John Y. Campbell is the Morton L. and Carole S. Olshan Professor of Economics at Harvard University. He has published over 100 articles and books on finance, including The Econometrics of Financial Markets (with Andrew Lo and Craig MacKinlay, 1997), Strategic Asset Allocation: Portfolio Choice for Long-Term Investors (with Luis Viceira, 2002), The Squam Lake Report: Fixing the Financial System (with the Squam Lake Group, 2010), and Financial Decisions and Markets: A Course in Asset Pricing (2018).
Campbell has served as President of the American Finance Association, as Chair of the Harvard Economics Department, and as a board member of the Harvard Management Company, and he is a founding partner of the asset management firm Arrowstreet Capital.
Navigating the world of finance and investing is undeniably complicated, sometimes unnecessarily so. And all too often the people who end up making the most costly financial mistakes are those who can least afford to do so. But what exactly needs to change in order for more people to make wise and well-informed financial decisions? And how do we go about implementing those changes? Joining us today to help us unpack this topic and the many decisions involved in the world of investing is John Y. Campbell, a British-American economist, professor of economics at Harvard, and founding partner at Arrowstreet Capital, a systematic asset management firm based in Boston. John has published over a hundred of articles on a range of topics including fixed income, equality valuation, portfolio choices, and household finance, all of which we explore in today’s expansive conversation. We kick things off by discussing utility theory, why it’s so important to the study of finance, and what it can teach us about risk aversion, before delving into portfolio structure, asset allocation, and hedging. John also expands on the study of household finance, the mistakes that households typically make, why household behaviour tends to differ from theoretical predictions, and how to bring theory and behaviour into alignment. We wrap things up by discussing how financial literacy, education, and regulation can improve outcomes for households before hearing John’s advice on selecting an optimal mortgage contract along with an overview of the type of risk that mortgage contracts expose you to. Today’s episode is jam-packed with information and insights from a profoundly knowledgeable figure in academia.
Key Points From This Episode:
An overview of asset pricing theory; unpacking the utility function in finance, what it teaches us about being risk averse, and how it is used to determine the value we place on any amount of money. (0:04:01)
The implications of using the Capital Asset Pricing Model (CAPM) for portfolio choice. (0:13:58)
The difference between arbitrage pricing theory and the Intertemporal Capital Asset Pricing Model (ICAPM). (0:18:15)
How predictability in stock returns affects portfolio advice for long-term investors and why John prefers the cyclically adjusted price-to-earnings (CAPE) ratio. (0:23:40)
Why a long-term inflation index bond is the ideal risk-free asset for a long-term investor, and how portfolio advice, concerning bonds, changes when inflation index bonds are not available. (0:28:32)
The impact that labour income should have on optimal portfolio choice and the relationship between human capital and financial assets as you age. (0:35:31)
Learn about John’s 2004 paper entitled ‘Bad Beta, Good Beta’ and how intertemporal asset pricing explains differences in returns between value and growth stocks. (0:38:33)
The benefits and drawbacks of value investing: why historically they do well on average, but extremely poorly over certain periods. (0:41:12)
A breakdown of stochastic volatility and how it affects portfolio choice for long-term investors. (0:47:16)
How long-term equity investors should approach foreign currency hedging in their portfolios, and how fixed-income investors should deal with foreign currency exposure. (0:50:07)
The study of household finance, what it aims to answer, and the major challenges in this area of study. (0:53:54)
An overview of the mistakes that households typically make, how costly they can be, and why household behaviour tends to differ from theoretical predictions. (0:59:57)
Suggestions on how household behaviour and theoretical predictions can be brought into alignment and the methods that have the most potential to improve outcomes for households. (01:04:47)
What households should take into account when selecting an optimal mortgage contract and the different types of risk that mortgage contracts expose people to.(01:10:18)
How John’s definition of success has shifted over the years, the joy of academia, and why he is especially grateful for the opportunity to connect with students on their educational journey. (01:16:04)
Read The Transcript:
Ben Felix: This is the Rational Reminder Podcast, a weekly reality check on sensible investing and financial decision-making from two Canadians. We're hosted by me, Benjamin Felix, and Cameron Passmore, portfolio managers at PWL Capital.
Cameron Passmore: Welcome to episode 250. And Ben, we've been to some meetups lately where we sat down and spent some time with listeners. And one of the common pieces of feedback that I heard was some of those interviews are so good. I have to listen to them two and three times. Or they're kind of complicated. I want to make sure I get all the nuance. Well, this is at least a two-time listen. This was a phenomenal conversation with Professor John Y. Campbell.
So interesting. So thoughtful. So many moments that I just found myself kind of lost in his answers. They're so good and so thought-provoking. This, everyone, is one of those episodes.
Ben Felix: Oh, yeah. I'll go back and listen to this more than once for sure.
Cameron Passmore: John's a British-American economist, a professor of economics at Harvard, where he's been since 1994. He's published over a hundred of articles and all sorts of topics in finance including fixed income, equity valuation, portfolio choices, household finance. And we talked about all these subjects. I mean, the questions that you put together, Ben, were phenomenal. John was very thoughtful and very prepared. This was a real home run conversation on topics that matter.
And the line that he said – like we talked about household finance. I wrote it down. Because he gave his list of the top mistakes that people make at household finance. And I thought one of his comments was so good, which is you can't solve problems you don't understand. Making a commentary on the industry, and complexity and things that people have to learn. And I just thought it was brilliant. But we got into details about portfolio structure, and asset allocation, hedging.
Ben Felix: Yeah, we talked about utility theory, which from reading John's textbooks, he starts there and builds from there. We covered utility, which is something that, as important as it is to the study of finance, and John talks about how important it is, it's not something that we've gone through with ourselves or with any other guests. But he's arguably the best person to do it. That was a great part of the conversation at the beginning.
And then we covered portfolio choice and asset pricing, household finance, which is another area he's done a ton of great work. And then we touch briefly on mortgage choice, which is another topic he's done quite a few great papers on.
Cameron Passmore: Yeah. John's the past president of the American Finance Association and former director of the program and asset pricing at the National Bureau of Economic Research. He's also a founding partner at Arrowstreet Capital, which is a systematic asset management firm based in Boston. Has a Ph.D. in economics from Yale.
Ben Felix: All right. Well, like you said, Cameron, this is one of those episodes that you might have to listen to more than once. It was an incredible conversation. Let's get to it with John Y. Campbell.
***
Professor John Campbell, welcome to the Rational Reminder Podcast.
Thank you for having me. It's a pleasure to be here.
We're very excited to be speaking with you. John, to start, what does asset pricing theory aim to accomplish?
Well, I guess the question is why are asset prices what they are? Why is the S&P 500 index at 4,000 and not 400 or 40,000? And an important aspect of that is why returns are what they are.
Historically, the S&P 500 has delivered an average real return of around, say, 7% a year. Why is it 7% and not 0.7% or even 70%?
Why is the theory of choice under uncertainty important to asset pricing?
Well, asset prices are set by investors who buy and sell assets in an uncertain environment. And their decisions determine the level of asset prices and the required returns on an asset. We have to understand what market participants are doing in order to understand the prices that they end up setting in the market.
What a great answer. How does standard microeconomics represent preferences?
Well, in economics we have this theoretical device of a utility function. We represent preferences with the utility function. And then what that does is it tells you any two options, any two choices. Which one is preferred? Or whether they're equally attractive.
Now if you apply that framework to risky gambles or sometimes called lotteries, then, under plausible conditions, you get a special kind of utility function, which is called a cardinal utility function that tells you the level of well-being that corresponds to any particular prize in the lottery.
In finance, we think about money, and we're going to get a utility function that tells us how much we value any amount of money. And then people can be understood as maximizing the expectation or average of that utility that they get from different amounts of money. And this expected utility theory, that's the bread and butter of academic finance.
But what does it mean to say that someone with preferences represented by a utility function is risk-averse?
Risk aversion means that you're going to prefer – it's like think about two prizes that are different, two amounts of money that are different. And if you’re risk-averse, you're going to prefer the average amount of money or the average of the two prizes to a 50-50 gamble on getting the bigger amount or the smaller amount.
And if we focus on this case where all prizes are just amounts of money, that corresponds to a utility function with the property that each extra dollar you get gives you a little less extra utility than the one before. In other words, it's a utility function with diminishing marginal utilities to use the jargon.
Now a utility function like that, if I can indicate it, it has a shape like this, which we call concave. Concave function. And we can study it using the mathematical theory of concave functions. There's an important tool in that theory called Jensen's inequality, after a guy called Johan Jensen who was a Danish telephone engineer.
And I sometimes tease my students in my finance class that we're studying the economics of Jensen's inequality. We're asking how does risk aversion play out in the marketplace?
Now how do you decide which utility function to use to represent people?
Okay. That's a great question. A good starting point is to observe that as the world has gotten richer over recent centuries – you know, we're all much richer in the 21st century than we were in the 18th century. But human attitudes towards risk measured in proportional terms relative to your wealth, those attitudes don't seem to have fundamentally changed. Things like risk premia. You know, same order of magnitude now that there were hundreds of years ago.
That tells you we need a utility model in which aversion to proportional risk, this is sometimes called relative risk aversion, is independent of the level of wealth. Now that doesn't pin down how big it is. It just says that relative risk aversion is a number. But we don't know what the number is. But that's really the starting point for most finance theory, is to say that wealth by itself is not going to change your relative aversion to risk over time.
How do you estimate how risk-averse an investor is?
That's another great question. There're various ways. One game I like to play with my students in my personal finance courses is invite them to do a thought experiment where a demon turns up in your room at night and forces you to gamble with the demon. And the gambling game gives you a 50% chance of winning. And if you win, now you're going to toss a coin. If it comes up heads, you get to spend 10% more throughout your life than you otherwise would have. And if you lose, if it comes up tails, you've got to spend 10% than you otherwise would have.
And if you're risk-averse, that's unattractive. You wish the demon would go away. So you offer the demon some amount of money, some amount of your wealth to go away and not make you play this game.
The question is how much of your wealth would you give the demon to get out of the gamble? And whatever percentage you decide, say, I give up 2%. Your risk aversion is twice that. Twice two is four. If you'd give up 2%, the risk aversion is four.
Now, needless to say, that is quite different from the investor questionnaires that mutual fund companies and investment advisors give their clients. And those questionnaires are qualitative. They're not tightly linked to utility theory. I think they provide pretty good guidance about who's relatively conservative and who's relatively aggressive. But they don't actually give you a number that you could plug into the theory the way my little thought experiment does.
That's an interesting observation. Do you think that investment advisors and financial planners should be trying to estimate that?
I think they could and should do a little more along those lines. I mean, it does depend. It's probably better done in the context of a personal relationship with a financial advisor than just filling in a form for a mutual fund company. But I think financial advisors can try to work with thought experiments like this.
And there're other ones too, right? I just like the one I gave you because it's so colourful. But there are other ways to do this as well.
Okay. Very, very interesting. On to some portfolio theory, how are utility and risk aversion applied to portfolio choice over a single time period?
Okay. The classic result is that if you are choosing between a safe asset and a single risky asset, then the share of your wealth that you should invest in the risky asset is the risk premium that's the expected excess return on that risky asset divided by the variance of the risky asset return, which is a measure of its risk. And also, divided by the coefficient of risk aversion. That number we were just talking about, if that's four, you have to divide by four.
Now that formula works. It's correct. Provided that variances of adequate measure of risk. For example, if returns have a nice normal distribution, a nice bell-shaped curve with that shape, then you're good to go and you can use this formula.
What are the shortfalls of mean-variance optimization?
Well, most obviously, financial markets are subject to extreme events. Wild things can happen. Like the stock market crash of 1987 when stocks fell 20%. The US market fell 20% in one day. I'm so old that I remember actually teaching in a classroom when that happened. And it was the morning, and I had a very boring lecture about the distribution of stock returns and the fact that it had fat tails. And I was droning on about that and a student in the back of the room raised their hand and said, "Professor Campbell, Professor Campbell, do you know the Dow is down 200 points already?" And I just said, "Yes, yes, that makes my point." And finished my lecture. But that was an example of tail risk.
Now that's a very old-time example. But of course, much more recently, we had the global financial crisis and we had the COVID-19 pandemic. And what happened in March 2020 is an example of tail risk. These tail risks are things that should worry investors and are not adequately captured by variants. That's the first-order point.
But there's another way in which mean-variance optimization can go wrong, and that's when you're trying to combine many risky assets to find a risky portfolio that has the highest ratio of reward to risk. If you succeed in finding that best risky portfolio, then you can treat it as the only risky asset of interest. It's your best mutual fund. And you can allocate between it and the riskless asset using the approach we were just talking about a minute ago.
But how do you find that best portfolio? The problem is if you use historical variances and covariances to combine risky assets along with historical mean returns, that procedure can go disastrously wrong if you're looking at too many assets with not enough historical data.
In that case, what you'll tend to do is find a portfolio just by chance in the historical period you're looking at happened to be almost riskless. And you're going to think this is an amazing deal. You're going to lever up, invest aggressively in it. And then ex-post in the future, it won't be riskless. It'll move around a lot and with high probability, you'll go bankrupt. That is a serious flaw in what I'll call unconstrained mean-variance optimization. They just take thousands of assets and look at their variances and covariances.
Super interesting to think about. Because you still see that we, as practitioners, maybe not completely unconstrained. But trying to optimize for sharp ratios based on historical data is still something that people try to do.
Absolutely. Absolutely. And you could do it, but you have to do it very carefully with constraints to avoid this trap of essentially fooling yourself into thinking that there are safe assets that don't really exist.
How do we get from mean-variance optimization to the capital asset pricing model?
The capital asset pricing model, or we say CAPM for short, that says the best risky portfolio with the highest reward-to-risk ratio is just the evaluated index of all assets that exist. The so-called market portfolio. And if that's true, investing is a nice simple process.
Now what's the argument for that? Why should that be? Well, it's a very simple argument. If all investors want to hold the single best portfolio and they agree on what it is, then, because total demand must equal total supply, that best portfolio must be the portfolio of all the supplied assets.
Now I just made some super strong assumptions with that argument. It's not obvious that all investors want to hold the portfolio with the highest reward-to-risk or Sharpe ratio. Some investors may be trying to hedge non-financial risks. For example, their labour income. And investors may differ in their information and their beliefs. People may be speculating against each other.
But still, I would argue that despite those important objections, the market portfolio is always a good reference point for investors because it's so well-diversified. You may not end up holding exactly that, but you're going to want to be not too far away from it.
What are the implications of CAPM pricing for portfolio choice?
Okay. I was sort of moving in that direction. I mean, if CAPM holds, then every investor who lives purely off financial wealth should just hold the market portfolio along with maybe some cash in a money market fund. Or if they're aggressive, they should try and borrow and lever up and keep the money in the market index.
Now you can ask, "Well, how do you lever up?" Well, maybe you have a house and you take a big adjustable rate mortgage and you're borrowing at that rate and you put your money in stocks. People do do that.
The standard advice that finance professors give, which is buy a low-cost equity index fund, that is based on the idea that such a fund is a good approximation to the market and that it is going to be at least close to mean-variance sufficient. It'll have an attractive Sharpe ratio.
Do you think it is a good approximation of the market portfolio?
I would argue you can do better with some portfolio tilts. And for example, I think taking a value tilt is a good idea. I think taking a low beta tilt is a good idea. So, low-risk stocks.
Actually, this is important for many individual investors because many of us have older relatives who have a retirement fund and they keep money in cash and then they have some money in the stock index. And anybody who's doing that could probably do better by having less money in cash and put a bit more money in, but put it in low beta stocks. You'll end up with the same risk and a higher average return. That's the so-called beta pricing anomaly.
There are ways to deviate. I would argue, the CAPM is not exactly right. Sort of thinking about the market first, that gets you in the right ballpark. And then you could try to be clever and take some factor tilts.
Okay. You touch on this. But I want to ask about it explicitly. What additional portfolio choice insights do we get from the introduction of multiple factors, pricing factors, in a single period setting?
Okay. Multi-factor models embody a very important insight that was originally due to the late Steve Ross, who taught me finance at Yale actually in the 1980s. It was an unforgettable experience studying finance with Steve Ross.
Now, diversifiable risk, the risk of individual stocks, can't be compensated in the market. Why? Because if it was, well, you could buy a bunch of high-return idiosyncratically risky assets, put them in a big portfolio and get an almost infinite ratio of reward to risk. Because the risk in the portfolio would diversify away. But the high returns would remain. And it would be too good to be true.
It follows from Steve's argument that even if the CAPM doesn't hold for whatever, reason we should expect returns to be associated with groups of stocks that move together. And there're two main ways to find these extra sources of return. One is to look for characteristics of stocks that predict their returns, like value, high book-to-market ratio, for example. Or momentum, meaning that the stocks have done well in the last year.
Steve's argument says that stocks with those characteristics, if they have high returns, should move together. And indeed, they do. Value stocks tend to go up and down together. Momentum stocks tend to go up down together.
Most recently, momentum stocks all crashed in recent weeks when Silicon Valley Bank went down. Tremendously negative returns to momentum. And they move together.
Now another approach to find returns is to find stocks that particularly move together. You can use a statistical method; principal components analysis or some variant of that. Find the most important ways in which stocks move together and then see if there's return associated with that common movement.
Steve's argument says that a portfolio like that may have a high return, although it doesn't have to. But indeed, very often, that method also finds high-return stocks. Whichever way you get there, these multi-factor models advise investors to load up on groups of stocks with high returns relative to their risks. And in the modern era, we have smart beta funds, for example, that do this. You have value funds, momentum funds, low beta funds and so forth.
We've been very deliberately asking about a single period setting. And I want to move to asking some questions about a multi-period setting. But before we do that, I just realized that you're the best person to ask this question to. And it's something that I just haven't squared away in my brain after reading about this stuff for some time. What is the difference between arbitrage pricing theory, that you're just talking about from Steve Ross, and the inter-temporal capital asset pricing model?
Arbitrage pricing theory, its strength and its weakness is its generality. Steve's argument that diversifiable risk can't be compensated. That's very, very general. But the theory is silent on which common risks are compensated and which are not? They don't have to be. They could be. But they don't have to be.
In fact, we don't even know in Steve's theory which risks get a positive reward and which, yeah, a negative reward, because some common risks might be providing insurance, valuable Insurance to investors. It says nothing about that.
The inter-temporal CAPM is an equilibrium theory about how investors think about risk and return over the long run, fine factors and what their risk prices ought to be. And whether the risk prices should be positive or negative. And how big? It's a more restrictive theory, which means it's more likely to be wrong. But on the other hand, if it's right, it's more useful. It's telling you more about the world.
Okay. That's very interesting. Would it be safe to say that the main difference is the generality of the models?
Yes. The Arbitrage pricing theory is an extremely general statement about the way the world has to be. The inter-temporal CAPM is a much more specific model.
Okay. No. That was a great explanation. Continuing on that topic, how does asset pricing change when we move from a single period, as we've been talking about, to multiple periods?
Well, I'll answer that with two points. I can think of two big differences. One is that single-period models essentially start from an exogenous description of what you're going to get next period. There's something called the payoff on the asset next period. And it's random but it's exogenous to the model. It is whatever it is. And then the point of the model is to tell you what the price of that is today.
Now in a multi-period grid model, next period's payoff is determined in part by the price of the asset next period. But that's endogenous to the model. The model has to tell you the price every period. That makes life much more difficult. Working with multi-period models is hard. You're going to have to either solve backwards from some final period and do that over many, many periods to work back to the present. Or else, find a solution that can describe an infinite lived asset's price in every period of its infinite life. Kind of a fixed point of the model. Okay. So that's problem number one.
Problem number two is that returns that you realize over many periods are partly the result of addition, right? Because you get a dividend that's added to your wealth each period. And partly the result of multiplication, because returns compound. You keep multiplying those returns to your wealth. You've got a mix of addition and multiplication. And that makes multi-period asset pricing models non-linear. And hence, hard to work with.
One of the important things that I did early on working with Bob Schiller was to find tricks essentially to approximate these relationships linearly so that you could study them and they become much more tractable if you have the right approximation.
Wow. Okay. Now how does portfolio choice change when we move to a multiple-period setting?
Well, a long-term investor has to think about risk very differently than a short-term investor. For a long-term investor, risk is not a volatility of wealth over some period…of the consumption that can be supported by that wealth. A standard of living that you're able to support.
Now if sustainable consumption is the product of two things. The return are that you get on your wealth and the level of wealth, W. It's R times W. That's what you could spend each period. Now that means that change in R has just as big an effect on what you can consume as the same proportional change in W…early in this century, that's cutting the real interest rate in half. That's cutting R in half. That's just as bad for a long-term investor. Long-term investors then have got to be very serious about thinking about that risk that the interest rate or the rate of return you can earn as you reinvest your wealth, but that's going to go up or down.
That's a conversation that we had lots when interest rates rose recently where we were trying to say, "No. This is good."
Yes, exactly. It's good, right? It's good.
What effect does time horizon have on portfolio choice for long-term investors?
Well, conservative long-term investors want to limit the risk of fluctuations in their product, R times W. And the way you do that is you find assets that do well raising your W whenever R falls. Or conversely, do badly with W falling whenever R rises. The two elements, the R and the W, are going to move opposite one another, right? And that's going to stabilize the product.
Now one example is a long-term bond, because long-term bond prices go down when interest rates go up. People holding long-term bonds may be feeling sore because they just lost money. But actually, it's hedging because they're going to earn a higher R in the future. The product R times W is safer than it otherwise would have been.
Bonds is sort of obvious. But even stocks, stocks are risky long-term assets. But they also have this property because there's evidence that the expected future real return on stocks declines when the level of the stock market goes up. With stocks as well, the R and the W are moving opposite. And that tends to be true for long-term assets. And that's why conservative long-term investors should like these assets.
Yeah, I want to push more on that. How does predictability in stock returns change portfolio advice for long-term investors?
Any asset whose return is predicted negatively by its price, like a bond, or I argue, like the stock market, is going to be safer for long-term investors than it is for short-term investors. And long-term investors who are conservative whose risk aversion is large enough, greater than one, they should hold more of these assets than short-term investors who have the same risk aversion. Horizon will influence assets you're in.
Now, on the other hand, if you're a long-term investor and you're aggressive with a risk aversion less than one, you actually want to hold assets that have high returns after their prices go up. Because those sorts of assets will enable you to invest more money at times when that money is productive and you can earn a higher return.
For example, hedge funds or private equity firms might want to have financial positions that deliver cash to them at the moment when they can then invest and earn a higher return. That type of strategy gives aggressive long-term investors a higher standard of living on average. But that standard of living will be very risky.
How predictable or mean reverting do you think investors should assume aggregate stock returns are?
My best guess based on work research I've done with a colleague, Luis Viceira, at Harvard Business School is that suppose you look out 30 years, suppose you have a 30-year investment horizon, our best guess is that the annualized risk of stock returns at that horizon is only about half the annualized risk at a one-year horizon.
Invest over one-year annualized risk, say, 16%. Invest over 30 years, more like 8%. If you look out ten years, I'd say the annualized risk is perhaps two-thirds or maybe 70% of the one-year risk.
Another way to put that is to say that only about half of the short-term volatility of stock returns is coming from news about future cash flows on corporations, which create permanent shifts in the level of stock prices. The other half comes from changing discount rates that investors apply to those cash flows. And those things are temporary.
The 16% vol that you experience over a year, half of that is permanent news about the very long-run profits of corporations. And the other half is just the ups and downs that get caused by are changing discount rates.
That is such an interesting way to look at it. Which variables predict returns?
I would always start with valuation ratios that measure the level of stock prices relative to some measure of cash flow or accounting value on stocks. You could start with the dividend yield. The problem there, of course, is there are also equity repurchases, which can distort that.
My preferred ratio actually is the so-called cyclically adjusted price-to-earnings ratio, or CAPE ratio. Bob Shiller and I, working in the late 1980s, popularized the use of that ratio. And the high level of that CAPE ratio in the late 1990s was what persuaded Alan Greenspan to use the phrase irrational exuberance for the market at that time. And I still follow the CAPE ratio quite closely.
Now there are other variables you can look at too to predict returns. But certainly, for individual investors, I wouldn't bother with them. But for institutional investors, there's some evidence of short-run momentum even at the aggregate market level. There are some seasonal effects. You can do things with interest rates. But I think for asset allocation for individual investors, I'll just focus on CAPE and leave it at that.
We use CAPE not to make asset allocation decisions, but in creating our estimates for expected future returns. CAPE is one of the inputs that we use there. Now this is important because – so you gave us some really interesting estimates and explanation of the difference between short and long-term risk and where those come from. And that will change people's asset allocation decisions. How confident do you think investors should be in that predictability?
Predictability is hard to establish definitively, right? You're trying to predict something that is itself extremely noisy. It's going to take a lot of data and analysis to do that. And it's always possible to argue that things are different this time. And there's been structural change and so forth. I think it is important to ask how well can you do out of sample.
I published a paper a few years ago with Sam Thompson where we expect basically looked at that. We asked whether predictive regressions work out of sample consistently. And we found that they do so long as you impose sensible restrictions on the forecast.
For example, that the equity premium is always positive. You're unlikely to be right if you're predicting a negative equity premium. But using a few simple restrictions from theory and combining them with the forecasts that you can make statistically, that seems to work out of sample pretty reliably.
What is the risk-free asset for a long-term investor?
The risk-free asset for a long-term investor is a long-term inflation index bond. Why? Because that bond is going to pay you an amount of money with a constant purchasing power every period in the form of a coupon. And that is known in advance. In the US, these bonds are known as TIPS, treasury, inflation, protected securities. But a number of countries have them. The UK, Canada, other countries. That is the simple way to think about it.
Imagine for simplicity that you live forever and you can buy an infinitely lived inflation index bond at perpetuity. That's inflation index. this bond is going to pay you a fixed real coupon, a fixed real income every year forever. You can have a very nice peaceful life. The check comes in the mail. You spend what you need to spend. You don't even know what the price is. You don't care. The price can go up and down. You know care. You have a stable income.
Now for a finite-lived investor, of course, the equivalent is an inflation-indexed annuity that pays you for the rest of your life in real terms. The US Social Security system is actually pretty close to that, which is one reason why it's so politically popular in this country, is it gives people a safe real asset, which otherwise it's hard actually to buy that in capital markets. But real pension income provided by the government is close to that.
I love the inflation index perpetuity example. John Cochran explained that to us, that that is – well, like what you just said, it's the risk-free asset. But in the short run, it can be extremely volatile in its price, which is so, so fascinating to think about.
You mentioned Canada. Canada's actually – that's where we are. They're shutting down their real return bond program. How does portfolio advice with respect to bonds change when inflation index bonds are not available?
Well, that depends on whether the Central Bank reliably keeps inflation under control. In the stable inflation environment of the early 21st century, inflation index bonds didn't add much honestly because nominal bonds were almost risk-free too. They're quite similar to inflation index if you know what inflation is going to be.
But if inflation is volatile, and in particular, if it may change in a persistent fashion. For example, if there's a risk that the Fed might raise its inflation target from 2% to some higher number, which some people advocate, then nominal ones are not safe assets for long-term investors. The inflation target could go up and then those bonds would pay a lower real income and lose value in real terms and would be risky. That's really the issue, is how much inflation risk is there?
I know. It's interesting that Canada is eliminating these bonds. I think that a lot of treasuries and finance ministries around the world think of their task as simply minimizing the interest costs to the treasury. But I actually think that providing a safe real asset as a sort of benchmark for the economy is also important.
In the US, it was actually Larry Summers when he was Associate Treasury Secretary, who persuaded the US Treasury to issue TIPS for the first time. And Larry, obviously very distinguished economist, but also quite politically savvy. I mean, I remember him telling me once, "I knew that within five or ten years we would be able to say either that we had saved the US Treasury a great deal of money. Or that we had created a wonderful new investment for the American people." Depending on whether they did well or badly, he would spin it one way or the other. But really, I think his purpose was to create this safe asset and hopefully help improve the financial system thereby.
If we took the example of like – say, we have inflation index bonds for one investor and the other investor doesn't have them, what would that change – holding all else constant, what would that change in terms of their optimal portfolios? Are we changing the bond duration? Are we changing the equity fixed-income mix?
Well, in the absence of inflation index bonds, a conservative long-term investor is going to have to cobble together as safe a long-term portfolio as you can. There are things like real estate that you can argue provide inflation hedge and stable real income. But of course, there can be shocks to the demand for real estate.
Think about the fact that the pandemic diminished the demand for prime commercial real estate and increase the demand for residential real estate outside cities. There are these other shocks that can create risk there. It's going to be a much tougher task to approximate a safe long-term real bond if we don't have these assets available.
Interesting. What drives the covariance between stock and bond returns?
Well, let's focus initially on nominal bonds. Let's think about stocks and then conventional US treasuries or Canadian government bonds. For those, the short answer is that if inflation and real interest rates are pro-cyclical and to go up during booms, then bonds do poorly in booms when stocks do well.
The boom is good for the stock market. But it's bad for the bond market because inflation is high and … are high. In that world, bonds and stocks move opposite one another. And that's been the world we've lived in from around the year 2000 through the COVID-19 pandemic.
Very salient example in the global financial crisis, stocks crashed. Most assets crashed. But the FED cut interest rates and there were fears of deflation. US treasury bonds did extremely well. They provided a hedge.
On the other hand, if inflation and real interest rates are counter-cyclical, then bonds and stocks both do poorly in recessions. And that was the world in the 1980s when we had oil price shocks that caused both inflation and recessions at the same time and the fed was willing to create recessions to bring down inflation.
There is some concern today that we might be re-entering a regime like that after the COVID-19 pandemic because the supply shock caused by the war in Ukraine. And also, the Fed’s need to re-establish its inflation-fighting credibility, which has been damaged recently.
I've been monitoring the day-by-day movements in the bonded stock market, and the covariance between bonds and stocks is certainly less negative than it has been on average in the 21st century. But it's not yet gone back to the strong positive levels that we saw in the 1980s. I think it's still kind of the jury is out on which way it's going to go.
If that – and it sounds like it is. If that covariance is non-stationary, how does that affect strategic asset allocation decisions?
Well, the world where bonds and stocks move opposite one another is a great world for long-term investors who hold both asset classes in a portfolio. For example, university endowment funds typically do that. In the world of negative covariance, the two asset classes hedge one another and you can invest aggressively in both. You don't need to have much in cash. Or you might even lever up and just sort of really go all in to bonds and stocks. If the covariance switches sign so these asset classes move together, then those investors need to rethink their risk exposure and pull back because they no longer get the natural hedge of the two asset classes for each other.
What impact should labour income have on optimal portfolio choice?
I would start by saying that most people's labour income is relatively safe. It's not totally safe. But certainly, much safer than the dividends on stocks. And that means that a young investor with many years of future earning power has an implicit asset that is relatively safe. This is what economists call human capital future earning power.
And when I'm talking to Harvard students in the classroom, I say your human capital is orders of magnitude larger probably than your financial assets. That's much less true for me. I'm in my mid-60s. A young investor with all this human capital can then afford to take risk aggressively in their financial portfolio because they've got this – the financial portfolio is just a little part of the iceberg above the water. Most of the iceberg is below the water and is safe. The overall risk exposure of your total wealth is quite reasonable even if you invest aggressively in your financial portfolio.
Now then, of course, over time, you have to dial back the risk as you get older and retirement approaches. That's the strategy that target date funds follow. And I endorse that basic approach. Target date funds are not perfect however because they don't take any account of how much have you saved and how well have you done.
If you go through a great stock bull market in stocks and your target date funds do really, really well, your financial wealth is now larger relative to your human capital and you should then dial back the risk a little bit. Target date funds don't do that. They just condition the asset allocation on age. But still, they're a good first cutter, first approximation for many people.
That sounded like the relationship between labour income and time horizon is changing asset allocation decisions. Do we know enough about the idiosyncratic labour income risks for one person versus another assuming the same time horizon to inform portfolio recommendations?
Yes. I mean, there are certainly some people. For example, stock brokers or people working in the financial industry who have income that is actually highly correlated with the stock market. And those people should be much more cautious about equity investments.
If one could go into more detail, actually, you could say, "Well, if you're in a particular industry, you should hold stocks that are outside your industry because your human capital is already exposed to that industry." And of course, the one thing you should never do is hold an undiversified position in the stock of your employer because you're doubling up the risk there.
It's also true that there are people who have extremely volatile income that's idiosyncratically risky. People who are very exposed to spells of unemployment. Or maybe you're working for a biotech startup and you have this drug that might be a blockbuster or might fail. A person like that is going to have to make sure to have enough liquidity held in safe assets to get through the periods of low income. They're going to need to be more cautious for that reason.
Yeah, that's interesting. That's a reason somebody might hold more cash. I think that also comes up when we get to mortgages. I don't want to jump ahead to that yet. But –
Yeah, absolutely.
Yeah. Okay. Now how does inter-temporal asset pricing? You mentioned value stocks earlier. How does intertemporal asset pricing explain differences in returns between value and growth stocks?
Okay. As we've already discussed, the inter-temporal perspective says that negative shocks to aggregate cash flows, which cause permanent declines in stock prices. Those are worse for long-term investors than just surprise increases in discount rates, which lower stock prices today but cause them to recover later.
An analogy I would like to make is with corporate bonds. Imagine that you were holding a portfolio of corporate bonds, and two things could happen that would lower the value today. One is some of the bonds could default. The other is that interest rates could go up.
Now if you have to sell a portfolio right now, all you care about is what you get for it. It doesn't matter whether it's interest rates or defaults. But if you're holding the bonds to maturity, defaults are obviously much worse because that money is gone and it's never coming back. Whereas if the bonds don't default, they will pay off at maturity if you just hold.
Okay. The theory says then that the cash flow risk is the really bad risk for a long-term investment. It turns out that value stocks are more exposed to that risk of fluctuating cash flows. While growth stocks are more exposed to the risk of changing discount rates. That kind of makes sense because growth stocks have long duration. A lot of the value is from distant future profits that they will make when they take over the world in the future.
Like long-term bonds, much of the value of a growth stock comes from a distant future. So then they're going to be very sensitive to discount rates. That pattern means that inter-temporal asset pricing theory predicts that value stocks will have higher average returns than the simple CAPM, the one-period perspective would predict. And that's just what we see in the data.
I wrote a paper on this in the early 2000s. And we showed that you can use an inter-temporal model to break the CAPM beta, the measure of risk in the CAPM, into two pieces. One, we call bad beta. And that's high for value stocks. And the other, a good beta, which is high for growth stocks. And the CAPM beta is the sum of the two.
We made an analogy which you kind of like with cholesterol. In the old days, cholesterol was the measure of heart attack risk. You went to the doctor. They did a blood test. They told you your cholesterol level. And then at a certain point, people discovered there're actually two kinds of cholesterol. And one is much worse for heart attack risk than the other. Now you have good cholesterol and bad cholesterol. We say you've got good beta and bad beta. And the bad beta is high for value stocks even though the total beta is low.
Yeah, that paper is like mind-blowing. The first time I read that paper was, yeah, very exciting. Really, really, really good stuff. Can you talk about – given what you just explained, what are the portfolio choice implications for, I guess, risk-averse long-term investors? But maybe because you mentioned this earlier, also the portfolio choice implications for risk-seeking long-term investors?
Great. Okay. Conservative long-term investors should be reluctant to take a value tilt in their portfolios despite the fact that there's this value premium because value stocks increase their long-run risk. In effect, a value portfolio has less of that mean-reverting risk reduction than I was talking about earlier than the market index does. It's going to look riskier to a long-term investor.
Now the counter argument is if you're a risk-seeking long-term investor, you'll want to hold growth stocks as a way to get dry powder when it's most valuable is the way I'd put it. You're not so concerned with reducing the long-run risk. You're concerned with having resources at times when you can reinvest those resources productively. It's sort of the opposite to the insurance argument.
That's really interesting. An aggressive long-term investor would want to tilt toward value stocks but also own growth stocks to rebalance into value stocks? Did I understand that correctly?
No. Start from the fact that just the average returns on value are higher. Why doesn't everybody want – why doesn't everybody want value? The argument for conservative investors is that, if you hold value, you end up increasing your long-term risk.
The argument for aggressive long-term investors is actually you do want value. Why? Because by going away from growth, you preserve the value of your portfolio when discount rates go up. And you want to have a portfolio that's valuable and available for reinvestment at those moments when discount rates have risen a lot. You want to be able to invest.
Yeah. It's the opposite of what I said. You want to have value stocks –
It's the opposite. It's the opposite.
Yeah, yeah. Interesting. I had not thought about that. Value stocks are less sensitive to discount rate shock. You get a discount rate shock, your value portfolio doesn't drop.
Right. Right.
Super interesting. What drives the boom and bust that we see in value stocks?
Value investing historically has done well on average. But it can do extremely poorly in certain periods of time. For example, in the tech room in the late 90s, disastrous for value investors. Global financial crisis, again, disastrous. And of course, the COVID-19 pandemic in 2020, really disastrous. There are conversely other times like the credit boom before the financial crisis when value was doing very well.
It's not surprising that value is risky. Because after all, the Steve Ross argument we discussed earlier says anything that has a return associated with it has to be risky. But what's driving the risk? What are the shocks that move value around?
I've looked at that in a current working paper that we just – it's written with Stefano Giglio and Christopher Polk, and we just started to circulate the paper last month. Hot off the press. What we show is that if you just use aggregate market movements, you can only explain about five percent of the variation in value versus growth returns. It doesn't tell you much.
But you can get that up to 50%. You can explain about half the variation using three shocks that are motivated by an intertemporal model. The cash flow and discount rate shocks we just discussed, and then one more which is shocks to the risk of variance volatility of stock returns. And essentially, whenever you have bad cash flow shocks that's going to drive down value relative to growth, whenever you have increases in discount rates, that's going to drive growth down relative to value. And whenever risk goes up, again, growth is going to do well relative to value.
So interesting. I love this one, because when we talk about value and growth, one of the things that I've seen people do is compare their standard deviations over time and say, "No. No. Value stocks aren't really that much riskier than growth." But this is a whole – I mean, this is the whole concept of, I guess, ICAPM pricing. We don't just care about variance. This illustrates it so clearly.
We even care about market conditions, right? And there're exposures to different things that move market conditions around. Yeah, that's the way to think of it.
Now those inter-temporal risk exposures that you mentioned, are those more pronounced intra-industry or inter-industry?
This is one of the things that we emphasize in this paper I just mentioned, is that you can break value into two pieces. There are the bets that you make across industries. Value Industries tend to be line physical industries, fossil fuel producers, utilities, things like that. Banks, very often. Conversely, the growth industries tend to be things like pharmaceuticals, and tech and all of that.
But then within each industry you also have value stocks and growth stocks. It turns out that these inter-temporal risk exposures are much more pronounced within industries. Once you look across industries, there are idiosyncratic sectoral shocks that are more important.
In the COVID-19 pandemic, for example, unsurprisingly, biotech stocks did well, right? Vaccine producers. And IT stocks, because Zoom was making the thing we were all using. While physical businesses, like utilities and fossil fuel producers, they did very poorly.
The particular pattern of sectoral shocks will always be important for the inter-industry value return. But this inter-temporal stuff we were just talking about, that really shows up very clearly when you control for industry and look at value versus growth within each industry.
Yeah, that's so interesting. If you're a value investor, if you want to tilt your portfolio toward value, I guess you would care more about the intra-industry systematic risk than you would about the idiosyncratic inter-industry risk?
Yeah. And it turns out, it's really the intra-industry value tilt that's rewarded on average where you get the risk premium. I mean, many institutional investors do this. They implement value intra-industry primarily. And they try to hedge out the industry tilts.
Yeah, okay.
Obviously, it's just a value mutual fund that you – a retail mutual fund won't do that. But that's actually what sophisticated institutions do.
Okay. Now how does stochastic volatility affect portfolio choice for long-term investors? And maybe it makes sense for you to quickly unpack what stochastic volatility means for the listeners.
All right. Stochastic is just random. And volatility means how much a stock return is moving around. And we know that we measure that with variance or standard deviation in the square root of variance. And we know that that changes over time a lot. Sometimes market is very volatile. Sometimes not so much.
Well, often we talk about that by referring to the VIX Index, which is the measure of volatility calculated from auction prices. But think of it as just the day-by-day fluctuations. The amount of risk that's hitting the market at any point in time.
Now that moves around. And when risk goes up, that's bad news for investors. Because for any given amount of return that you're getting in the stock market, if the risk is high, then you're having to take more risk to get the same return. An increase in risk, controlling for return is a deterioration in your investment opportunities. It's a bad thing for long-term investors who are reinvesting their wealth in the stock market. They're going to want to hedge against increases in risk by buying assets that go up in volatile times.
For example, equity index options do that. Or it turns out that growth stocks also do that. And the reason probably is that growth companies have options embedded in their businesses. They have the option to implement a new technology now or put it off, wait, and do it later. They have a lot of flexibility.
And so, in a risky time, they actually can exploit the shifting opportunities more nimbly than a value stock that's all about the assets in place. It does turn out that whenever volatility is high, growth stocks tend to do well. Whether the market is up or down. In the tech boom, the market was up. Volatility was high. Growth stocks did well. In the global financial crisis, the market was down. But again, growth volatility was up and growth stocks did well. It's a very systematic pattern.
That means that, for long-term investors, growth stocks are a hedge because they hedge against increases in risk. And so, that effect can further help the intertemporal model explain why value stocks have higher average returns.
That is incredible. Absolutely incredible. When we spoke with Bob Merton, it came up in the conversation, growth stocks for long-term investors. And we talked about your ‘Good Beta, Bad Beta’ paper. Or I made reference to those findings. And his comment was, "Well, I would actually think about growth stocks more from the perspective of options." But you just explained that when you take that perspective, you get effectively the same result as being a long-term hedge.
You do. You do. And since the ‘Good Beta, Bad Beta’ paper, I've written another paper, which it sort of adds in that third element of the risk and sort of puts it all together. And both things matter, I would argue.
Incredible. How should long-term equity investors approach foreign currency hedging in their portfolios?
Okay. Great question. And a big change in topic. But very interesting. I would say equity investors should primarily ask how currencies move with the equities they're holding. Because there's only modest return differences across countries. Yes, there's a little bit of return difference. Deviations from uncovered interest parities and profit opportunities. But it's pretty modest.
The main thing you're doing when you're holding a currency is your taking this risk exposure. Now some currencies move with the world stock market and their own domestic stock markets. And the Australian dollar and the Canadian dollar are two of the main examples. The Canadian dollar tends to strengthen when world stock markets are up.
But there are other currencies, like the Euro, the Yen, and the US dollar, that tend to move the other way. The US dollar very obviously strengthens when we have financial crises. And these currencies that move against the stock market are particularly attractive for long-term investors because they hedge the equity exposure in a relatively cheap way. Yeah, you own a lower return on those currencies, but only slightly lower. And so, it's a fairly affordable hedge.
Given what you just described, how would optimal currency hedging be different for a Canadian investor versus a US investor?
So I would argue that both the Canadian and the US investor should want to have US dollar exposure in preference to Canadian dollar exposure. But what that means is that, for a US investor, they should currency hedge their Canadian equity investments. They might want half some Canadian stocks. But then they want to hedge that back into US dollars.
Whereas for Canadian investors, you want to not currency hedge. You want to leave your US equity investments in the US dollar and thereby take this US dollar exposure.
How stable do you think we should assume those types of relationships are?
Well, I haven't done research on this very recently. But I last published a paper on this around 2010 or so. But since then, I think if you look at the COVID-19 pandemic, once again, the US dollar strengthened a lot in that downturn.
Obviously, things could change if the US government becomes utterly fiscally irresponsible. It could lose its position as the sort of reserve currency and ultimate safe haven currency. Things could change. It isn't that there's some God-given property of the US dollar that does this?
But that status of the dollar and of the other large reserve currencies I mentioned seems to be fairly stable. It's not that easy to change it up. The properties of the Canadian and Australian dollars are quite likely linked to the importance of commodity exports both the Canadian and Australian economies.
Now, over time, you could imagine that Canada and Australia may diversify their economies and become less commodity-dependent. That might change the relationship. But the extent that the sort of industrial structure of the economy is pretty persistent and changes only slowly, I would expect this to be a fairly persistent attribute of the Canadian dollar.
How should fixed-income investors deal with foreign currency exposure?
Well, when we looked at this, we found currencies and bond returns are not very strongly correlated. Fixed-income investors should basically currency hedge. If you're buying a foreign bond, the currency risk exposure of that is just sort of pure noise. Like the pure noise, just hedge it out.
I mean, it's odd in a way that fixed-income mutual funds don't offer that more often. One can find currency-hedged bond funds. But you have to sort of know what you're looking for. But I would argue those should be attractive.
All right. I want to do a bit of a shift. We've been talking about asset allocation-related stuff. I want to do a bit of a shift into household finance, which is another area that you've done just incredible research. What question does the study of household finance aim to answer?
Household finance is the study of how households use financial markets and financial instruments to achieve their objectives. Just as corporate finance looks at how a corporation uses financial markets to achieve their objectives, we're just asking the same questions about households. There are two main aspects of that. One is normative or prescriptive, which is what should households do? And the other is descriptive. What do households do?
What are the unique challenges for household finance as an area of study?
There are challenges, major challenges, both on the normative side and on the positive side. On the normative side, to make good recommendations, you have to be able to solve a model that captures the full complexity of the household's financial problem. Now that's tough because households have labour income, which is random, and it evolves with your age. They have spending needs, which also evolve with family structure, for example, if you have kids.
Households can only borrow at high-interest rates only up to a certain limit. Some borrowing as floating rates. Some as fixed rate. They face the complex personal income tax system. They need housing and durable goods as well as day-by-day non-durable consumption and so forth.
Ironically, a household's financial problem is actually much more complicated and harder to solve really than a corporations problem. Even though the corporation is much more sophisticated and employs professional well-trained staff to do the financial management. That's the normative challenge.
Now on the positive side, the descriptive side, you need to be able to measure what people do. And ideally, also, what they believe. That's really tricky. Because surveys can't go into sufficient detail. If you really asked all the detailed questions you want to ask, people would become disgusted and stop answering the survey.
And administrative data are usually partial, right? You might get data from, say, a brokerage firm. But that's just getting one brokerage account. It's not looking at the whole set of financial accounts that a person has.
We've made some progress in recent years though, and this is a large reason for the growth by using data from private sector firms that aggregate people's financial information or by going overseas, particularly to Scandinavia, where governments collect very comprehensive information about people's finances and make that available to researchers.
You mentioned households are differing in their beliefs as being one of the challenges of household finance. Do you have a sense of how much households differ in their beliefs and preferences?
The evidence on beliefs is just beginning to come in now as people – really, the Holy Grail in the field is can you link self-reported beliefs to what people actually do? And you're always going to have a small sample where people report their beliefs. And it's hard to link to what people do. But we're beginning to learn about that.
I would say there are some persistent differences in belief. Some people are just always more optimistic than others. It's sort of like temperament. Some people are more up. Some people are more down. And then when people say their beliefs are changing, they do seem to change what they do but only a little bit. Not nearly as much as the sort of portfolio choice formulas that we discussed earlier would imply. It's as if the beliefs are a very noisy measure or sort of amplified measure of what they truly act upon.
Now with regard to preferences, that's another great question. And I have a working paper that studies this in Scandinavia where we can measure quite accurately both how much households save and also how they allocate their assets. This paper, which is written with Laurent Calvet, Francisco Gomez and Paolo Sodini, a kind of very European author team. It's called The Cross-Section of Household Preferences.
And we study Swedish households that are middle-aged and have sort of accumulated some retirement wealth. And we use a life cycle model, which sort of fits that part of life pretty well. And what we find is a great deal of variation in the rate of time preference how impatient people are.
Some people are very impatient. They save early in life. They do their retirement saving. And then other people are impatient and put it off and have a great scramble as retirement approaches. We also find a lot of variation in the responsiveness of saving to interest rates, what economists call the elasticity of inter-temporal substitution.
And we find a bit of variation in risk aversion, but actually less in that. And the reason is that almost all of these households in Sweden do take some risk through home ownership and equity mutual fund ownership also by leveraging using mortgages.
Most people have a sort of moderate coefficient of risk aversion somewhere between four and seven. There's some variation there. But you don't see too many people with risk aversion of one, or less, or 50, or more. It's not that kind of range.
It's interesting that we find so much difference even when we're looking at these middle-aged Swedish households that have some stock market ownership. Because if you looked at the full population, including the very young, the very old, people who don't have any stocks, you'd probably find even more variation. But that's for us to look at in future research.
Fascinating. How well does financial theory describe the observed behaviour of households?
I think it depends a lot on how well-educated people are. I would like to hope that the behaviour of those of us on this Zoom call would be reasonably well-described by financial theory. And more generally, people who have more education, income and wealth tend to do quite well. But less educated people, much less so.
And in many different arenas, you can document this, that the theoretical prescriptions are a good first cut. Furthermore, educated and less so for … and less educated people.
One of the problems is the financial system is unduly complicated and that further disadvantages people with lower education. And one of the things I am trying to do in my work is to advocate for simplicity in financial product design. We should be able to make things a little bit easier for ordinary people.
I want to come back to that topic later. But I'm going to save it because I've got some more things I want to ask you. First, what are the mistakes that households do to typically make?
There's a long list. I would say the easiest mistakes to sort of prove unambiguously are the cases where people leave money on the table. Where they just – for example, failing to exploit an employer match for a retirement fund even when you actually have the ability to withdraw the money if you need it. Or failing to refinance a mortgage when you could reduce your interest costs substantially and you have the right to do it and you just don't do it. Or not shopping around and ending up paying a very high price for a completely standard financial product like an index fund.
Those mistakes are really unambiguously mistakes. They may not be the biggest mistakes though. And let's make the analogy with how did the federal government end up putting Al Capone in jail? Not from murdering people, although he'd done that. But for tax evasion. Because that was what they could prove. And so, these sort of leaving money on the table things are sometimes quite small. But you can unambiguously prove that they're mistakes.
It's a great analogy.
Now, broadening out, what mistakes do people tend to make? I mean, here are a few. Taking out student debt to go to a school that you're not going to be able to complete, you're going to drop out or take extra time to graduate. And you're going to end up in a very serious debt trap.
Failure to create a reserve fund of, say, three months’ spending to cover emergencies. Repeatedly borrowing for short periods of time at a very high-interest rate. Growing over payday loans, for example. That's a debt trap. Not saving adequately for retirement. Mismanaging risky investments. Either taking no risk at all — just leaving your money under the mattress — or failing to diversify. Buying the stock of your employer. Or trading too much. Fancying yourself as a day trader and trading too much. Paying high fees on financial products either because they're too complex to understand or because you don't shop around.
There are housing mistakes like thinking housing is a great investment, so you should buy a house that's much bigger than you need. That's a mistake. Picking the wrong mortgage. Failing to refinance at the right time. And then at the end of my list, I'll say ensuring small risks by buying extended warranties on durable goods while leaving large risks like disability or long-life uninsured. Lots and lots of mistakes.
Yeah. Do you have a sense of how costly these mistakes can be?
They can be very costly. The mortgage, for example, that's the biggest liability that most people have. It can easily be hundreds of thousands of dollars. And if you don't refinance, you could easily pay an interest rate that's, say, 1% or 1.5% higher than it should be on a debt of $100,000. That's substantial.
Not saving enough for retirement can again have enormous implications. Not annuitizing and outliving your wealth. Or ending up house-rich and cash poor and reluctant to move but trapped in a house within an adequate cash flow. These can be really pretty serious. Other things, like wasting money on an extended warranty. I mean, it's a mistake, but it's small. There's a wide range.
Yeah. It's fascinating to hear the list of mistakes though. Why do you think that household behaviour does differ from theoretical predictions?
Of course, there's a huge field of behavioural economics and behavioural finance that studies that. And I think that it's made some progress. For example, there's the concept of present bias, which means succumbing to temptation and sort of putting off painful things and indulging yourself today. And this helps to explain behaviour like getting into credit card debt and rolling it over persistently, which a lot of people do. That's one example.
There are also – there's this phenomena of loss aversion. The tendency to sort of think about risks in isolation relative to a reference point. That can explain why people who buy stocks, they like to realize losses and they hang on to gains. I think, for some behaviours, behavioural finance tells us a lot.
But actually, I think a lot of it is simpler than that. It's that people are confused that this financial thinking does not come naturally to people. They just don't understand the problem. You can't solve a problem and come up with a rational answer if you don't understand it. People find finance scary and off-putting. They don't like to think about it. So they don't think about it. And I'm not sure that the behavioural economics research agenda really tells us much about that.
Okay. Now based on the way that you had to answer that question, I think I can predict you answer the next one. But I'm still going to ask it. Do you think that households should aim to behave more like theoretical predictions? Or do we need to adapt the models to how households behave?
I think for a lot of household finance, we need to try to find a way for households to do more like the theory. I think it's going to involve designing simpler financial products. It's going to involve some regulation. I think there's a role for government mandating financial education. I think there's a role for financial advice. We can talk about ways to do it. Or I think it needs to happen.
Now having said that, I don't want to imply that financial theory is always perfect. And let me give you an example. Years ago, my colleague, Greg Mankiw, published a paper called The Asset Allocation Puzzle, in which he pointed out that financial advisors routinely tell their clients they'll give model portfolios for conservative, moderate and aggressive clients.
And in those model portfolios, the ratio of stocks to bonds is higher for the aggressive than for the conservative. And his co-author pointed out that that's inconsistent with CAPM thinking. Because in CAPM thinking, there is some portfolio of stocks and bonds that's best. And if you are more conservative, you just dilute that with cash. And if you're more aggressive, you lever up. But you don't change the ratio of stocks to bonds.
He said there's a puzzle here. Why are people doing this crazy thing? Well, the inter-temporal model that we were discussing earlier actually says, "You know what? That's actually right." Because if investors are long-term, then the safe asset for long-term investors is actually bonds. Not cash.
Asset allocation puzzle, not a puzzle. If people were behaving that way – and the simple theory, the CAPM theory, said that was wrong, the problem was with the simple CAPM theory. Not with what people were doing. That's an example. And we should always be alert to other possible examples where actually we need to work harder on the theory and make it fit household behaviour better.
Between financial literacy education and regulation, what do you think has the most potential to improve outcomes for households?
I certainly think there's a role for financial literacy education. I think it's important. One thing I'm doing these days is teaching a personal finance course to Harvard undergrads with no prerequisites. Anybody can take it. You don't need to know any economics. And I feel that's very important. And I'm very excited about doing it.
I think a number of US states – I don't know what the situation is in Canada. But a number of US states mandate financial literacy education in high schools. And some of those curricula are quite good and can make a difference. I will say that in high schools, the problem is that the students are not yet making most of the financial decisions they're talking about. Perhaps they're going to have to decide how much student debt to take on. But that's about it.
Imagine if driver's ed were only in the classroom with no road component. How much would anybody learn about road safety from, right? It's the experience of driving that teaches you how to drive. And similarly, I think financial literacy education is better done at a time when people have the decisions to make.
For people who go to college, taking it senior year is great because they're about to graduate. They have to think about cost of living, taxes, 401K, retirement plans, all kinds of things. The problem is the majority of people don't go to college. They graduate high school. They start working.
High school may be too early for the education to be effective. But once people have left school, how do you ever reach them? And it's kind of an unsolved problem. In certain special circumstances, we can address it. Like the US Army, for example, has a sort of financial literacy boot camp that they put the recruits through. And that's great. I think that's a good thing and very helpful.
I think we should do what we can. But it is not a panacea. And I think there's also a role for regulation to try to squeeze out the most complex and difficult-to-manage financial products and promote the use of simple products, which needn't necessarily be the most familiar ones. There's a role for innovation as well.
And finally, there's a role for financial technology. I mean, I think robo-advising, for example, is pretty helpful in spreading low-cost reasonable quality financial advice to more people.
Yep. We've occasionally picked on a Canadian robo-advisor because they started out doing exactly what you just said. But they've slowly over time started adding more high fees, speculative products, and crypto trading and stuff like that because they need to make money. There's an interesting problem there too.
Yeah. The problem in the financial advice industry is that advisors have a business. And for that reason, they need to cater to the biases of their clients up to a certain point. They can gently pull their clients in a good direction. But if they tug too hard, they know that the fish is off the hook. The fish swims away. It is very tough.
What role do you see for financial advice that is assumed to be good?
I think it can be great. I think the challenge is to make it affordable enough. I mean, if you've got a financial advisor who's well-educated and is a fiduciary, so no conflicts of interest, it can be very good. But the trouble is people who've studied financial advisors, some of this research is in the Canadian context, find that advisors have all kinds of idiosyncrasies.
It's not that they are corruptly cheating their clients. It's that they believe weird things. They invest for themselves the same weird way they invest for their clients. But it's not necessarily a great outcome for their clients. How we solve that? I'm not sure. But good financial advice can be very valuable if we can provide low enough cost.
We have just a couple questions left on mortgages for you. What considerations does a household need to make to optimally select a mortgage contract?
Okay. I've thought about this over the years. And there's this choice between fixed rate and adjustable rate mortgages, which exists in many countries. Of course, the mortgage system is very different in Canada. But in the US, we have these very long-term 30-year refinanceable fixed-rate mortgages. And then we have adjustable-rate mortgages.
We also have in the US this phenomenon of points, which I've learned more about recently, which is very weird when you think about it. Because what are points? You have the ability to borrow a little extra. But instead of increasing the face value of your debt, what that does is increase the interest rate. You're borrowing more money. But it doesn't change the officially recorded debt. It just changes the interest rate. Well, that's odd.
Now, ultimately, what you might care about is the monthly payment you have to make. It maybe doesn't matter if the debt is big and the interest rate is low. Or the debt is low and the interest rate is big. It's the product that sort of matters. But you always have the right to refinance.
If you think you're going to refinance, you might do well to take points, borrow some extra money, promise to pay a high rate. And then shortly thereafter, turn around, refinance and never pay the high rate, or not pay it for very long. This is a weird and complicated system. And I think one of the things we ought to do is to eliminate the system. Because there's evidence, savvy people with education do well. And less sophisticated people, and particularly minority, Black and Hispanic borrowers, do worse and end up suffering in this system. There's effectively a cross-subsidy from the poor to the rich with this system.
But to get back to your question, what are the considerations to optimally select a mortgage contract? If you know you're going to move soon, or if you are very short of money right now because you are stretching to buy your first house, or you expect your income to be going up rapidly in the future, all you care about is the interest rate right now. And that's usually lower on an adjustable rate mortgage than a fixed rate mortgage. Not always. But usually. And that will push you towards an adjustable rate mortgage.
If you – in a relatively stable financial position, and you like predictability of payment, then a fixed-rate mortgage is attractive. Finally, if you are a wealthy middle-aged person who has a big house and a financial portfolio, you may – and you're relatively aggressive with low risk aversion, you may want to lever your financial portfolio in effect. The way to do that is to borrow with an adjustable rate mortgage. Borrowing at the short rate as cheaply as an individual payment.
And if interest rates go up, you can always de-level by selling some of your bonds and stocks and paying off the mortgage. This advice corresponds to the patterns we see in the data. The adjustable share is higher among first-time home buyers, young buyers, subprime borrowers. And it's also higher among jumbo mortgages that are the very largest ones often with middle-aged borrowers with substantial financial assets.
We see that pattern in the US. It's sometimes attributed to government distortions of the market. But actually, we see the same pattern in Denmark where there are no distortions. I think it's a fundamental phenomenon that's driven by the considerations I just mentioned.
Interesting. Can you talk about the different types of risk that mortgage contracts expose people to?
Sure. An adjustable-rate mortgage, the risk is that inflation and interest rates will go up and your mortgage payment will go up very rapidly. Many people are living through that right now. I became very aware of this because, back in the 1980s, I grew up in England. Went to college in England. I moved to America. But my college friends all moved to London. Started families. Bought houses around the same time in London in the 1980s.
Then they all had adjustable-rate mortgages, which are dominant in the UK. And then in the late 80s, the government lost control of inflation. Interest rates went way up. House prices fell. They didn't have any more borrowing capacity. They had negative home equity. The rates went up. They had babies, and toddlers and they just had to tighten their belts and take the hit to the family budget. It was very painful. That's a risk.
It's a risk if you have no borrowing capacity. In that story, the problem was that they didn't have any other way to borrow. If you have unused borrowing capacity, it's really not a problem. The higher adjustable mortgage payments in the higher inflation environment are just compensating lenders for the inflationary erosion of the debt. And you can just adjust your borrowing and it'll all be fine.
For fixed-rate mortgages, you're obviously protected against that upside risk in mortgage payments. If rates fall, you can refinance. But one risk that can be very important is if – and I don't quite know whether to call it a risk or not. But here are two things that can go wrong. One is rates can fall and you can forget to refinance. You just don't realize you should. That happens a lot actually. Lots of people fail to refinance and pay much more than they need to.
In the US, that's particularly true again among Black and Hispanic borrowers. Research I've done in Denmark shows that it's low-education, low-income people who make this mistake. And also older people.
The other problem can occur when rates rise. Because under US rules, if you move, you give up your old cheap mortgage and you have to get a new expensive mortgage. And that can create locking where people are very reluctant to move. That hasn't been studied in recent years. Because for decades, mortgage rates have predominantly fallen. Now they've spiked up a lot and may go up further. Is there a risk that labour mobility will decrease and it'll jam up the labour market? And people won't be able to move to better job opportunities because it's too expensive to give up the old mortgage? Time will tell. But that's going to be a very salient research topic I think.
Very interesting. Our final question for you, John. How do you define success in your life?
Okay. Well, that's a deep and very personal question. When I was young, I was an ambitious young man with an Oedipus complex. What did I want to do when I was young? I wanted to become a more famous scholar than my father and grandfather who were both academics. And I also wanted to make money like my mother's father who was a Wall Street banker. I wanted to sort of live up to what I saw in my family. And I guess that made it inevitable that I'd become a financial economist and start an asset management company.
Well, now I'm older. And I've come to realize those things are actually not that important. The joy in academia – there's a lot of joy in academia. But it's about discovery. Not fame. It's about figuring something out and having the penny drop. It's really wonderful.
The joy in business is building a company that serves its clients and its employees and seeing how that is a living entity. And the deepest joy in life is about sharing life with a partner you love and trust and helping younger people grow up to be happy and productive adults.
I'm deeply grateful for the relationships with undergraduate and graduate students that I have had in my academic career. It's one of the great joys of an academic career is these lifelong friendships with young younger people. And I'm even more grateful for my wife, my four children and my two grandchildren.
I want to specially mention my oldest son who has Down syndrome. He has taught me a lot. If I'm wiser of an older adult than I was as a young one, that's partly due to him. I mean, he taught me that IQ is not the measure of worth, contrary to what many academics think. And that no disability needs to stand in the way of a passion. In his case, it's the passion for music and playing piano to share that music with other people.
Wow.
Those are my thoughts.
What a wonderful and beautiful answer. John, this has been an incredible time together. Thanks so much for joining us.
It's been a pleasure. Thank you very much.
Is there an error in the transcript? Let us know! Email us at info@rationalreminder.ca.
Be sure to add the episode number for reference.
Participate in our Community Discussion about this Episode:
Links From Today’s Episode:
Rational Reminder on iTunes — https://itunes.apple.com/ca/podcast/the-rational-reminder-podcast/id1426530582.
Rational Reminder Website — https://rationalreminder.ca/
Shop Merch — https://shop.rationalreminder.ca/
Join the Community — https://community.rationalreminder.ca/
Follow us on Twitter — https://twitter.com/RationalRemind
Follow us on Instagram — @rationalreminder
Benjamin on Twitter — https://twitter.com/benjaminwfelix
Cameron on Twitter — https://twitter.com/CameronPassmore
John Y. Campbell — https://scholar.harvard.edu/campbell/home
'Who Should Buy Long-Term Bonds' — https://www.nber.org/system/files/working_papers/w6801/w6801.pdf
'Inflation Bets or Deflation Hedges? The Changing Risks of Nominal Bonds' — https://scholar.harvard.edu/files/campbell/files/campbellsunderamviceira_20160523.pdf
'Growth or Glamour? Fundamentals and Systematic Risk in Stock Returns' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/gorg20090319_copyedited.pdf
'Bad Beta, Good Beta' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/bbgb_2004_nberw9509.pdf
'An Intertemporal CAPM with Stochastic Volatility' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/cgpt_volatilityrisk20170123final.pdf
'Global Currency Hedging' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/globalcurrencyhedging_20090128_manuscript.pdf
'Biases in long-horizon predictive regressions' — https://www.sciencedirect.com/science/article/abs/pii/S0304405X21004013
'What Drives Booms and Busts in Value?' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/cgp_valueboomsbusts_20230311.pdf
'Household Finance' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/householdfinance_jof_2006.pdf
'The Cross-Section of Household Preferences' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/calvetcampbellgomessodini_20221027.pdf
'Restoring Rational Choice: The Challenge of Consumer Financial Regulation' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/elylecture_march2016.pdf
'Down or Out: Assessing the Welfare Costs of Household Investment Mistakes' — https://www.journals.uchicago.edu/doi/abs/10.1086/524204
'A Model of Mortgage Default' — https://scholar.harvard.edu/sites/scholar.harvard.edu/files/campbell/files/mortdefault13022014.pdf
'Household Risk Management and Optimal Mortgage Choice' — https://scholar.harvard.edu/campbell/publications/household-risk-management-and-optimal-mortgage-choice
'Predicting the Equity Premium Out of Sample: Can Anything Beat the Historical Average?' — https://www.nber.org/papers/w11468
'An Asset Allocation Puzzle' — https://www.nber.org/papers/w4857