Host Intro (Gene Kim)

So I am so excited about our next guest. He is Dr. Joe Davis, Global Chief Economist and Global Head of the Investment Strategy Group for Vanguard. For those of you who were at ETLS Las Vegas, you might have seen Devlin McConnell's talk, where he showed clips of this incredible session that Dr. Davis presented at their internal technology conference, which I had the pleasure of attending earlier this year.

Vanguard, for those of you who don't know, popularized the index fund in the 1970s and now manages over $9.5 trillion in assets for over 15 million clients with a mission to help investors achieve their financial goals. Dr. Davis is responsible for the firm's internal and external economic and market outlook.

His talk on the potential economic impacts of AI blew me away. My takeaway from his presentation and his writings is that he holds a minority view that AI has the potential to significantly boost global productivity, potentially even addressing key structural concerns that have challenged the economy for the past 15 to 30 years. Over the last decade, he has regularly spoken alongside former Fed chair Janet Yellen, Fed chair Ben Bernanke, and so forth. So I'm so delighted that he can share some of his learnings and opinions with you here today. Joe, I'm so happy to see you again.

Q&A

Gene Kim: Maybe we can start off by asking: what does a Chief Economist do?

Dr. Joe Davis: That's a good question. I think some of my colleagues probably wonder as well. There are really three primary responsibilities. One is what you would expect: to really get some sense of where the risks are to the economy, whether it's the United States, China, or other markets around the world. It involves a lot of analytical work and forecasting, like what GDP growth or inflation will be. Sometimes advisors care about that because it has implications for the financial markets, the stock market, or active managers who care about what the Federal Reserve may do, which will affect portfolio management.

Secondly, I lead a research team here. We're nearly 80 employees around the world. It's not just looking at short-term events, but big picture, longer run. That's where I got into AI: what could the implications be and how may it change growth and the financial markets?

Thirdly, it's a combination of those two. I work at Vanguard, so we provide financial products for people to invest and save for whatever goals they may have, and also ask how they should put together a portfolio of stocks, bonds, and other investments. The economy and the financial markets, how they may perform in the future in a range of outcomes, they're tied at the hip. That has implications for the portfolio, whether how successful it'll be, or perhaps you should consider another one. It's really all those three together.

You hear about the first one in financial media, like what was the jobs report, and that's important. It's the second and third ones that I think have more enduring value, particularly for investors and for everyone on the call, if they're thinking about what it may mean for their career.

Gene Kim: In your talk and in our conversations, you said something to me that now seems so obvious, but I'm embarrassed that I hadn't internalized it until you mentioned it, which was that there was this inextricable link: productivity is closely linked to technology advances. Can you say more on that? Why is that important to you?

Dr. Joe Davis: Why should we care? We hear a lot about economic statistics in the near term, but I can tell you for societies, for us as parents and employees, what really matters is productivity growth. You can think about that as technological change. It's output per person, and how much more output we are able to generate per year for the same unit of time. Why that matters is that ultimately drives standards of living.

It explains why some countries such as the United States are wealthier than other countries. It's because the productivity advances over time have just been more powerful. That's been the story since the Industrial Revolution two or three hundred years ago, and will be true going forward.

The big question has been in the United States, for example, and in other markets, we have technology all around us. Yet productivity as conventionally measured has been really low since the Global Financial Crisis through 2007. That's 15 years. That's like half or a third of a working career. Why is it low? And secondly, is it at all going to change? What role could AI play in it? That's why we should care.

Gene Kim: Can you talk a little bit more about that graph? When you first corresponded, I mentioned to you that one graph you showed. It was so startling to me because I've seen lots of graphs on GDP and productivity, but I've never seen one like that. Specifically, the fact that deviation from GDP growth per worker has been negative since 2007. Why is that so surprising and startling?

Dr. Joe Davis: It's below average, and this gets to where are we going as an economy, and where is the world going. This is an innovative framework we're using. We believe we're the first institution or research team, academic or otherwise, to actually look through this lens. We all know that productivity growth hasn't been improving as quickly as it used to. The question is why.

The chart you saw at the conference was a decomposition explaining the factors that are keeping trend growth output per person low. I'm netting out the little ups and downs day to day and asking, what is the trend? It used to be really high for decades, then it slowed in the seventies. It bounced up with the personal computer. We all recall the late nineties. Since the Global Financial Crisis, it's been plateaued. No one's been able to explain why.

What you saw there was not only that decomposition; it was the fact that it's the lack of technological progress. Specifically, there are three dimensions we measure, and we're the first ones to do it. One, we're getting better at what we do: think of a power tool like a computer, an augmentation of our work. Secondly is automation: we don't have to do stuff, we don't have to spend time doing it. Those are two different forms; the computer can do some of each of them. The third one is new technologies emerging that could power growth. It used to be the automobile in the 1920s, or electricity. It was the personal computer and the internet.

So AI: the question is, that's the picture. It's been really low. Most economists when you talk to them outside of Silicon Valley are expecting that line to remain low, which is the foundation of every forecast that I've seen pretty much on the planet: the Federal Reserve, the Congressional Budget Office, just because they have to plan for long-term growth. The question is, can you predict the change, and what direction will it go?

Gene Kim: It is not a surprise to you that I was surprised by that graph, because I had never seen a representation of that, and yet it had a lot of explanatory power for me. It seemed to explain why there is some different dynamic going on versus the period before, which you mentioned as the 1980s, where there were definitely productivity gains being had. Am I saying that correctly?

Dr. Joe Davis: You're right. I'm not expecting many people on the call to follow GDP statistically. I have to do it; it's part of my job. But output per person ultimately drives sustained wage growth. That's the paradox we live in: we have technology all around us. My kids never separate from their smartphone, and yet you don't have it propelling faster growth.

What I found is the paradox: we actually don't have enough automation in society. That's the factor showing up in the chart you mentioned. It's striking, and I'm talking outside of manufacturing, outside even the IT profession. It's the broad service economy. Eighty percent of the jobs in the United States are service economy: retail, business services, financial services, Vanguard, education, and healthcare. We're working really hard. I'm not saying people aren't working hard, but we haven't had the technology to accelerate, where I can do two times the volume that I used to do.

That's the sort of power. We saw some of that for a time with the personal computer. Just imagine all of us trying to do our job without a personal computer. But everyone's using it now, and that's kind of plateaued. The question is, is technology going to come around? The consensus view is that AI is here, but the general view is it's going to be fairly marginal, which is why economic growth projections haven't really changed materially in five or 10 years.

Gene Kim: In your talk, you mentioned the work of Dr. Daniel Rock, who came up earlier today with Professor Dr. Matt Beane. You mentioned the distinguishing factor. He framed it as the famous paper, "Are GPTs GPTs?" Are GPTs general purpose technologies? You use that to distinguish a mere automation versus those that augment. Can you talk about the distinction and why you think it's so important?

Dr. Joe Davis: It is really important. You want both. Automation gets a little bit of a bad name, although if you're in manufacturing in the United States over the past 50 years, you have felt it. I don't want to sugarcoat it. Automation can sound really scary. The fact is, though, more often than not, most of our jobs, we don't do things as manually as we used to do. Jobs change more than they disappear.

The question going forward is how much AI is going to change that arithmetic. There are really two broad spectrums. There are some who view AI as marginal and we're not going to change much. That's the economist consensus. I'm being unfair to my profession, but most people are in there. It's not that we're not growing, but it's the status quo.

Then the other extreme is AI is going to be dystopian, which means massive automation because you don't need a human in the loop. If you've used ChatGPT or code co-chairing and all that, and you play that forward, it's going to eat a lot of jobs.

What our analysis and research says is definitely some jobs will be automated away, and certain tasks, but more of the jobs will benefit through augmentation. I will bottom line this: AI, in our simulations, and we're modeling this, this is not my personal view that I'm just throwing out there. We are modeling technological change, and dare I say, we're able to predict to some extent the magnitude and the direction of change five to seven years in advance, based upon some of the theories Daniel Rock and others have done. We're building upon academic, really good science. What I'm telling you is that more likely than not, AI is going to be more transformative than the personal computer and the internet.

Gene Kim: I find that really exciting. To put context on that, the whole notion of whether our kids are going to have a better life than us: if you're sufficiently pessimistic, the answer is no. We love seeing evidence that there is actually a brighter potential future. As a community of technology leaders, we want to move the needle on important goals of the business. To say there are ways we can materially impact that is an important thing for us to factor into decisions.

Dr. Joe Davis: Just to amplify that, because Professor Rock is great. Similar to him, we looked at what AI could potentially do to every job description, every task within every job. Think of your own job description. We look at AI intensity across each one of those dimensions, and then play forward across all occupations.

What I don't want to say is that it will be transformative in growth effects but not disruptive. It's not going to be dystopian in our view, unless we get into how it could be, but unlikely right now for the next 10 years. However, we will very likely see the greatest change in work in probably 50 years. I'm talking for my job. For every IT job code that we have looked at, there's significant change. I'm not saying there's massive job loss, but what we spend our time on versus what we use technology for, by our estimates, the mix of time will change. For more professions than not, they will benefit from that change, but it'll be disruptive.

Just think of when the personal computer first came along when I was younger, where I looked at it like, do I have to use this? What about Excel? Do I need this? I don't want to dismiss it. There's a lot of change. Roughly 20% of jobs will have significant job loss because the automation intensity is above 50%. When you're above 50%, the number of jobs go away over time. But is it beneficial for society on net? Yes. Like any form of technological change, there will be winners and losers from an occupation perspective.

Gene Kim: A cynical read of the AI jobs report that Rock and peers did is that the only jobs left over are floor sanders, food service workers, and a couple others. Knowledge workers disappear, of course. That's a...

Dr. Joe Davis: That's closer to the dystopian. How you get dystopia, and I'm not critical of Professor Rock's work, a lot of this is intensity. For us to get massive job loss, one of two things has to happen. Either you get general artificial intelligence and not just GenAI. You get singularity. It can reason exactly like a human being. AI is moving fast. I just don't know if in the next 10 years you're going to get that far.

And/or AI merges with robotics, because now you get physical. You mentioned the professions that are safe. A lot of them involve physical movement. Think of a nurse in a hospital. That's a lot of complex movement. You're moving a patient around, things such as that. Or even package delivery: how do you get the package from the car, which can be autonomous, to the front door? You need robotics. If those two would merge and you have close to a sentient being, then yeah, I don't know what any of us are doing. But we're certainly not there yet. Four out of five US workers are not even using artificial intelligence in their job, at least not consciously. I think we have some time before we rush to judgment.

Gene Kim: To put the technology profession front and center, when we talked I mentioned the work of Dr. Eric Meijer, who I've brought up three or four times today. He said we may be the last generation of coders, and if that's the case, let's have fun doing it. His view of the future is that he doesn't want to write code manually anymore. He's one of the pioneers of programming languages, and he said time can be neither saved nor created. This is the best gift that technology can give us. I thought that was a beautiful depiction.

Dr. Joe Davis: I think that's one of the higher automation scores in our work. We're coming at it two ways. We're looking at every job occupation. Our system that you saw, those graphs, is projecting the entire economy over the past hundred years. That's what gives me confidence that we're going to see some significant change.

But a lot of these skills are still valued. There's a difference between time spent on something and skills valued. I see, for more jobs, humans still in the loop, but you've got to get time savings. If we don't have that, then we're doing the same job that we were doing five or seven years ago, which means our pay is not going up. I don't want to sugarcoat it. There is some change. I was just out in Silicon Valley last week, so I've seen it up close and personal.

Gene Kim: I shared with you last time that I've fallen in love with the way labor economists talk. There seems to be something in the precision they can frame things. Two questions: why are labor economists showing up so much in the AI discussion? And are there things that, as technology leaders, we should be learning from labor economists in ways to articulate important concepts?

Dr. Joe Davis: When we talk about technology, what's always been true and will continue to be true is that at the end of the day, it's always been both a battle as well as a complement of our skills as human beings, the value we can bring to a company, and the tools we use, going back to fire, I suppose, but the technology we use to get the job done. Labor economists focus at the micro level, the individual worker level.

Technology is particularly important because you can look at job descriptions. The crazy thing is, if you get a decent handle on what a technology can do to certain tasks, think about every line in a job description: if you're a crane operator at a construction site, that's moving heavy equipment, overseeing, and so on. If you think of those verbs in the job description, and you get some sense of what technology can do to each one of those tasks, you can actually see into the future with a decent degree of accuracy.

That's why our model is able to pick up in 1910 that electricity is going to transform the economy in the twenties. It knows in 1989 that technology... It doesn't know the personal computer, what it is. It's picking up those signals that in the late nineties we're going to have a personal computer, which had been around for some time, and the internet is going to lift growth for a time being. It also tells you which occupations are going to benefit and others that are going to go through more disruption. Again, at a broad brush, it doesn't tell you the date, but it tells you the direction.

It's really coming from labor economists who look at the tasks and the nature of technology. Some technology is good for your job, Gene. It may be a little bit more substitute for mine. This is all around complements and substitutes.

Gene Kim: Complements and substitutes. That's all this is?

Dr. Joe Davis: That's all this is. Now, we are all making more money than, say, two generations ago for the same job, because we've had both complements in our work as well as substitutes. Manufacturing has seen this. You go to an automobile plant, there are not nearly as many workers as there used to be 50 years ago. Yet the average auto worker will make more money than they did 50 years ago. I'm not Pollyannic. There's been job loss through robotics and trade and so forth, but that's always been this horse race. You look to labor to get some sense: how is AI going to change my job, and is it going to hurt more? Is AI going to be marginal like social media, or is it going to be transformative?

Gene Kim: I would be remiss if I didn't ask, can you give a brief definition of complements and substitutes?

Dr. Joe Davis: A complement means I do better at my job. I think everyone on this call has benefited from the personal computer. There's probably more IT professionals in the world than there were before the personal computer. Economists have benefited. The personal computer has been a complement. I can do better research, get better insights, do more per unit of time, generate more reports or forecasts. I could have done the same thing with graph paper and a pencil, but I'm way more productive. Usually when that happens, the number of jobs go up, even though I personally can get more done during the day.

Substitutes say, I don't have to do that manually. Data entry at Vanguard has been automated for 20 years. We get data from APIs, from data providers. There used to be people hand coding in the price of our funds that you would read in the Wall Street Journal. We don't do that anymore, and a lot of firms haven't done that for years. That was a substitute. Some of those jobs disappeared. Others spent more time doing other work. Our fund accountants used to price funds manually with calculators 25 years ago. Today, they don't do that. They're looking at fraud and detection of patterns, and it's much more analytical.

Some things were substituted by the computer, yet they're doing more work complemented by the computer. The nature of that job changed in 15 or 20 years. That's what's going to happen, more often than not, in most occupations with AI. The time spent will shift. In IT for sure, less time coding. It does not mean that IT professionals will be less valuable. It means it may be more on the generation of... I am not an IT professional, but looking at the job codes, I can tell you that's where it points.

Gene Kim: One of the things I found so exciting about your talk was that you're talking about knock-on effects, whether it's electrification, the assembly line, or the personal computer. There is something about these general purpose technologies that create huge changes in the economy. Can you talk about what those are?

Dr. Joe Davis: That's going to be the open question. The automation we're going to get, and some augmentation, like we may be able to write better code, fewer bugs, in a shorter amount of time because AI will be doing some of it for us. But how you really transform economies is you have to change daily life. Technology innovation changes how we work, is how I think about it. The other side of the coin is transformation. That's different. That changes how we live.

Electricity is a simple question. There are knock-on effects. Without electricity, in the United States, we don't have the entertainment industry. Electricity didn't create the motion picture in the 1920s. Editing was involved in both. But without electricity, there's nothing to plug into the wall. We're not having air conditioners run the movie theaters. People wouldn't go to movie theaters before air conditioning in the summer. And you didn't have the motion picture industry. Electricity didn't create it, but it enabled other things.

Personal computer, for example, and the internet: think of what that has enabled. We don't have e-commerce without the computer and the internet. The internet didn't create e-commerce. So the question is, what platform is AI going to be to enable new products, new insights, and dare I say new industries? Healthcare comes to mind. We haven't seen a lot of evidence today, but does it help with drug discovery? Does it transform education in terms of a different way? And then there are other things I just can't think about.

It's a platform. The human mind is still going to be involved, but if you take that technology out of the loop, I don't think we can do even current genetics research without personal computer technology. Human creativity is going to have to come in here, and it's going to have to build on AI. That's the open question: are you going to get closer to electricity-type effects? We saw major transformations, at least I wasn't aware of until I did the research, that electricity enabled.

Gene Kim: One of my favorite ones is the automobile, combined with the interstate highway system, leading to the modern supply chain, which had incredible knock-on effects.

Dr. Joe Davis: And the American suburb. We don't have American suburbs without the automobile.

Gene Kim: You had this wonderful self-effacing depiction of what GenAI could do for an economist. Can you share that story, and what does one make of it as the person overseeing the department of economists as well as from the economist's read on it?

Dr. Joe Davis: I was trying to automate my own job, so it's kind of fun. Economists are easy pickings. We get our forecasts, half of them are right, half are wrong. It's like a coin flip. On one hand, on the other, sometimes you can't get a straight answer. So I was going to pick on my own profession.

I used ChatGPT as a simplistic way to say: could it do five hours of my job, a day's project, like forecasting inflation? I went through that exercise. I asked it all those questions, and I know how to do that. I've been doing this over 20 years. I asked it to get data to generate an inflation forecast, and then it pulled all the data and knew what models to run without me force-feeding it. Then it wrote all the report. I made the fun joke of, let it write an email to my boss to share the report, which it had written up, because I've been so busy during the day. It did it in one minute.

Then I went through those tasks and said, listen, you can save roughly 50% of the economist job, where AI is a substitute for half the tasks. For the other half, it's a complement. It was like 50%; it can do half the job. And my colleague always reminds me economists are only half of a job anyway, so I guess half of mine is gone.

Then I did it for nursing, where you save a little bit of time. There's the paperwork and electronic health records. If you have friends and colleagues in the health profession, you know it's really taxing. You have a little bit on fewer errors. There'll be a 20 to 30% lift in productivity just on nursing. I went through a few of those examples to see how much time are we saving, or could save. Again, we haven't seen this. I'm projecting forward where I believe AI's abilities will be in five to seven years' time. That's why there's a range of outcomes.

Gene Kim: In that nursing example, what I found so compelling was where the time is being saved. It's in filling out reports and trying to find information.

Dr. Joe Davis: That's where labor economists are really important. We're using data that labor economists use. We're a really good research team, with advanced degrees, but we're building upon the work or reading the work of Professor Rock and others. That's why we've talked to them. You've got to compare notes. These are smart people.

In the healthcare profession, some professionals spend more time on paperwork and records than they do actually dealing with patients. Just think about that. That's why I said there's actually a lack of automation in today's society. I know it's a scary word, but at least for the next five or seven years, I'm actually for that. This is actually going to be a boost. More often than not, some jobs are in the crosshairs if 80% of their time is doing that sort of activity. That's been the case for 70 or 100 years. It'll continue to be the case.

Gene Kim: I'm wondering if you can role play. You're the head of nursing, or I'm the head of the economist department and you're my boss. Give me the best possible reaction to these findings, and the worst possible reaction to these findings that leads to all sorts of bad outcomes.

Dr. Joe Davis: For most employees, that's why where we've been working at Vanguard is thinking about use cases and learning from this technology. We're a regulated business. You mentioned all the assets we have under management. It's not our money, it's our clients' money. Maybe for you on the call, it may be your money. We have to take it very seriously.

Can we think of safe use cases and get it in the room? I always want an IT professional in the room. They know these technologies. I know just enough to be dangerous. Get the IT expert, get the HR expert, and those three individuals: business leader, IT expert, and HR. Think in those three ways and try to get a handle on where we can use this technology for the betterment of the workforce.

For the foreseeable future, for the next several years, this is more about getting it and deploying it at scale, and getting the use cases. After that, yes, I think we're having some organizational change and maybe redeploying some workers that have IT skills here and moving them up here, because we can automate this fraction of the jobs you do. But from a nurse, we may have to shift how we're doing time shifts because now we have less time on the paperwork, and that may have downstream implications.

I'm not worried about dystopia right off the bat. There are one or two occupations with high automation intensity that I think will be, but that's not in healthcare. It's not in education, it's not in finance at least, and it's not even in IT other than basic coding. But you could say that about a lot of things. IT skills are in incredible demand.

Gene Kim: The term you taught me was the lump of labor fallacy. Let's switch roles. I am your boss, and you're the head of nursing, head of technology, or head of the economics department. Joe, I hear that you've been able to automate 30% of the tasks away. I relish this opportunity to shrink costs by 30%. What is the best, most thoughtful response you would give?

Dr. Joe Davis: That shouldn't happen. That would be myopic. The lump of labor happens in this dystopia debate: the more the machine can do, think of a pie, think of this as a lump, and the more technology can do, we're just left with the scraps. What's wrong about that is that the amount of work we need to do at a company and in society, think of how many problems we have in society. How much work is there needed to do? The pie grows over time.

If you say to me, Joe, we can get 30% more time out of what you were doing, hopefully that unlocks so I can get to more higher-end stuff and leave the other stuff and go. In rare examples, if whatever you were able to automate is the only thing I was doing in my job, that job is gone. But more often than not, 80% of jobs have multiple critical tasks. The problem is we don't have enough time getting to the top.

I mentioned the nurse. A cardiologist spends as much time on healthcare records as in surgery. That's crazy. How can you shift the time? They've done it in manufacturing; robotics are doing half of the physical movement of certain things. Can we get that in the service sector? Can we get that in IT? Can we get that in finance? We're sophisticated at Vanguard. We're using a lot of computer software and data scientists. In my group, we have a lot of sophistication, but we could always be more efficient at scale.

I think it would be a redeployment of some of those tasks, but we have to get a better handle on what AI can do in a sandbox environment before you just start massively deploying it.

Gene Kim: If what you say is true, your forecasts are correct, and this is an epochal shift of the last time we saw 50 years ago, what advice would you give to people just entering the workforce: high school students, college juniors, people just entering the workforce?

Dr. Joe Davis: I've said this for a long time. I'm not dismissive of technical skills, because I went to graduate school. Whether it's IT or analytical skills, my advice has always been, and I'm just telling you what I give to my children in college: read as much as you can. That gives you context. I have a goal of three hours a day minimum, and I tell my kids social media doesn't count. Actually, that would be a penalty. Social media and sports, whatever, that doesn't count. I don't even care what topic, because that can give context.

What's key here for AI is that it's not the answer, it's asking the right question. That's both if you have a STEM background or if you have a liberal arts background. If I get you with AI, let's say you can do some of the stuff you were doing before, but I get more time of your creative ability to ask the next several questions. Now we're cooking. That's the advice I'm giving: read as much, because you need context. I need to read to have context in some of these things I'm talking about. I didn't grow up knowing this. It's by doing research and reading. So reading is my long-winded way. Three hours a day.

Gene Kim: I have to ask one last question. I told you this story about how in two hours I got to do something that I thought would have taken a week. But even deciding to do the work was hard. For me to consider writing what I wrote, it was just never a good month; maybe next month. That seemed to generate a reaction from you. Can you describe what that reaction is and why that's important?

Dr. Joe Davis: That's a productivity lift in real time. Economists don't do a good job, myself included, by talking in this wonkish term. Who's walking around talking about productivity? It's just: can I get things done faster to get on with the value? If you can deliver that same value, same output, but in half the time, that's how the United States went from an average income of $30,000 per worker to $100,000. How did that happen? It's because of things like that across all different jobs, and you add them compounded over time.

We want that. We never want change too much too quickly, and that's where AI and ChatGPT, I think, caught the social consciousness. Not on this crowd because everyone knows it, but it's more the average person saying, whoa, what can it do? That's what we're talking about.

Gene Kim: You're talking to a group of technology leaders who have very much skin in this game. Is there any help you're looking for from this community to help in anything that you care about?

Dr. Joe Davis: I think it's seeing what's beyond the initial, where everyone's starting to deploy it and use it. Where are the genuine use cases beyond the first automation round? I'm looking for that. Secondly, it was the question you asked me: where are some of the knock-on effects? You merge AI with something else, and I don't know what something else is. What can that start to unleash?

I throw out healthcare as one, but I would really appreciate, this is a really smart group, what are some of the other mergings, whether that sounds scary or optimistic? That's what I'm on the lookout for because that's what changes society. The internet and computer did that with hindsight. That goes beyond just changing how we do things. Those two things would be really helpful.

Gene Kim: Fantastic. Dr. Davis, thank you so much for teaching us, and actually not just about general purpose technologies, but also who economists are and what they do. Thank you so much.

Dr. Joe Davis: Gene, always a pleasure engaging with you. And it's Joe. Forget doctor. Just the dismal science.

Gene Kim: Thank you so much, Joe.

Dr. Joe Davis: Thanks, Gene. Cheers.

Closing (Gene Kim)

Oh, that's so fantastic. I get to hang out with a lot of people with economics backgrounds these days. One of these things I did not see coming.