Listen to the podcast:
There is a new leading indicator that uses intra-family flows into high yield — or junk — bond mutual funds to foresee credit-market overheating, which typically precedes economic downturns. Research by Wharton finance professor Itay Goldstein, and co-authors Azi Ben-Rephael and Jaewon Choi, uses their model to also predict the business cycle by forecasting GDP growth and unemployment — up to one year earlier than other indicators.
“It’s the first study that links the credit markets to flows in mutual funds,” Goldstein noted. The paper is titled, “Mutual Fund Flows and Fluctuations in Credit and Business Cycles.” He discussed the highlights of the study with Knowledge@Wharton. (Listen to the full podcast above).
An edited transcript of the conversation follows.
Knowledge@Wharton: There are a lot of studies out there, as you note in your paper, linking the business cycle to credit markets. One broad conclusion that you note — from the literature — is that growth and leverage in the financial sector combined with negative shocks leads to financial crises. The upshot in your paper is that while an overheating of credit markets typically precedes an economic downturn, a big question remains: Is there a way to know when overheated credit conditions are likely to develop before they actually develop, and if so, then the government might be able to take some action to blunt a possible downturn.
Please give us a short overview of what you have identified in the paper.
Itay Goldstein: As you note, the credit market is cyclical and it tends to be correlated with the economy as a whole. We tend to see that credit market is overheating before a downturn in the economy. And many scholars in finance and macroeconomics are trying constantly to find why this is the case, and also to find leading indicators to tell us when it is coming. People have looked at [mostly equity] mutual funds over the years, but did not find quite the right signal coming out of mutual funds that will tell us sort of a leading indicator for the credit cycle.
And we identified what we think is a good leading indicator, and this is flows into high yield corporate bond funds. But not just flows. What we have is a particular component of flows, which is the intrafamily flows into high yield corporate bond mutual funds [that occurred between January 1984 and December 2012].
… Our leading indicator has really two components. First it looks at interfamily flows, and then it looks at flows into high yield funds. Let me first talk about the interfamily aspect of it. The ICI, the Investment Company Institute, publishes data on flows in general across different funds, different types of funds, and also gives us information as to what flows are within the family, and what flows are outside of the family. When I say outside of the family it means money flowing into the fund family and out of the fund family.
“We identified what we think is a good leading indicator, and this is … intrafamily flows into high yield corporate bond mutual funds.”
So a fund family is something like Vanguard for example. So Vanguard is a fund family because they have many funds, and they are all under the big umbrella of Vanguard. And Vanguard would have corporate bonds, government bonds, equity, high yield bonds, investment grade bonds and so on.
And what we look at is flows within Vanguard into the high yield funds. And basically what we find is that when you look overall across all families, the intrafamily — the within-family flows into high yield bonds — are highly predictive.
Now why is that the case? So why should we care about the intrafamily component? Because we think that this is the first component to move, to show signs of changes in demand for risk, for changes in taste, for investment. Think about it, when investors have money in Vanguard or other fund families, and all of a sudden they have, say, a greater appetite for risk, they want to invest in high yield bonds. The first thing they will do is take money they already have invested in other funds of this family, and move it towards the type of fund that they are interested in.
And we think that this is why the intrafamily flows into high yield funds is the thing that is the most effective leading indicator. Because when something changes in underlying tastes, in capacity for risk, and appetite for risk, the first thing that will move, that will show signs of changes, will be the flows within the family towards that type of investment.
Knowledge@Wharton: So when you talk about those interfamily flows, you refer to them as the smart money. And I guess in this case, it doesn’t necessarily mean they are correct, it means that they are sophisticated investors who tend to move faster than others. … So how does focusing on that then get you to predict these other macroeconomic markers?
Goldstein: It has been noted before that the credit market is correlated with the economy. When there are signs of overheating in the credit market, generally it is correlated with contemporaneous, good economic conditions — high GDP, low unemployment. But then it leads later on to worse economic conditions.
“It’s the first study that links the credit markets to flows in mutual funds.”
… An overheated credit market would be one where credit spreads tend to be low, there is high volume of issuance of bonds — the share of high yield bonds out of overall bonds is generally high. Financial intermediaries are going to expand their balance sheets. So all of these indicators generally point to potentially overheated credit markets.
… But then it also predicts a downturn. So, how is all of this related to mutual fund flows? Basically what many researchers are trying to find out is, what are the origins of these credit market conditions, or credit market overheating.
We think that it originates from some change in demand by investors. Investors all of a sudden have maybe a more positive outlook for the economy, they have greater appetite for risk, and as a result they shift their investments towards the high yield funds, the high yield bonds. But the challenge is really to identify this signal coming out of market data. And what we found is that looking at this component of intrafamily flows into high yield funds is really an indication of this buildup of demand, and greater appetite for risk.
This predicts the improving conditions in the credit markets potentially overheating, and then leading to the downturn in the economy. So I am not trying to say that these flows within the family of mutual funds into high yield bonds are causing the whole thing, but they are just an indicator that within the economy people start building this appetite for risk, and want to invest in these high yield bonds. This is a clean indicator that this is happening. And then for us, researchers looking at that, it helps us predict the things that are about to come.
Knowledge@Wharton: What you have found seems to be an earlier indicator than anything else that is out there right now.
Goldstein: Right. Other indicators have been offered before, for example, the high yield share, which is the share out of total bond issuance that is coming out of high yield. Our indicator leads that by a year or even more. Another indicator that has been proposed is the excess bond premium — basically, the premium on bonds that is not coming from credit risk. And again, ours leads that by a few quarters. So basically it is a leading indicator that seems to predict all of the other leading indicators that have been offered in the literature.
Another thing that I think is interesting about it is that it also helps us maybe interpret better: What are the foundations for these credit markets overheating? Because it is a clear indication of an increase in demand from the side of investors to taking on more risk and investing in high yield bonds.
Knowledge@Wharton: Does it work in reverse? If that group of investors that you are talking about, the in-family funds, if they start to withdraw from these high yield bond funds, is that a sign that they think the problem is much closer?
Goldstein: Yes, that is a good question. So it works in both directions. We looked at symmetry, it’s not completely symmetric but it does seem to work on both ends.
Knowledge@Wharton: Is there some way to say, now they are leaving, that means we are going to have a problem within six to eight months, or two to three months. What is the correlation there historically?
Goldstein: Generally the timing of it would be that when the flows into high yield funds within the family start building up, it basically means that the other indicators of credit market overheating are going to show signs of going up within a year. Which means that GDP will go up within a year, and unemployment will go down within a year. And then the reversal happens about a year and a half to two years later.
“It is a leading indicator that leads all of the other leading indicators by about a year — or even more.”
So generally if you want to think about flows into high yield funds within the family today, when will GDP start showing signs of weakness or reversal? We are talking about three years later. Now clearly we are talking about statistical relations. So it’s not like every time you have flows into high yield funds within the family it means that GDP will go down three years down the road. We are talking about correlations over 25 years or so.
… It’s all statistical relationships, and it is all trying to control for other things that are changing around the same time. In general the nature of the exercise is that you run these regressions, but then you are trying to control for other things. So if you are an investor and you are trying to predict something, you would like to take into account other things, not just this particular factor.
Knowledge@Wharton: Before the podcast started I asked you what your model would be telling you right now, and you said haven’t run the current numbers just yet. Is it your plan to do that soon?
Goldstein: Now you’ve got me curious, maybe we should do that. Generally the primary goal of it is trying to understand relationships between variables over time. The way that we are looking at it is, we are running these regressions over time for 25 or 30 years or so. But certainly if you are trying to predict the future, then yes, you would like to take a look at what these variables are today, and what they tell you for the future.
We have this thing in the field where we say, okay so it predicts everything within a sample, does it do a good job also predicting things out of sample, and generally it does. It does work, but we haven’t looked at this particular point in time.
Knowledge@Wharton: You have to come back soon and tell us about that. In the meantime, tell us how your study differs from others that look at the relationship between overheating credit markets and the possibility of a downturn in the future.
Goldstein: It’s the first study that links the credit markets to flows in mutual funds. And the reason that flows in mutual funds are interesting is that they give you a clear indication as to where investors are moving their money. So people were trying to link those credit markets overheating into flows of money by investors, and I think we are the first ones that are able to do that.
One clear advantage of what we do is the fact that it is a leading indicator that leads all of the other leading indicators by about a year — or even more. Another clear advantage is that it lends itself to a very interesting interpretation, by which we see a group of investors, we can’t quite put a name on them, but we sort of know their characteristics, generally what they look like, how they behave. When these investors have more appetite for risk, those are the indicators that the whole credit market is going to start going towards the overheating.
Knowledge@Wharton: When you see this money within the family flowing into the high yield bond mutual funds, do you also look at where that money came from?
Goldstein: We did look at this. These inflows into high yield funds, are they correlated with outflows out of investment grade bonds or government bonds or equity funds? There isn’t such a strong correlation there in that it is sort of hard to predict — the inflows into high yield funds — and where they are coming from.
Part of the story is, and maybe this is also related to why the high yield funds are working so well, it is a very particular clientele of investors. When you think about [the investors in] high yield funds, those are typically savvier investors, more sophisticated investors, wealthier investors. Whereas if you think about, say, equity funds, everyone invests in equity funds. You have some sophisticated investors there, you also have some very non-sophisticated investors there. And this is why it is more difficult to get a clear indication of flows into and out of equity funds.
… One thing that we were thinking about is, is it really just a change in taste, demand or appetite for risk, or maybe these investors are just super smart and they predict everything going forward? And you can see that those things are slightly different. The first story is, they’re not necessarily super smart but they just give us an indication that there is appetite for risk out there.
“They’re not necessarily super smart but they just give us an indication that there is appetite for risk out there.”
Whereas the second story would say these are just incredibly smart people who are able to forecast the whole cycle and make money out of it, and no one else knows how to do it. And to some extent I would say that our evidence is consistent with both interpretations, but as we think about it and reflect on it, we think it is a little bit difficult to believe the second one. So this is why we are leaning more towards the first one. That is, the findings seems more to say that they just provide indication to the overall appetite for risk in the economy in the market rather than to say that these are incredibly smart individuals who can do something else that no one else can do.
Knowledge@Wharton: To the extent that this model proves to be accurate, it can be used perhaps by policymakers to provide some kind of policy change that could blunt or maybe even prevent a downturn in the future.
Goldstein: That is absolutely true. Certainly one of the primary reasons that this is so interesting is that policymakers, in particular central banks that are setting monetary policy, are continually looking for signs of overheating in the credit market and upcoming downturns in the economy because they want to adjust policy accordingly.
To the extent that this measure is something that they would use or accept going forward, certainly it can help guide their policies. So as you mentioned, one thing that we did was to use these flows within the family into high yield funds to predict future tightening in monetary policy.
That basically says when investors put their money more and more into high yield bond funds, then in the future policy makers will start tightening monetary policy. But if they accept our results and look at this measure in advance, then they could take those steps in advance and in some sense they could do better.
The fact that we can predict the policy suggests that this is useful information for them. If they have this information and use it in the future they could start adjusting the policy earlier. So I think that is potentially one of the more exciting ways that people can use this research.
Knowledge@Wharton: Can anyone use your paper to run the numbers for themselves?
Goldstein: Yes. I think it is fairly transparent. As I mentioned, a lot of it is based on the data of the Investment Company Institute, because they are the ones that publish the data on intra-fund flows. Now this data comes out a little bit with a lag, so it’s not like you can do it on a daily basis. But with that caveat in mind, this is certainly something you can use.
Knowledge@Wharton: What will you look at next?
Goldstein: One thing that I think would be very interesting — the data right now is not available, but maybe this will give people an appetite to go look for it — is look at this across different countries. Because overall the robust relation between the credit market and the economy has been examined and tested across different countries.
What we do here is only for the United States, because this data is only available in the U.S. So clearly taking it a step further, one would like to see that this is not just a U.S.-specific relationship, but also something that you can extrapolate across different countries. But this is a very long term-project.