Martin Steward on 35 years of curve flattening.
After watching 10-year US Treasury rates flirt with the sub-2% threshold and UK gilts hit Victorian levels yesterday, I thought it might be interesting to look at previous instances of yield curve flattening and think about what they suggest about today’s.
First, let’s define our terms for the sake of clarity. The yield curve is usually described as ‘flattening’ when the spread between the yield on an issuer’s two-year bond and 10-year bond (and/or between its two-year and 30-year) is decreasing. There are two types of curve flattening: ‘bear flattening’ is when the nearer end of the curve rises more than the longer end (so most of the flattening is accounted for by a rise in shorter-dated rates, or a bond sell-off); ‘bull flattening’ is when the longer end of the curve falls more than the nearer end (so most of the flattening is accounted for by a fall in longer-dated rates, or a bond bull market).
Since the beginning of July, we have seen a ‘bull flattener’. Two-year yields have fallen by 32 basis points, but 10-year yields have fallen by 119bps and 30-year yields by 97bps.
Going back to 1977, there have been many periods during which the US yield curve has flattened (for the purpose of this blog, I looked at the 2Y/30Y spread, but the correlation of that spread with the 2Y/10Y is very high). Between April 1977 and the end of 1978, the two-year yield rose 425bps while the 10-year rose 190bps and the 30-year 130bps. There was a similar move over the late summer of 1979. Through the second half of 1980, the two-year rose 616bps while the 30-year moved up only 253bps. These were all bear flatteners. I found another nine significant bear flattening periods up to the present day (‘significant’ in the sense that one can eyeball them on a chart - which generally means extending beyond about 20 trading days). The most pronounced lasted from June 2003 until February 2006 and saw the 2Y/30Y spread narrow by 449bps.
That tells you most of what you need to know about curve flattening. Usually, it’s the bear variety and associated with periods of economic recovery. That stands to reason. One expects most of the movement on a yield curve to occur at the shorter end (it takes either extremely upbeat or extremely gloomy assessments of long-term economic prospects to effect a big move in 30-year rates), so as central bank rates rise to keep a lid on expansion, the curve (bear) flattens, and as they fall to stimulate a faltering economy, the curve (bull) steepens.
So if bear flattening indicates growing confidence in an economy and bull steepening growing anxiety, what does bull flattening indicate? It’s not good because it suggests significant downward movements in long-dated yields - or gloominess about long-term economic prospects. But how gloomy? Well, one way of assessing that is to consider other periods of bull flattening from the past 35 years.
This is where it gets worrying. The fact is, significant periods of bull flattening seem to be very rare. I found a 77bps narrowing of the 2Y/30Y spread in November 1982, coinciding with the tail-end of the long early 1980s recession in the US. Similarly, while the long trend between October 1998 and May 2000 was, overall, a bear flattener, within that, the spring of 2000 did see a 44-day bull flattener of the same magnitude of the 1982 move. This coincided with the top of the dotcom equity bull market.
More significant was the bull flattening that began in September 1985 and ended in May 1986, which saw the 2Y/30Y spread reduce by 159bps and included a 311bps drop on the 30-year yield (and was followed by another, less pronounced bull flattening during Q4 of 1986). What was happening in the wider economy at the time? US growth was slowing down from the heady heights of the 1983-4 recovery.
So while bull flatteners do indicate gloominess, it has not been depressive gloominess. But the point, I think, is the rarity of those bull flatteners.
Between 1977 and 2001, I found three (maybe four if you count September 1985 to end-1986 as two separate trends). I could find no significant bull flatteners during the 1990s. The pattern returns in - you guessed it - November 2008, when we see a whopper in terms of the size and speed of the flattening: the 2Y/30Y spread narrowed by 128bps over 31 trading days and the 30-year yield dropped 161bps (from levels already less than half those of the 1985-6 period).
But 2009 was a year of relief and optimism, right? Maybe, but that didn’t stop the occurrence of another, summer bull flattener. The curve flattened again between February and August 2010 - 82bps over 136 trading days - and, yes, that was another bull flattener, which saw 30-year yields drop 121bps. And finally we come to the latest bout - a 65bps narrowing of the 2Y/30Y spread and an 87bps narrowing of the 2Y/10Y over 34 trading days (and counting).
So, from three bull flatteners over 24 years, we have gone to four over just 34 months. It’s important to note that this has not been one long, volatile, 34-month bull flattening (in the sense that the 1985-6 move was a long, 16-month bull flattener). The 2Y/30Y spread has in fact widened since November 2008, even taking the latest flattening into account.
This is a game of cat-and-mouse between the Fed and investors: rates get slashed and QE gets launched (twice) in a desperate attempt to steepen the curve, push capital into risky assets and put the wealth effect under our sluggish economy; but then the next bit of terrible news comes out, and investors run back to those US-dollar cash flows and further out on the curve in a scramble for yield - however paltry.
The most worrying aspect of this latest flattening? As well as its speed and magnitude, which recalls 2008, it has come immediately upon the Fed’s latest wheeze - telegraphing its intention to keep rates at zero for two years - which was supposed to have led to at least a bit of steepening.
Of course, one can easily argue that all of this is technically interesting, but fundamentally not very informative. What can the relative movements of long bond yields tell us when short rates are anchored so solidly at zero? Economic research house GaveKal has been lamenting the ‘looser-for-longer’ policy for some time now, and for many reasons - one of them being that, “as Japan showed, when short rates get to zero, the yield curve loses most of its ‘predictive’ power”.
But surely the keyword there is ‘Japan’. The predictive power that the US yield curve has for us today is the extent to which the US (and by extension the global) economy is heading toward its own ‘Japan syndrome’. After all, the most dramatic bull flattener in history surely remains that of Japan’s curve over the past 20 years. With that in mind, the probability that the bull flattening of the past 34 trading days is the beginning of a much longer trend - with no steepening relief rallies to interrupt it - seems much greater.
I will leave you to ponder what that means for the debt dynamics of sovereigns, and the balance sheets of banks.