How central banks distort the predictive power of the yield curve

Gaps between long- and short-term bonds may flatten for financial not economic reasons

Megan Greene


As the Federal Reserve raises interest rates during an economic recovery, the short end of the yield curve goes up. But the central bank has little control over longer-term rates © AFP


I have a bet with some colleagues that we will see the yield curve invert before the end of next year. This hardly seems a bold call given that the gap between two-year and 10-year Treasury yields narrowed to below 20 basis points last week to a post-financial crisis low. So I’m going to propose another question on an even thornier issue — if the yield curve inverts, will we actually care?

In theory, we should. The yield curve charts the difference in compensation that investors receive for holding debt of different maturities. It usually slopes upwards to the right, to reward investors for locking up their money for longer. But if investors think the economy is about to go into recession, they demand a higher premium for holding short-term bonds than longer-term debt and the curve is inverted.

Yield curves have been better than most economists at predicting US recessions, although there is some debate about which part of the curve to use. Investors typically focus on the gap between two- and 10-year yields, while academics favour the spread between three-month bills and 10-year notes.

A paper from Federal Reserve board researchers suggests that a spread of short-term interest rates may be best. All three metrics turned negative a year or two before each of the seven US recessions since the 1970s. You can argue that one curve is better than another, but all of them put us economists to shame. As the Fed lifts interest rates during an economic recovery, the short end of the yield curve goes up. But the central bank has little control over longer-term rates. Those are determined by expectations for short-term rates, inflation and the “term premium”, the bonus investors demand for the risk of holding an asset for a longer period. If the economy seems to be faltering, investors expect inflation to ease and so longer-term yields fall.

While the yield curve has a successful record predicting downturns, seven recessions is not a huge sample size. Some economists argue that policy shifts and structural changes mean the yield curve’s signal is now distorted. The US Treasury is currently borrowing more money to finance predicted big budget deficits and it has mainly done so with short-term debt. That increase in supply has pushed the price of bills and short notes down and their yields up (bond prices and yields move inversely).

At the same time, global quantitative easing has created a seemingly insatiable demand for five- to 10-year Treasuries, pushing down yields. This means that the yield curve may be flattening for financial rather than economic reasons. Meanwhile, expectations for the neutral rate (when real gross domestic product grows at trend while inflation remains stable), inflation and term premia have been consistently lower than before the financial crisis. That means longer rates have been lower, making it easier for the yield curve to invert as short rates continue to rise.

Finally, deflation fears have made long-term bonds a useful hedge for equity investors concerned about a stock price correction. That also boosts demand and represses long yields.

All of these factors make it easier for the yield curve to invert, but none seems to reflect particular concerns about the economy. An inversion this time need not necessarily indicate a recession is nigh. I am sympathetic to the arguments that this time is different, but reluctantly so. Do not forget that in 2006, then-Fed chair Ben Bernanke insisted the yield curve’s signal was being distorted by structural factors, two years before the 2008 crash.

It may not matter whether the yield curve’s predictive powers are distorted. If the markets, companies and individuals believe an inverted yield curve means there will be a recession, they will behave accordingly. An economic downturn will then become a self-fulfilling prophecy — and I will win my bet.


The writer is global chief economist at Manulife Asset Management

America’s Global Engagement

The myth of disengagement derives from American rhetoric and not American actions.

By George Friedman

Is the U.S. Headed for Another Mortgage Crisis?




Ten years after the mortgage-fueled Great Recession, several of the market and structural components remain in place that could set the environment for the next crisis. In her latest research, Wharton management professor Natalya Vinokurova takes a historical look at the development of mortgage-backed securities and finds fascinating parallels to the present day. She spoke to Knowledge@Wharton about her papers, “Failure to Learn from Failure: The 2008 Mortgage Crisis as a Déjà vu of the Mortgage Meltdown of 1994” and “How Mortgage-Backed Securities Became Bonds: The Emergence, Evolution, and Acceptance of Mortgage-Backed Securities in the United States, 1960–1987,” and why one should heed the warnings of history.

An edited transcript of the conversation follows.


Knowledge@Wharton: The inspiration for this research goes back to the introduction of mortgage-backed securities. Can you tell us about that?

Natalya Vinokurova: One of the things many people do not realize is this last mortgage-backed securities market got started in the 1970s. Specifically, the goal was very much to find funding for the baby boomers as they were buying houses. What’s interesting about the market is that for the first 15 years or so of mortgage-backed securities being around, bond investors did not believe that these were bonds. The big project was convincing bond investors that they could treat mortgage-backed securities as bonds. One of my papers on this topic looks at the process by which mortgage-backed securities issuers convinced bond investors that these were, in fact, bonds.

Knowledge@Wharton: What did they do?

Vinokurova: Mortgage-backed securities and bonds are different on a number of dimensions.

The dimension that bond investors were most concerned about in the 1970s and early 1980s was prepayment risk, with the idea being that mortgage-backed securities are bundles of mortgages. Your average borrower can repay his or her mortgage at any time. Most of the time, these borrowers would not incur a penalty. As an investor, this meant that if you were buying a security backed by 30-year mortgages, there was a very small chance that the security would still be around 30 years out. It could have been repaid in seven years, 12 years. There was a lot of uncertainty about when these mortgages would actually be repaid. As a bond investor, somebody’s trying to sell you something and they can’t even tell you how long this thing is going to be around. Obviously, bond investors pushed back.

One of the interesting things about my research is I get to go back and look at the various mortgage-backed securities that issuers tried to sell to bond investors. I documented at least seven or eight different types of mortgage-backed securities, and that’s just what was offered to the public. I have no way of tracking the many private experiments.

Knowledge@Wharton: How did market participants come to accept these mortgage-backed securities as being like bonds?

Vinokurova: An important step in the acceptance of mortgage-backed securities as bonds was developing tools that bond investors believed would manage prepayment risk. The tools that won the game, which were introduced in 1983 in a public security called the collateralized mortgage obligation, was tranching. The idea behind tranching is that you can slice investors into different classes. Tranche is the French word for “slice.” Each of these slices of investors would theoretically be exposed to different levels of risk. You had the junior tranches, which were supposed to absorb the risk. The senior tranches were protected by the fact that you had the junior tranches as part of the security.
The security that started it all only had three tranches. By the early 1990s, you had securities with something like 68 tranches…. But the precursor for this flourishing of all these tranches was when Freddie Mac issued the collateralized mortgage obligation in 1983. It was the first publicly issued security that used tranching. Bond investors said, “We now believe you,” because each of these tranches was given a ballpark range of repayment. If you were a pension fund looking for longer-term securities, you knew that the security you were buying would be safe from prepayment risk for the first five, seven, however many years you needed.

Knowledge@Wharton: Now we get into the 2000s and the financial crisis. You found that past experience didn’t play very much of a role in what happened next, correct?

Vinokurova: In the early 1990s, the Fed undertook a series of interest rate cuts, so just about every mortgage borrower in the United States had an opportunity to refinance their mortgage. The reason they had this opportunity is because the bond investors’ capital flowed into the mortgage market. Prior to the bond investors believing that mortgage-backed securities were bonds, you had these concerns about insufficient financial capital flowing to the mortgage market. Once the bond investors believed these things were bonds, once they believed that tranching would work, just about enough money flowed into this market to enable as many people to refinance as wanted to. You had 70% of all borrowers repaying their loans in some securities. 
What that meant was no matter how well the tranching of prepayment risk was structured and how much fancy math went into it, at the end of the day, the junior tranches disappeared. They were kind of overwhelmed by the risk. The senior tranches found themselves vulnerable. If you were a pension fund investor who was comfortably sitting in the knowledge that whatever it was you owned would not be repaid for the next five years, you found yourself in the same boat as the junior tranches.

What I argue in the paper is the series of events — starting with the faith and the efficacy of tranching, leading to the influx of bond investor capital into the mortgage market, leading to this self-destructive loop — is exactly predictive of the events of 2008. Moreover, what’s interesting about the meltdown of tranching for prepayment risk is that it is what encouraged people to invest in subprime mortgages because some of these subprime mortgages had prepayment penalties, which prevented the borrowers from refinancing. In a way, the movement towards these nongovernment-backed, mortgage-backed securities with default risk was driven by the fact that they were seen as being a safer bet.
Knowledge@Wharton: What were the most surprising conclusions in terms of how market participants reacted to this?

Vinokurova: One of the reasons why people did not update their beliefs about whether mortgage-backed securities were bonds was because people constructed narratives that were very specific to the crisis. The early 1990s events were variously called “the meltdown” and “the mayhem” of the mortgage market. But the explanation that got constructed for the mayhem had to do with the fact that mortgage lenders lowered the fees associated with refinancing. Instead of blaming the use of tranching for what happened, instead of seeing the systematic causes of the events of the 1990s, the market participants settled for this very local explanation.

I think we see something similar with the 2008 crisis, where instead of looking for structural causes, what we see is a search for this very specific explanation.

Knowledge@Wharton: Why do you think people want that very specific explanation? Is it because that seems easier to fix than a structural problem?

Vinokurova: I think it’s a combination of that. It’s also a combination of the fact that people have a lot of time pressure. People in these jobs, whether they be investors, rating agencies, investment bankers, are all working with time pressure. Frankly, most of them do not last in their jobs for long enough to remember the previous cycle, and the ones who do kind of partitioned their experience. I didn’t really find anybody trying to look for these global lessons.

I think one way in which the failure to look is evident — and this is something I found surprising — is if you examine the arguments in the 1970s in favor of developing these mortgage-backed securities, there is absolutely no reference to the prior U.S. markets in which mortgage-backed securities played a part. One such market developed in the 1870s. Another such market developed in the 1920s. Even as these markets have certain parallels to what happened post-1970s, once the securities were introduced, you don’t see any of these actors making specific appeals to these prior experiences. As they don’t remember the history, they literally repeat it.

For instance, some of the names of the mortgage-backed securities issued in the 1970s are exactly the same names that were used in the 1920s. In 1975, Freddie Mac issued something that was a precursor to the collateralized mortgage obligation, a security that they called a guaranteed mortgage certificate, which is exactly the name that was used in the 1920s by mortgage insurance companies issuing mortgage-backed securities. The parallel does not end there because what happened in the 1920s as part of the mortgage crisis was these mortgage insurance companies went bankrupt, and this is very much what happened to AIG in 2008. The history literally repeats itself.


Knowledge@Wharton: Is there a way to build institutional memory into the system?

Vinokurova: Absolutely. I think a good analogy here is the Food and Drug Administration. This is an entity that tries to force memory. If your drug failed to do certain things or it poisoned people in the 1960s, you can’t reissue it and say, “Oh, let’s do this again.”

In my research, I take the prospectuses — the documents explaining what the securities are and what they should do — as almost a fossil record because that’s my way of reconstructing what these securities did. Just locating these documents is surprisingly difficult. This is even the case within the firms that pioneered these securities. For instance, I contacted Citibank or Citigroup, which bought Salomon Brothers on the merger path a while ago, and I asked them for prospectuses of some of the securities that were pioneered by the Salomon Brothers in the 1980s. The reply I got was one, they couldn’t give me access to their archives because I wasn’t a client; and two, they do not keep their prospectus records in the archive for longer than three months. Now, we are talking about securities with 30-year maturity. The fact that we live in a regulatory environment where an issuer or an underwriter of a security can discard what is effectively the contract between them and the investors seems very surprising.

Knowledge@Wharton: Does there need to be more regulation requiring that this type of documentation be kept?

Vinokurova: Unfortunately for the mortgage industry in the United States, the problem runs deeper than the regulation of documentation. I can think of no other country in the world where you have these fly-by-night mortgage originators that disappear after every bust. And there are trade-offs in terms of wanting more people to have access to credit. We want the credit to be cheaper. But it seems that in designing the system, the access to credit considerations perhaps get prioritized ahead of the safety of the system.
Knowledge@Wharton: There was a lot of finger-pointing during and after the crisis to assign blame. One of the interesting things about your paper is that you find that while people may want to ascribe animus to some of these market participants, that wasn’t necessarily the case.

Vinokurova: Right. The perspective from which I approach my research is to imagine the best-intentioned actor in the system. Imagine somebody at Citigroup who is trying to learn from the 1970s. We are not going to make them learn from the 1920s, but they are trying to learn or build on the knowledge that Citigroup accumulated. At some point in tracking down these various people, you would have to go on the circuitous path of locating the people you would learn from. So, there are systematic problems that do not get addressed. In a way, I feel like the incentives narrative is not helpful. This is not to say that there wasn’t fraud, that there wasn’t ill will, but the system has structural problems. The fact that we repeat mortgaged-backed securities markets every 50 years suggests to me that it takes everybody to die who remembered what it was like, and then we try again.

Knowledge@Wharton: As we’re coming up on 10 years since the start of the Great Recession, there has been a lot of talk about what will cause the next crisis and when. Do you see any warning signs? Are there changes we could make to protect against that?

Vinokurova: I think that house price inflation is an incredibly potent signal of us being in a bubble, of us being on the verge of a crisis. I think the quantitative easing, which is effectively creating liquidity by printing money, has been shown by folks like Markus Brunnermeier at Princeton to bring about crises. When you have too much money chasing too few attractive options, you end up in the bubble. And the bubble will have to burst.
In terms of what we know about righting financial crises, a lot of it is not rocket science. The banks are too big. The banks that were too big to fail 10 years ago are even bigger now. The Consumer Financial Protection Bureau has been rendered ineffective by the current changes, and Dodd-Frank has been rolled back. It was a set of regulations that people didn’t think were strong enough to remedy what we saw happen. I think there are things that can be done, but it’s not clear that the current administration is interested in pursuing these paths.

Knowledge@Wharton: If you had control, what would you do?

Vinokurova: What I would do is further restrain the leverage of the banks. Banking used to be a boring 9-to-5 job. It needs to go back to being a boring 9-to-5 job.

Knowledge@Wharton: What’s next for this research?

Vinokurova: I’m interested in the structural parameters of the system. One of the projects that I just submitted to a journal looks at the history of mortgage ownership recording. One of the very interesting things that came out of the 2008 crisis is that it turned out that, in many cases, banks didn’t know who owned what loan. In my research, I trace the development of that system that keeps track of land ownership recording and mortgage ownership recording to the 1630s. That allows me to say, “Look, these are structural problems. They have been around through these multiple generations of reform.” I feel like one of the challenges reformers often face is that they think they are the first people who tried to reform the system, and I feel like learning from the people who came before them would actually be helpful.

The Chinese Profits Puzzle

Threats to Chinese growth are multiplying but some big companies are still posting their best results in years. What’s going on?

By Nathaniel Taplin