TREND WATCH: What’s Happening? The spread between long- and short-term bond yields has hit its narrowest point since the summer of 2007. Media outlets and policy experts fear that a few more hikes of the fed funds rate will send the term spread into negative territory, a scenario which has preceded each recession ever since Elvis recorded “Heartbreak Hotel.” Others aren’t convinced that the term spread-business cycle relationship will hold up this time around, for reasons including the outsized impact of quantitative easing, the low relative level of interest rates, and negative “term premia.”
Our Take: The claims of these critics don’t hold up to scrutiny. The term spread reflects a rich variety of business-cycle dynamics—everything from central bank policy-setting, the timing of firm capex, and bank lending to consumer sentiment and rational investor expectations about the future. So when the yield curve inverts, the negative term spread cannot be explained away by any one errant variable. All eyes are now on the Federal Reserve, whose growing inflation hawkishness may be locking it into a rate-hike playbook that does indeed invert the curve and (alas) send us into the next recession.
Is a U.S. recession imminent? That may depend on what you mean by “imminent.” The term spread between 10Y and 2Y Treasuries is rapidly shrinking—and is on pace to enter negative territory by Christmas of this year if its trajectory over the last 18 months continues.
Such an inversion would in turn spell trouble for the U.S. economy because the term spread is one of the most trustworthy of all leading indicators. Some forecasters prefer to use the 10Y-90D spread, which offers a more complete historical picture. Others use spreads with slightly longer-dated maturities (10Y-1Y or 10Y-2Y), which invert earlier than the 10Y-90D metric. But whichever spread you’re using, the fact remains: Each time a major term spread has turned negative since just before the 1957-58 recession, a recession has ensued within the next two to six quarters.
Over these same sixty years, there has only been one “false positive”: In 1967, after several Fed hikes put a nasty squeeze on investment, GDP faltered—without however triggering a recession.
In earlier decades, as far back as the data allow, the same basic business-cycle pattern in term-spreads prevailed: Spreads widened coming out of recessions and narrowed at the end of recoveries. But exact “inversion” comparisons before the late 1950s are difficult for many reasons—including reshuffled bond maturities; the Fed’s subservience to the Treasury during World War II and the Korean War; the severity of the Great Depression; and different tax treatment for short- and long-term government debt.
Unmistakable inversions, however, did occur. The Crash of 1929, for example, was preceded by a massive term inversion: For nearly two full years (January 1928 to November 1929), 3- to 6-month Treasury yields exceeded long-term Treasury yields—and did so by an average of 91 basis points. The maximum inversion, reached in the summer of ’29, hit an impressive 145 basis points.
Term spreads don’t just predict recessions. They also track closely with future economic growth—suggesting that, even if the yield curve doesn’t invert in 2018 or 2019, the course is already plotted for an impending business-cycle slowdown. Plenty of experts and media outlets are thus pointing to the narrowing term spread as a giant amber warning light. Others aren’t convinced, declaring that things are different this time around.
So who’s right? Let’s investigate.
TRACK RECORD AS A LEADING INDICATOR
Economists and academics have been studying the relationship between the yield curve and the business cycle since the LBJ administration. In 1965, University of Chicago economist Reuben A. Kessel became one of the first to tie term spreads to phases of the business cycle. He wrote that, since the end of WWII, the gap between 20Y and 90D yields tended to widen coming out of a recession and to narrow heading into a recession, with outright inversions only occurring at business-cycle peaks.
More recently, the term spread has become a staple of econometric models that attempt to predict oncoming recessions. In 1998, researchers Arturo Estrella and Frederic S. Mishkin tested various leading indicators to gauge which ones were best at signaling future recessions. They found that the 10Y-90D term spread performed better than any other indicator, including variables such as stock prices and changes in the money supply. The Federal Reserve Bank of New York even publishes a model that uses the 10Y-90D spread to predict the likelihood of a recession occurring within the next 12 months.
There’s a reason why forecasters are in love with the term spread: Historically, this variable has been an almost flawless recession indicator. Researchers at the Federal Reserve Bank of San Francisco noted in March that every U.S. recession since 1955 has been preceded by a negative 10Y-1Y term spread. They also noted (as we have seen) the single false positive in 1967. The rule appears to hold even when substituting 1Y Treasuries for 90D or 2Y Treasuries.
To be sure, this sort of binary “probit” modeling has its downsides. After all, there have been only a handful of recessions in the past half-century—and thus only a handful of data points on which these models can be trained. Another drawback is that the odds of a recession in these models don’t rise dramatically until the next recession just about underway. It’s like turning the signal lamp red when the locomotive is only 30 seconds away from an oncoming train.
But, as it turns out, the term spread is useful not just as a recession indicator. It is also a superior leading indicator of future GDP growth. In 1988, Robert D. Laurent compared the term spread between 20Y Treasuries and the federal funds rate with three other leading indicators—the level of the nominal federal funds rate, the level of the real federal funds rate, and the growth rate of the money supply (real M2). He found that, between 1964 and 1986, a predictive model based on the term spread would have been the best at forecasting near-term real GNP growth.
We tested this idea ourselves by comparing the 10Y-90D term spread with future GDP growth. Going back to 1959, we see that the term spread decently tracks the 6Q-forward GDP growth rate, both in direction and magnitude. This relationship puts to rest at least one big criticism levied against the term spread—that it is not informative until it turns negative.
The term spread stacks up well against other leading economic indicators. Unlike many of the others, which hinge on monthly or even quarterly data, the term spread is real time. It can be calculated at any moment. It is also context-agnostic: Policy decisions and macroeconomic trends which may affect the absolute level of interest rates, for example, often have little impact on the spread between rates.
Let’s compare the term spread with two other trusted leading indicators. First, we look at the “consumer confidence spread,” which is the difference between people’s future expectations and their perceived current condition. (Once people start feeling that they’re doing much better today than they will be doing tomorrow, a recession is typically close at hand.) Over the past several decades, the term spread and the confidence spread have moved in lockstep. Second, we look at the 3-month average initial jobless claims. This is a decent indicator, but not as good as the term spread—since jobless claims sometimes don’t begin rising on a YoY basis until well into a recession.
WHY IT WORKS
Normally, a long-term security carries a higher yield than an equivalent short-term security, even in the absence of default risk (a plausible assumption for U.S. Treasury debt). This higher yield, reflecting a “positive term premium” or bias, compensates investors for the downside risks of locking their money up longer term—risks that include crisis illiquidity and rising inflation expectations. Most of the time, in other words, investors demand a premium in return for lengthening the duration of their commitment.
But most of the time is not all the time. Why does the yield curve sometimes invert? And why is an inversion so closely associated with the next recession?
Business-cycle effects: central bank policy. Monetary policy is the most direct and obvious driver of the yield curve cycle. Going into a recession, the central bank serves as a “lender of last resort,” dramatically lowering short-term rates to ensure that commercial banks remain liquid and consumers remain inclined to spend. Conversely, when the economy is running hot, the central bank hikes short-term rates to curb aggregate demand and thereby stay ahead of inflation.
While the long end of the yield curve is not immune to Fed manipulation (witness all the QEs in recent years), it is mostly driven by investor expectations stretching far beyond the current phase of the business cycle. It is therefore slow to respond to Fed policy. The short end, however, is closely riveted to the Fed’s interest-rate target (typically, the federal funds rate). So when the Fed puts on the brakes and raises the short end 200 to 400 basis points over the CPI, that usually pushes the short end higher than both long-term future expectations of short-term rates and the term premium.
In other words, the inverted yield curve happens when the Fed is throttling the economy out of fear of rising inflation. That throttling in turn initiates the next recession—hence the adage that expansions never die of old age but are rather “killed by the Fed.”
Business-cycle effects: the real economy. Real economic activity has an additive effect on the yield curve.
Over the course of the business cycle, the national savings rate rises. When the economy just emerges from a recession, there are lots of borrowers—households, firms, and (especially) the federal government—and not many savers. Even with weak capex, this demand for credit and liquidity stresses credit markets. While the short end is held low by the Fed, the long end is kept higher than it otherwise would be by the multitude of borrowers. These include firms and households that want to take advantage of relatively low long-term rates by refinancing. The duration of corporate bonds typically jumps early in a recovery.
All this reverses late in the business cycle, when there are more savers and fewer borrowers, when employment and production are decelerating, and when long-term financing seems a lot more expensive.
Early in the cycle, the yield curve is steep but still low at the long end, encouraging deals from borrowers who think it can’t go much lower. Late in the cycle, the yield curve is flat but still high at the long end, discouraging deals from borrowers who think it can’t go much higher.
Business-cycle effects: commercial bank lending. Consider how commercial banks operate. They are in the business of “maturity transformation”—taking short-term credit and transforming it into long-term loans. The more dramatic the upward slope of the yield curve, the more profitable this business is. But what happens near the end of the business cycle as the yield spread flattens and ultimately inverts? The industry’s “net interest margin” (the difference between income from loan assets and payments to depositors and other creditors) gets squeezed. Soon, banks are paying out more in interest on short-term loans than they are receiving in interest on long-term loans. This eliminates the incentive to lend, which suppresses economic activity.
Business-cycle effects: investor speculation. Finally, let’s look at speculation, which both amplifies the term-spread cycle and accelerates its timing. This is in some ways the most decisive force, because global speculators have bottomless pockets when they sense they can make a sure duration bet.
What’s the rationale for this sure bet? It comes from traders knowing or at least sensing where they are in the business cycle. When traders think that a recession has just passed and the worst is over—and that the entire yield curve is about to start rising—they will short the far end and park their extra money in the near end. They will simultaneously make money while steepening the curve.
Likewise, when they think that an expansion is getting old and the best is over—and that the entire yield curve is about to start sinking—they will go long the far end and avoid the near end. Once again they will make money, only this time they will help to flatten or invert the curve.
RESPONDING TO THE SKEPTICS
Most economists and policymakers are aware of the historical relationship between the term spread and the business cycle. But not everyone agrees that it still applies today. Let’s listen to the skeptics.
According to the Street, this time is different because the economy is still humming and rates are still low. They tell us to pay particular attention to the low overall level of the yield curve. So long as rates are sub-3%, they say, an inversion is no cause for alarm. Others are pointing to negative term premia as a positive differentiator. Still others, looking for “special” reasons why the yield curve is flattening, are pointing to QE or global turmoil and or big bond purchases by insurance companies. The Fed, meanwhile, wants to distract the pessimists by pointing to a new, shorter-term recession bellwether.
Let’s respond in order.
Sure, the economy is still expanding and earnings are still rising. But when you’re certain that it’s no longer doing either, you have already lost in the marketplace. That’s the whole reason why investors look for leading recession indicators. (We will say more below about how just far the inversion leads the recession.)
Is it relevant that rates are generally lower today than in past business cycles? Probably not. Analysts at the San Francisco Fed specifically examined this question. They found that only the difference between short and long interest rates—not the level of the rates themselves—matters for recession and GDP forecasts. Refer back to Chart 3: Any time a yield curve inversion occurs, whether rates are at 5% or 9% or 15%, it is followed by a recession. In 1956, in the inversion that preceded the 1957-58 “Sputnik” recession, short-term Treasuries shot past 20-year governments at only 3.3%. That’s not much higher than we are today.
Many plausible reasons can be (and have been) adduced to explain the generally lower level of interest rates during the current expansion. Heading the list would be lower inflation and inflation expectations. Other reasons might include a greater supply or a lower demand for global savings, driven perhaps by demographic aging, slower productivity growth, or technological change. But it’s not clear how any of these forces would affect the relative yield-curve dynamics of the business cycle.
Other skeptics are pointing specifically to the anomalous plunge over the past two years in the so-called “term premium.” This difficult-to-measure entity (it can only be estimated by surveys, not through market prices) suggests that 10Y yields are now very low in comparison to investor estimates of future short-term yields. According to the popular New York Fed measure, the 10-year term premium is currently around minus 50 basis points. This means that investors are accepting a half-percent less in their 10Y yield than they are expecting on the future rollovers of 30D T-bills. Yes, this is weird: Term premia should normally be positive.
So what’s going on? Hard to say. It could be that the New York Fed measure is inaccurate. (There are competitors; here is one that generates a higher number.) And, even if the measure is accurate, one wonders how we should interpret its significance. In their historical analysis of term inversions, the San Francisco Fed analysts wondered the same thing. After splitting up long-term yields into “term premium” and “expected future short-term rate” components, they found that neither component mattered more than the other in their forecasting models—it was the overall term spread that mattered as a predictor. In other words, it doesn’t matter why the term spread is falling, only that it is falling.
The skeptics won’t have any of it. They think today’s negative term premia point to historically anomalous market conditions. They say that long bonds are getting starved on the supply side by years of Fed QE and are getting overwhelmed on the demand side by worried global investors looking for a safe haven and glutted pension and insurance plans forced to match assets to liabilities. The implication is that, once this transient market imbalance expires, long-term yields will surge, and the risk of inversion will disappear.
That’s one interpretation. But here’s a different one. Maybe the market accepts all of the above. But maybe it also believes that, with the Trump fiscal stimulus in high gear, clear and present inflation dangers will force the Fed to accelerate its rate hikes even sooner than the negative 10-year term premia disappear. And maybe it also foresees divergent (and not convergent) global “slowth” plus a dramatic slowdown in U.S. employment growth (see: “Howe: Unemployment At 20-Year Low. What’s Next?”) plus a splendid little trade war—all of which will put the economy on the brink of recession in 2019.
Market reaction? Don’t fight the negative 10-year term premium. Endorse it. Fall in love with it. If the 10Y yield does not rise much between now and then (and who knows exactly when “then” will arrive), you are perfectly positioned to profit. As for the impact of QE, this is entirely hypothetical. QE is now historical fact, and the Fed’s current QT campaign is on track to reverse it so slowly (raising 10-year yields by an estimated 43 basis points over the next four years) that the plan seems almost certain to be superseded by future events. Care to bet whether the Fed will confront another crisis before July 2022?
According to this contrarian interpretation, in other words, the market isn’t worrying very much about insurance companies or QT or safe havens. Sure, they have some effect in pulling the long end down—but that effect isn’t going away any time soon. Ditto for yet another driver of negative term premia: the rise of risk-parity investing, which is driving up the demand for long-term Treasuries due to their deliciously negative correlation with equities in recent years. This too won’t last forever. But again it probably won’t fade until the markets crash or the inflation rate really jumps—in which case all bets are off.
What’s more, so long as we are entertaining theories about why long-term yields could soon rise, let’s take a look at the other side: reasons why short-term yields may rise even faster. In case you haven’t noticed, the U.S. economy—with eight years of expansionary running room and fiscal afterburners—is now steaming ahead at full employment. According to the CBO, in fact, the economy has been running “above potential” since last spring. Both inflation and inflation expectations are accelerating. But the Fed’s target short-term (fed funds) rate remains well under current (CPI) inflation.
How often has this happened historically? How often, to be precise, has the U.S. economy experienced a positive gap above potential while the fed funds rate has been lagging significantly behind the CPI? The answer is: Never. The only possible exception came in the 1970s, when the Fed clearly let inflation spiral out of control.
Even after two more hikes by the end of 2018, the Fed will still have its heavy foot on the monetary accelerator while the economy continues to race forward. By “race” we don’t mean setting any real GDP growth records (that depends on productivity and demography). Rather we mean straining at the outer limits of its production potential.
THE FED’S CONUNDRUM
Here, then, is the case for those who argue that the yield-curve inversion remains a meaningful indicator. The FOMC is not speeding up its rate-hike schedule because it thinks the inversion theory is wrong. These policy wonks, career bureaucrats, and bank presidents are speeding it up because they are increasingly convinced that they have no other choice.
It is one thing to be blamed for helping to trigger the next recession by attempting to get out in front of quickening inflation. It is quite another to be blamed for unleashing accelerating inflation expectations that could require excruciating monetary medicine, including perhaps a severe downturn and years of structural damage to the economy. “Heedless to obvious signs of an overheating economy and overvalued markets, he repeated the loose-money excesses of Arthur Burns and led America into another decade of grinding stagflation.” That, surely, is not the epitaph Fed Chairman Jerome Powell is looking for.
Fed officials are long-term, cost-benefit pragmatists. Milton Friedman, the patron saint of every monetarist, once declared that “the notion that we can avoid a recession indefinitely is, I believe, wishful thinking.” The Fed as an institution agrees. Sometimes the risk of recession is the necessary price you pay for keeping inflation in check; without paying this price, the next recession could be far more severe.
So even as the yield curve flattens, the odds of faster rate hikes are rising, not falling. By September 26, the market odds of a third hike in 2018 are now 86%. By December 19, the odds of a fourth hike have soared to 56%. That would take us to the 2.25-2.50% range.
Further into the future, the Powell Fed shows few signs of easing up. According to the latest dot plots, the median FOMC member thinks the fed funds rate will reach 3.0% by the end of 2019 and 3.5% by the end of 2020. Only by 2020, then, does the FOMC finally expect its short-term rate to exceed its current “long run” rate expectation (3.0%) to a significant degree (50 basis points). Ditto for the gap between its short-term rate and inflation expectations by end of 2020 (2.2%). That’s a gap of about 130 basis points.
But let’s be realistic: The Fed may have to move quite a bit faster than this. As a historical rule, the Fed feels compelled to push short-term rates much higher than current inflation whenever the economy is overheating—higher by 200 to 400 basis points. And it feels compelled to raise them quickly.
What’s more, the Fed’s favorite inflation indicator, the “core” PCE deflator, has already hit 1.96% YoY growth and is accelerating fast. Its monthly rate has recently been averaging around 2.5%. Let’s say it starts tracking around 3% YoY in early 2019—and a nervous Fed, turning in an ever-more hawkish direction, feels it needs to go 200 basis points over that. So is a 5.0% fed funds rate by the end of 2019 unthinkable? By no means. It’s only unthinkable to those who have been lulled into complacency by the exceptional torpor of our long post-GFC recovery.
One development that would rule out any such accelerating rate-hike path, of course, would be clear evidence that the economy is slowing to stall speed. (By 2019, this evidence may be hard to assess due to a sharp and unavoidable deceleration in employment growth.) But of course this would be the one development that tells the inversion school that they were right all along.
The Fed is standing at a crossroads. Path One is to stay ahead of inflation by hiking short-term rates—even if it hastens the arrival of the next downturn. Path Two is to avoid a yield curve inversion by delaying rate hikes, or even reversing them, in an effort to stave off the next downturn.
Given the term spread’s perfect history as a recession predictor, Path Two would seem to be a reasonable course. But even as some eminent leaders within the Fed acknowledge the yield curve’s impressive track record, many others brush aside its significance. Some even want to scrap the term spread as a leading indicator entirely and want to put in its place a nearer-term 3M T-bill spread (6Q forward minus 1Q).
This new indicator defies reason—since the 6Q T-bill “spread” is basically just the market’s perception of upcoming Fed policy. Wrap your mind around this: The Fed is suggesting, as an alternative to the standard term spread, that the market instead watches its own expectation of what the Fed is about to do! The Fed then points to this indicator as assurance that, right now, the market expects the Fed to keep hiking, which means that a recession is not imminent. If you suspect that the Fed is totally obfuscating the essential logic of term-spread dynamics, you are correct.
The Fed is twisting and turning on this indicator. It trying to distract the market from any talk of future economic weakness. Because it knows that inversion psychology is self-fulfilling, it is trying to get investors to focus entirely on its own determination to fight inflation. Bottom line: The Fed is girding itself to press on, undaunted, with its rate-hike agenda.
We’ve already discussed what’s motivating the Fed. Let’s recap it here in three steps. The Fed recognizes (a) that inflation is accelerating; (b) that by any historical standard it is way behind the curve in pulling it in; and (c) that failure to control inflation could lead to unstable stagflation and its various crash-prone sequelae—and thence to a catastrophic loss of support for the Fed by the public and Congress. The Fed absolutely cannot risk going there.
Meanwhile, as investors begin to sense that the Fed is locked into this inflation-fighting strategy, many are becoming more comfortable moving into long bonds even as inflation still heats up. If investors regarded the Fed as just another player, game theory suggests that they should be careful—since the Fed could always pull the rug out from under them before the inversion fully matures. But in this case, the game is asymmetrical: Investors pretty much know that the Fed can’t take that risk. It’s like a prisoner’s dilemma in which one person knows the other won’t bluff or snitch. That advantage serves as an automatic get-out-of-jail-free card.
If the inversion school is right, what does this imply for the timing of the next recession? As we’ve already seen, a straight-line 18-month trend in the 10Y-2Y yield spread gets us to inversion by Christmas 2018. If that happens, the model suggests a recession starting anywhere from July 2019 to July 2020.
To get a better feel for where the economy may be heading, let’s track the last twelve years in two dimensions: term spread and VIX. And then let’s track it in another two dimensions: term spread and credit spread.
Either way, we see an economy that has about two quarters to go before coming “full circle” back to where we were in mid-2006. Imagine Powell slowing down (or stopping) rate hikes in 2019 after the curve inverts. Then imagine the economy sliding into recession in early 2020, setting the stage for a raucous election year—just as the Crash of ‘08 lent a grim backdrop to the Obama-McCain debates.
This situation isn’t unsalvageable for equity investors. A recent BMO Capital Markets analysis shows that stocks can still perform well in the flattening phase of the yield curve. But the real opportunity is clear. Once the curve fully inverts, you want to be long in long bonds. And you want to start battening down the hatches on everything else.
