All of this is true. The path to get to those might be different for different companies. And HR needs to be the function that improves things dramatically. So I think one is, there's the index. Two is, I want companies to start carrying, elevating the role of HR. And the last thing is that I'm hoping that people who want to invest seriously in a company Would do their homework, including homework on human capital before they invest. So let's say you have an activist investor, they want to buy a percenta…

AI assessment note: “So I think about three separate ways. One is we have an index”

Q So I want to turn to some closing questions, but before I do that, I'd be remiss if I didn't ask what else is exciting and new in your research?

A Wow. So my life, my professional life has three parts. I have a research lab at Duke called the Center for Advanced Hindsight, and we mostly do research on financial decision making for lower income, middle to lower income, trying to help people save, spend a little bit less, spend better, pay debt on time, and so on. We have research on health, getting people to exercise, take their medication, sleep better, And so on. Mostly in the US, some in Africa, some in Europe, and tomorrow I'm off to China. We might start something there. So that's mostly changing health habits and financial habits. That's one life. The second life I have is I have a group in Israel who is working with the Israeli government. So once a year we meet with all the officers, almost all the officers of the government. They tell us what their big problems are, and we kind of bring the behavioral economics perspective. We go ahead, we do research, And this ranges from education to land disputes, to traffic, to tax evasion. I mean, we, we cover the whole realm and that's amazingly exciting. And then the third life is that I have a few startups. And from time to time when I think I have a good idea and nobody else is picking it up, I, I do it myself. So I have now a startup that does food. We have capsules about the size of a mug. And we have a machine called Genie. And within 60 to 90, sometimes two minutes, w…

AI assessment note: “my professional life has three parts. I have a research lab at Duke”

Q Once you've done your work and ranked these companies based on whatever the model tells you in terms of the success of their human capital management, how do you then take that and construct this index?

A So we're working on our pure index. And in our pure index, we basically take the top 20% of companies we have, we don't rank them within that 20%, and we just invest equally in each of them, and we rebalance quarterly, and that's what gives the excess return that we talked about of slightly more than six percent a year. We have worked with some partners That have other requirements as well. So one partner, for example, said that they do want to keep the balance of the different sectors. So in that case, what we do is we don't say, here's a 20%, but we basically use their constraints as well. So we say, let's take each sector, and within each sector, we'll give you the top companies. If somebody came and they said, oh, but we want another ESG led. We could do that, but our pure index, which is both interesting and important because it tells you What is the signal? Now you can decide that it's not the only signal in the world that you want to combine it with something else, which is perfectly fine, but the pure signal is let's take all the companies, rank them by human capital, take the top 20%, equally weight them, rebalance once a quarter, and we want to make sure that that gives high returns. Now, after that, you could do other things with it to take other things into account.

AI assessment note: “we basically take the top 20% of companies... invest equally in each of them”

Q You've done experiments in all kinds of things. Where do you start? And what's that process to get to some kind of research driven result?

A Many times it can start at almost a random place about curiosity. So the first thing that needs to happen is, is some curiosity, and that curiosity can happen from, like, I see something bizarre, and I say, why does that happen? So an example for this is that when I was still in hospital, we had the ration of how many painkillers we could get. We would get six shots of morphine a day. That was the Our limit. And, you know, of course, me and all the other patients, we would track how many injections I got, how many I have left, want to make sure I have enough for tonight to sleep, and because I'm a little obsessive, I would count my medications, but I would also track other people's medications, just to make sure they're not getting too much. And one night, the guy next to me, Ronnie, screams at night, and he asks for a painkiller, and the nurse goes to his room, and Silence, and he goes to sleep. And I knew he got his seven injection. So I called and I said, what's going on here? I want another injection too. And she said, no, no, she just gave him placebo. She just gave him IV fluid. And, you know, it's, it's one thing to read about placebos. It's another thing to experience pain, pain from burns, and to know that the person next to you is experiencing similar pain, and to know that they are quietly going to sleep after that. So that's an example of something that I experience…

AI assessment note: “Many times it can start at almost a random place about curiosity.”

Q What is the index? Is it the narrowness of the gap leads to outperformance?

A We take the 20% companies who have the narrowest gap between men and women. And we said, what if we invested in those 20%? Imagine you take the S&P, And you say, let's take all the companies in the S&P we have data for, and we will take the 20% companies that the gap between men and women is the lowest. Now, it's, there are very few companies where the gap is zero, but that creates a real substantial return. Now, how do you deal with these two seemingly opposing facts? You look at our approach, we find big returns to treating women well, You look at the Xi index, it looks like a negative effect. What's the truth here? How do you look at it? And it turns out that, and this is really important, that often we measure what's easy in terms of what's important. And if you think about the SHE index, nobody, nobody in their right mind would think that if you assign more women to the board, that by itself would change a company. It's a perfectly great first step, but it wouldn't change the company. What people think is that this would lead to more equality of women down the line. But it turns out that if you have a check the box approach, it doesn't because you stop at that. So I want to make two points. The first one is that we need to measure what we really want to measure. And when we create proxies to remember that we make this assumption. So we say, okay, we're measuring women at t…

AI assessment note: “We take the 20% companies who have the narrowest gap between men and women.”

Q How do you think about the refreshing of this data in terms of frequency and how dynamic or stable is it?

A Yeah. So in general, what we find is that this data has a half lifetime of about 18 months. And it's actually what you want, right? You want the data to be valid, but you want the value of it to go down. Why? Um, I think about it a little bit like rain and coffee. At some point, the coffee reveals itself. You can see how much coffee beans there are, but the rain gives you a prediction in advance of what will happen, and motivation is the same thing. Human capital is the same thing. At some point, these things will be revealed. At some point, human capital would mean that there's better products, new procedures, all kinds of things that happen, and at some point, other people should be able to view it and make judgment. So you want that data, you want that usefulness to have a limited amount of time. It's the question of what's the cycle of human creativity and so on. I think a year and a half is about right. So with that in mind, I think that refreshing it every year, Is probably sufficient. If I could refresh every half a year, that would probably be better. Right now we refresh once a year.

AI assessment note: “Right now we refresh once a year.”

Q You also mentioned with Glassdoor, which we're all familiar with, you have to kind of know what you're looking for. What's an example of something that you could take the data from something available like that and know what you're looking for?

A So I'll give you an example. One of the things we find is that it's not about just the overall level of, let's say, feeling appreciated. So we find that feeling appreciated is a very good signal. But it's not just the overall level of being appreciated. It's the difference between how management and rank and file employees feel about being appreciated. We call this alignment. And companies that the gap between how they treat management and how they treat employee is very large are companies that are actually less productive in the stock market. Because there's a gap, and we mentioned the issue of relativity. Relative pay is more important. Relative fairness is more important. Fairness is all about relativity. The same thing is true about employees versus management. So if you look at Glassdoor, and you want to understand better the information, it is really good to try and separate the role that people have. And if you find that people are complaining equally at management and at the level of the employee, this is not as a bad signal as if management is not complaining and employees are complaining much more. So that's an example for what to look at.

AI assessment note: “difference between how management and rank and file employees feel about being appreciated”

Q Do you have either a favorite experiment you've done or a favorite finding that came out of it?

A It changes By the day, depending on what I'm thinking about and so on, but if I think about what did I, for me, find that was the most deeply interesting for a long time, I think it was some of our findings on placebos. Placebos really work, right? People don't recognize that placebo really work, and placebo is basically the idea that the mind can change our physiology and change our physiology In, in a way that solves the problem. So, the Pygmalion effect is one example. You take teachers, and you say to them, half of these kids are just not that good. The kids sitting here on the left are just not that smart. The kids sitting on the right are really smart and intelligent, and then you wait, and at the end of the semester, you find exactly what you told them, even though you told them random, you randomly assigned kids to that. Not very ethical, but, but it works, and, and that's because what happened is the teachers act on their beliefs. Right? So a kid who is supposed to be not that smart says something, and they read into it, not being so smart, oh, it's a mistake. The kid who is very smart, they say, oh, maybe there's something in there, encourage them more, and so on. Another part of the placebo, the self-fulfilling prophecy, it's a little bit like Pavlov's dog. So if you remember the story with Pavlov's dog, you have bell meets saliva, bell meets saliva, bell meets saliv…

AI assessment note: “I think it was some of our findings on placebos.”

Q What teaching from your parents has most stayed with you?

A My mother was very trusting. She told me that she doesn't like me to take, like, an hour off from school here and there when I need a break, but she said when I need a whole day off, she'll write a note to the teacher, and I could take the time to do something useful. So, it was an interesting thing where she said, when you need time off, let me know, and as long as you use the time well, I'll let you have that. So that was, I think this trust was something very Very important to me. And my father was always not just a father, was also a very good friend. I felt I could talk to him about everything. And that type of conversation has stayed with me with other people as well.

AI assessment note: “that type of conversation has stayed with me with other people as well.”

Q How do you think about meaningful changes at the individual company levels? You could think about things like a new management team comes into play or a transaction and a merger. How do you go about factoring those into the pace of change for human capital?

A So both of those can be incredibly useful, incredibly destructive to companies, and the decision we've made is that when the company's nature changes very dramatically, we no longer think that we have data about that company. So imagine company X, and we have all the data from this company in January, and then in February they were acquired. What can we say about that company? Or they acquired another company. If we don't think that we can say something about the underlying culture of the company, we basically say we don't know what to do with them anymore. And the same thing can happen for a big change in management. Recently, a big company came to me and said that they heard rumors that somebody doesn't like their culture. And they're trying to acquire them. They came to me and said, oh, before somebody acquires us, we want to try and change our culture internally. That's great. Now, it was in a different hat that they asked me to do this, but if a company is trying to change something radically, Then we have to say that our data is less good, not as accurate, and I would not have as much faith in this. So I would say, if a company just changed management in a dramatic way, merger in some way, we don't know about them, let's give the new management on the new merger some time, get the data again, and see what can we say about them.

AI assessment note: “when the company's nature changes very dramatically, we no longer think that we have data”

Q been a little while. And I would love to start with when you first came on the show, we talked a bunch about your research and some of those incredible stories and insights. And then we also started talking about what was the beginning formation of a rational capital and this model you had built. So what's happened in the last couple of years with the business and the model?

A So first of all, just kind of to remind all of us, we started on this journey about four years ago to try and quantify human capital, and it started from my basic research on human motivation, and we moved toward understanding how companies treat their employees, how employees feel about the company, and what impact this has on the performance of the company in the stock market. And we had some really good, interesting results suggesting that it's a really wonderful place to look for excess returns. And when we talked last, we, we found all kinds of interesting things that fit with social science, right? So we found that overall level of compensation doesn't matter so much. Fairness, perception of fairness in compensation matters a lot. So we found lots of these nuggets and the argument was that it would be Incredibly important to, to invest based on that. How could you not? And one of the things that puzzles me is that if you gave people that information, how many people would say, oh yes, I, I want to ignore this. I don't think that human capital is important. I don't think employee motivation is important. Of course, it has to be, it has to be crucial. Since then, we've done a couple of things. We've expanded our data, one, and We did a huge study during COVID to understand what's going on during COVID. We tried to understand what does this have to do with ESG, and we tried …

AI assessment note: “Since then, we've done a couple of things. We've expanded our data”

Q So when it comes to creating investment products out of this research, How do you decide if you found causation?

A We use everything we can in terms of a good statistical knowledge, and particularly for these kind of things, we use models with a delay, so what you want to do is you want to say, oh, it's not just a correlation, it's causal, right? What does it mean causal? One thing happened before the other one. If they happen in the same time, you might say, oh, I don't know which one is causing which. Most of the things we have are not quick Right? It's not, oh, there was a press announcement, and now something happened. So we work with models that have a delay in them, and we try to estimate the delay, and if you got in a model like this, the delay was reversed, right? You said, oh, first the company made money, and then they make the environment safer. Then you would say, oh, it's not causal. So that's one part of it. We also look at what we know from research. So we didn't approach all AT questions with equal belief about the work, right? We have some things For example, sense of purpose, right? We know that the sense of purpose works for all kinds of other things. Finding itself with purpose works is not very surprising in some way. And then we also look at different dimensions. So for example, the dimension of benefits, that one changes much more as a result of how companies do financially rather as a lead indicator. And actually in our data, we don't see much of an effect for that. …

AI assessment note: “we use models with a delay, so what you want to do is”

Q Do you find that the companies that treat their gender segmentation the same, so would make it into the gender index, that also would make it into the true index because they're doing well on both sides, do better than the ones who treat people the same, but they're sort of equal opportunity offenders?

A I haven't looked at that. I think you could basically say there are all these companies, right? The companies who make it to the regular index and don't, companies make it to the gender index and not, and we can look at the four quadrants of this. My guess, and I'm just guessing here because I haven't done this analysis, my guess is that the companies who are doing good on both are, are much better because, you know, these things, the thing about treating people well, it doesn't happen automatically. This is, comes out of intention and care, and it comes with, for example, feeling appreciated. People have to have a different approach to management. Reducing bureaucracy has to come with, like, real trust of people. It's effortful to figure out how to treat people well. It's not magical because a lot of people's intuitions are wrong about what motivates people. Right? People say, oh, let's just get more benefit, and let's increase average salary, and let's get better coffee. The things that you really want to do are different than that, and people need to work hard at that, and my guess is the companies who are working hard on this are, are doing it across the board. Just that there's some people who focus more on gender, and some people focus more on everybody, and so there's some differences as well.

AI assessment note: “I haven't looked at that. My guess is that the companies who are doing good on both are much better”

Q And so these different, many different metrics that you've used, are they difficult data sets to attain?

A Yes. Almost all of them are basically a proprietary data set. I think the good news Is that each company that wants to do it can relatively easy start an internal process of improving human capital, right? And I think step one is to put HR. HR, I think, is maybe the worst, least reputable department in almost every company. Maybe compliance is worse, but, but you know, it's a hierarchy of the company. HR, rather than Being considered the R&D arm of the company is considered a bureaucracy that people don't like. But I think the good news is that I think there's lots of steps that companies can take internally. As investors, we have this data and we're using it, but it's tougher to get. And it also represents this approach, I think, of investing by getting outside data. So, you know, there's some people who are investing in agriculture based on satellite images. To some degree, this is the same thing.

AI assessment note: “Yes. Almost all of them are basically a proprietary data set.”