Interest rates and the face/vase problem

The relationship between interest rates and money is pretty confusing.  So I’m not surprised most people are confused.  I’m somewhat confused.  But I’m nowhere near as confused as some of my somewhat confused commenters (Bob Murphy) assume I am.  In some respects it’s pretty simple.  We’ve always known the following:

1.  Moves toward easier money usually lower short term rates.  The effect on long term rates is unpredictable.

2.  Moves toward tighter money usually raise short term rates.  The effect on long term rates is unpredictable.

3.  Extremely easy money policies (hyperinflation) almost always raise interest rates.

4.  Vice versa.

And those are still true.

When I talk about money and interest rates I am sometimes talking about cases 1 and 2 “Interest rates are an unreliable indicator of the stance of monetary policy.”  And sometimes I’m talking about cases 3 and 4 “Ultra low interest rates are a sign that money has been tight.” I should make that distinction clearer.

Nothing that has happened recently has changed my basic views on these 4 points.  We have a long data set to look at, and recent events simply add a few points, consistent with what we already knew.  So why does it look like it conflicts with market monetarism?  Several reasons:

1.  We emphasize the contrarian cases, where tight money lowers rates, in order to differentiate our brand.  People begin to think we believe that always occurs. If we are going to do that, we better be ready to take a hit when things go the other way.  I plead guilty.

2.  We don’t have a good theoretical model explaining why the liquidity effect sometimes dominates at the longer maturities.  We observe that fact, but can’t really explain it.

But remember that I’m 58 years old and have been observing monetary policy my whole adult life.  I certainly knew that tighter money can raise long term bond yields.

Now here’s where commenters seem to get confused.  Even when the liquidity effect is important, other effects are important two.  Yesterday monetary policy returned to the status quo ante.  Last September the Fed told us that QE would be data-driven.  Bond yields were very low, around 1.7%.  (But 10 year bond yields actually ROSE on the news!) In the middle of this year there were hints that the Fed was switching policy, and that they’d taper despite the fact that the data did not call for tapering.  Bond yields rose to a peak of roughly 2.9%.  Then they returned to the earlier policy of a data-driven QE.  Bond yields are now about 2.7%.

Many people assume that tapering rumors caused bond yields to rise from about 1.7% to 2.9%, and then the return to the earlier policy caused rates to fall back to 2.7%.  But that is almost certainly false.  Yes, the decision not to taper did cause rates to fall, but not very much.  That means the hints that they would taper caused rates to rise, but not very much.  Most likely we had a 100 basis point rise in yields over the past year due to macroeconomic forces, and another 20 basis points due to fear of tapering.  Now that 20 points has been unwound.

Note that my hypothesis is not just pulled out of thin air, it’s confirmed by other markets. Consider equities.  We know for a fact that equities were hurt by taper talk and helped by yesterday’s decision not to taper.  Suppose it really were true that taper talk had explained the huge run-up in bond yields over the past year. What should have happened to equity prices? What did happen?  Obviously the same economic forces that were pushing bond yields 100 basis point higher were also pushing the S&P sharply higher.

And we know what some of those forces were.  I recall one strong jobs report (early July?) where bond yields rose sharply, and so did stock prices.

Why did so many people miss this?  Because:

1.  The liquidity effect is real.

2.  Once you start thinking in terms of the liquidity effect, it’s hard to think of anything else (the face/vase problem–cognitive illusion.)

For instance, people will cite the fact that the path of short term rates fell yesterday, i.e. the Fed is now expected to raise short term rates later than before, and assume that means money is getting easier.  As far as yesterday is concerned that’s true.  But only because we know what caused the change yesterday—monetary policy.  Over the long term however, the expected path of shorter term rates is mostly endogenous.  You might think the Fed has “complete control” over short term rates.  But that would only be true if they didn’t care about the macroeconomy. But they do care.  If there are forces expected to raise NGDP growth to excessive levels, the Fed would respond by raising rates.  In that case the “cause” of the future expected higher rates is not an “expected liquidity effect” it’s an “expected income and inflation effect.”

So don’t fall into the trap of thinking that all rate changes reflect Fed policy, just because you clearly observe that some of them do, and also because the Fed has near total control of short term rates in a technical sense.

PS.  Here’s what I said minutes after QE3 was announced last year:

Here’s what the Fed says it’s trying to do:

These actions, which together will increase the Committee’s holdings of longer-term securities by about $85 billion each month through the end of the year, should put downward pressure on longer-term interest rates, support mortgage markets, and help to make broader financial conditions more accommodative.

Nope.  Long term yields increased on the news, just as market monetarist’s would have expected.  And thank God they did!  The higher yields are an indication that markets have (slightly) raised their NGDP forecasts going forward.  The jump in equity markets suggests that RGDP growth will also rise (albeit modestly.)  The bad news is that 100 points on the Dow is indicative of a really small change in the RGDP growth rate, basically within the margin or error.  So we’ll never know any more than we know right now about whether the policy will “work.”  Of course that won’t prevent hundreds of economists from making silly pronouncements a few months from now, based on actual changes in RGDP.  I beg you to ignore them all.

I should not have said “just as expected”, when what I meant was “just as MMs suggest often occurs with easy money policies.”  So the confusion is partly my fault.  I gloat when things go my way.

Note that the instant reaction of stocks is a more reliable indicator of monetary policy that long term bond yields.  Long term rates rose on the announcement of QE3, and rose again on taper talk.  Why is the long term bond market so schizophrenic?  I have no idea.


Tags:

 
 
 

28 Responses to “Interest rates and the face/vase problem”

  1. Gravatar of Ironman Ironman
    19. September 2013 at 05:32

    A couple of visual references – this is perhaps only evident in retrospect, but in looking at the bond market back in June 2013, we see that it immediately jumped about 20 basis points in reaction to Bernanke’s 19 June press conference as the Fed would appear to have lost control over future expectations (it took the Fed about a month of damage control to put the genie back in the bottle).

    Second, in looking at the information that stock prices provide, it’s more of a mixed message – the reaction to the Fed’s announcement yesterday is a lot more muted that it would seem from just the intraday change in stock prices.

  2. Gravatar of Matt McOsker Matt McOsker
    19. September 2013 at 05:36

    The rise in rates coincides with the increase in teh stock market. Portfolio shift from bonds to stocks (or people using their reserves from QE purchases to bid up stock prices?

    http://research.stlouisfed.org/fredgraph.png?g=my8

  3. Gravatar of Benoit Essiambre Benoit Essiambre
    19. September 2013 at 05:42

    Hey how is a jump in the stock market a predictor of RGDP instead of NGDP? Or is it about the ratio between the bond stock market movements?

  4. Gravatar of ssumner ssumner
    19. September 2013 at 05:50

    Ironman, Your first paragraph certainly supports my post. I don’t follow the second paragraph.

    Matt, Portfolio shifts are not really a useful concept, as in aggregate they don’t occur. When one person sells another buys. Focus on what causes the price to change.

    Benoit, I think it does predict NGDP to some extent, but in the short run NGDP and RGDP are closely related.

  5. Gravatar of Morgan Warstler Morgan Warstler
    19. September 2013 at 05:56

    Scott, I get what you’re trying to say, but 3 and 4 and not even near mirrors of each other like 1 and 2.

    Meaning, you are essentially saying that RIGHT NOW is the economic opposite of Zimbabwe.

    then also:

    “If there are forces expected to raise NGDP growth to excessive levels, the Fed would respond by raising rates.”

    Say RGDP is 1.5% and inflation is 3%.

    Say Democrats win both houses of Congress, and announce their intention to double Federal employee wages.

    CBO predicts this will causes NGDP to hit 5.5%.

    The NGDP Futures Market goes bezerk, and suddenly the cost of money / interest rates shoot up.

    The cost of Federal Debt rolling goes up, the CBO announces greater future debt service and thus debt levels.

    All of this happens in an incredibly short time window, right?

    Am I missing anything here?

  6. Gravatar of Benoit Essiambre Benoit Essiambre
    19. September 2013 at 06:19

    I’m still trying to understand the logic. How can we know if the market is predicting real value instead of just nominal increases with these movements?

  7. Gravatar of njk njk
    19. September 2013 at 06:26

    I sat around a table earlier this year and the question was asked, “what if the Fed doesn’t taper?” One person said (paraphrasing), “I don’t rates will fall much because taper will still be looming, but it will simply be delayed.” So the Fed’s taper signal affected the way he thought about rates. Now, I think you’re saying that the Fed has completely restored its credibility. Maybe that’s true to an extent, but whether my colleague’s logic is right or not, rates acted exactly as he predicted they would.

    Tell me why he’s wrong.

  8. Gravatar of Matt McOsker Matt McOsker
    19. September 2013 at 06:28

    Scott, if bond buyers sell that will drive prices down and yields up (stocks have gone up in price with steady demand). Bond holders have been selling, here is the data:

    http://corporate.morningstar.com/us/documents/AssetFlows/AssetFlowsSept2013.pdf

    Stocks are seeing net inflows during this time. Does not take that much demand to bid up prices.

  9. Gravatar of Jonathan Finegold Jonathan Finegold
    19. September 2013 at 06:32

    Is this another way of explaining the same point? Say you’re talking to an Austrian, and we have a hypothetical economy where the “natural rate,” or equilibrium rate, is an r of 1%. Based on this reference, an Austrian would say that a manipulation of the interest rate to 1.5% would be tight money, while decreasing it to 0.5% would be easy money. In our case, there are many who believe that this “natural rate” is below zero, so any rate of interest above zero is going to be the result of tight money. (Of course, we can disagree on where the equilibrium rate lies.)

  10. Gravatar of dlr dlr
    19. September 2013 at 07:01

    Scott, I think everything you said here is roughly accurate but understated. A more precise explanation of why the rate story has made MM look off is because you, for reasons I don’t really understand, seem to have a bias towards downplaying the recent size of the liquidity effect in the short run. Even now your 20bp medium-run story is way too simplified by plowing over interesting short run reactions (which you usually like).

    From March 11 2013 through May 7 2013 the 10-year fell 53 points from 2.06 to 1.63. During that time the S&P rose 4.4% from 1556 to 1625. The 10YR then rose by 90 points to 2.53 by June 24, plainly buoyed by the prospect of taper, while stocks declined to 1573. These are big moves in rates with zero confirmation from stocks that real rates were being driven by anything but a liquidity effect. Sure people can make all kinds of (generally unconvincing) tertiary arguments that other factors besides stocks supported an expectations effect here, but stock market reactions usually appeal to you. In the long run, I strongly agree that it is easy to overestimate the importance of the liquidity effect on interest rates. But I think you mainly hurt your case by (for some reason) refusing to embrace the strong and surprising effect that Fed LSAP moves have had on rates. I think for anyone watching markets it has been a pretty straightforward empirical surprise (to me) that the Fed has had a lot more sway over the entire real rate curve through some kind of segmented market / portfolio balance / who knows channel than expected. I doubt it matters much in the scheme of things, but you still seem to run from it a bit.

  11. Gravatar of phil_20686 phil_20686
    19. September 2013 at 07:02

    It seems to me that the best way to explain it is to say that “better monetary policy always raises the long term real interest rate”. But “better” can mean tighter or looser policy in the short term. We have lived through a decade in which “better” was almost always “looser”, but in the 1970’s, “better” was almost always tighter, and I imagine that if we had TIPS from there, we would see that the real long term interest rates rose in response to tightening, and fell on loosening, but its hard to know as we don’t have the equivalent inflation expectations data.

    I think of the real long term interest rate has basically having an upper limit set by technological progress/demography/supply side, and bad monetary policy just holds it below that rate, better policy can force it up to that rate, but never higher.

    So the real message from the taper/no taper, was that
    (1) Better GDP figures have raised the real interest rate
    (2) Markets still believe that looser policy will raise the real interest rate.

    Also, I don’t think that the minute by minute data is very important on these event studies. In particular because large bond positions take so long to unwind. This means that there is virtually no gain in e.g. PIMCO trying to change its positions on the day of the taper, instead it takes a few days to make a plan and then several weeks to execute it.

    Its quite common in equity and bonds to see traders/HFT move the markets one way on the day of some news, only for a big player to come in to build a stake over the next few weeks driving the price higher, right up until they meet the selling price of another large player.

  12. Gravatar of brendan brendan
    19. September 2013 at 07:14

    Scott, I think you focus too much effort explaining the small price movements. Asset price movements make more sense if you assume that:
    1) Small price movements are noise and don’t need to be explained.
    2) Large price movements in the days and weeks after a policy event- i.e. post-news-announcement DRIFT- are more reliable signals of policy effects.

    Yes, EMH violations. But the EMH regularly is violated by reality in these ways, across time, and across asset classes.

    Think about it- were you surprised equities trended up after QE2? Or after the Evans rule? Or after Abe? You were not; in fact, you noticed Japanese equities diverging in real-time and told us about it. And if you believe in the drift EMH violation, then why not the excess-noise violation too?

  13. Gravatar of John Hall John Hall
    19. September 2013 at 07:23

    I double-checked some numbers and I’m still confused. I think a lot of this would be clearer if we had NGDP/RGDP/PGDP futures markets.

    Last year’s QE3 announcement occurred on 9/13/2012. On 9/12, 10 year yields were 1.758% and TIPS were -0.631%. On 9/13, 10 year yields fell (they only rose initially) to 1.724% with TIPS also falling to -0.755%. A modest decline in nominal yields, but a strong decline in real yields. This means that implied inflation rose to 2.479% from 2.389%.

    On 9/17/2013, 10 year yields ended the day at 2.848% with TIPS at 0.679%. Yesterday, 10 year yields ended the day at 2.689% with TIPS at 0.475%. Strong declines in both nominal and real yields. This means that implied inflation rose to 2.214% from 2.169%.

    So in both periods, the announcement of more QE or not halting QE purchases led to a rise in inflation expectations (consistent with a shift in aggregate demand) and a strong decline in real yields (more of a consistently strong decline in real yields than nominal yields anyway).

  14. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    19. September 2013 at 07:24

    It would be better to simply get out of the habit of talking about monetary policy in terms of interest rates. Completely drop it. Finito.

    Chant every day; ‘Interest rates are not the price of money.’

  15. Gravatar of John Hall John Hall
    19. September 2013 at 07:28

    brendan, 10 year TIPS fell by their most since February 2011. Is that not significant?

  16. Gravatar of Two Views of the Non-Taper Rally | askblog Two Views of the Non-Taper Rally | askblog
    19. September 2013 at 07:32

    […] 1. Scott Sumner: […]

  17. Gravatar of brendan brendan
    19. September 2013 at 07:46

    John, so TIPS implied inflation expectations rose more than they had in nearly 3 years. Notice that an equivalently violent rise in equities implies maybe a 3% up day; but we got only 1%. Seems inconsistent, no?

    I do not suggest that the Fed’s decision was meaningless. I suggest that markets are noisy and that better info will flow from larger movements in the coming weeks.

  18. Gravatar of Brian Donohue Brian Donohue
    19. September 2013 at 08:40

    Excellent blogging! I would say you seem less confused than most of your colleagues. I don’t recall a more lucid description of the difficulty of interpreting changes in long-term interest rates.

    You write: “And we know what some of those forces were. I recall one strong jobs report (early July?) where bond yields rose sharply, and so did stock prices.

    Why did so many people miss this?”

    For the record, I didn’t. 2013 has struck me as a sort of ‘inflection point’ on the road to recovery.

    My business is pensions, so I’ve spent a silly amount of time over the past 25 years looking at interest rates too. Here’s what I said in July:

    “For now, pension sponsors have enjoyed a substantial boost in 2013, with equity values holding up in the face of higher rates.”

    http://www.octoberthree.com/Cms_Data/Contents/o3_cms/Media/Articles-2/Pension-Finance-Update_7_31_13.pdf

  19. Gravatar of ssumner ssumner
    19. September 2013 at 08:50

    Morgan, What if that does happen? I don’t see your point.

    Benoit, I have said 1000 times on this blog it’s criminal negligence that the Fed doesn’t subsidize and run a NGDP and RGDP prediction market, so we can know in real time.

    njk, He’s right.

    Matt, You said;

    “Scott, if bond buyers sell that will drive prices down and yields up ”

    Sorry but that’s a common fallacy. Each day the number of shares bought and sold is identical. That’s the sort of thing people say on Wall Street, but it’s nonsense. It’s not just that it isn’t true, it can’t be true.

    Jonathan, That’s right, but can be misleading. Many people think the Fed just influences the market rate, but they have even more influence over the Wicksellian equilibrium rate. That’s what Austrians sometime miss. (And Keynesians.)

    dlr, You are doing what I warned against. People on Wall Street who claim to watch markets closely assume all rate changes are due to the Fed. So when the expected date of Fed rate rises moves up, they always attribute it to Fed policy, not the economy. In fairness you tried to overcome that by cherry picking some periods where the liquidity effect may have been driving things, and correlating with stocks. The problem is that it was driving things in both directions in your sample period. I don’t doubt that 100, 200, or even 300 basis points of bond yields moves can be attributed to the liquidity effect. But they are in both directions. As I tried to show in this post it nets out to roughly 20 points.

    Think about it. If QE really had such a massive (say 100 point) liquidity effect, then how the heck did bond yields RISE when QE3 was announced? Stocks confirm it was a pleasant surprise to markets. And even worse, why haven’t rates fallen back to 1.7% when we returned to the original Sept 2012 QE3 policy of a data driven program. It makes no sense.

    brendan, You said;

    “Think about it- were you surprised equities trended up after QE2? Or after the Evans rule? Or after Abe? You were not;”

    Yes, I was. If I wasn’t, I’d be rich. (Instead my commenters are getting rich.)

    John, Thanks for that data.

  20. Gravatar of ssumner ssumner
    19. September 2013 at 08:51

    Thanks Brian.

  21. Gravatar of njk njk
    19. September 2013 at 09:22

    This is great:

    Credit Agricole’s David Keeble: “Fed credibility and its communication strategy are in tatters.”

    Scott Sumner: “The Fed did what it said it would do, make its decisions based on the data. Good for them, and good for policy credibility.”

  22. Gravatar of Suvy Suvy
    19. September 2013 at 09:48

    To explain the flat yield curves across the ZLB as a sign of very tight monetary policy, it’s really simple. What asset classes perform well in an environment of deflation and negative growth? Cash and government bonds…

  23. Gravatar of Matt McOsker Matt McOsker
    19. September 2013 at 09:56

    Scott, in my above quote I am speaking of bond prices not stock prices. If bond prices go down then yields go up period – there are no stock shares involved that are exchanged. The Fed buying I would think is supporting and driving up bond prices.

  24. Gravatar of Morgan Warstler Morgan Warstler
    19. September 2013 at 10:04

    Scott,

    I just want to know if I’m think correctly about CBO announcements, is that what they’d see:

    Pay increase for government announced while NGDPLT has been running at trend… without Fed action, the new predicted NGDPLT is higher (via inflation), so the Fed would need to raise rates to stay on trend.

    Raising rates means the cost of rolling over government debt is higher, so we might see the actual deficit increase as well.

    Am I missing something? Or is this a decent playout of events assuming a big pay bump.

    Is this

  25. Gravatar of dlr dlr
    19. September 2013 at 11:06

    In fairness you tried to overcome that by cherry picking some periods where the liquidity effect may have been driving things, and correlating with stocks. The problem is that it was driving things in both directions in your sample period. I don’t doubt that 100, 200, or even 300 basis points of bond yields moves can be attributed to the liquidity effect. But they are in both directions. As I tried to show in this post it nets out to roughly 20 points.

    Scott! You pointed to a (totally random) period where real long rates and stocks had a pretty clear negative short term correlation (cherry picking!), especially around Fed announcements/rumors, but argued that these netted out over the year, presumably because both rates and stocks rose substantially for the period. Yet from Jan 1 2011 to September 2012, you would regularly point to what were typically positively correlated co-movements in rates and stocks around Fed news/rumors as evidence that the Fisher Effect was validating the MM view of the world. Taken as a whole, though, those positive Fisher Effect correlations also “netted out” such that the 10YR went from 3.5% to 1.7% while the S&P went from 1260 to 1470. I can’t help but think you went from event-studier to event-netter somewhat conveniently.

    Think about it. If QE really had such a massive (say 100 point) liquidity effect, then how the heck did bond yields RISE when QE3 was announced? Stocks confirm it was a pleasant surprise to markets. And even worse, why haven’t rates fallen back to 1.7% when we returned to the original Sept 2012 QE3 policy of a data driven program. It makes no sense.

    Exactly. It makes no sense. And I have not seen a paper, blogger or anyone else who has been able to fit the real interest rate pattern of the last view years into a simple model of the world, nor have I found any of the non-simple ex post fitted models interesting. But I think maximum credibility (I say this carefully, as I think you have plenty of credibility) requires you be more straightforward about the mysteries inherent in these varying patterns between the real rate curve and stocks instead of overfitting them into your 1-2-3-4 model of the world. This is coming from someone highly sympathetic to your view of what principles probably most matter in thinking about how these prices should relate over time.

  26. Gravatar of WeaK Signals | Historinhas WeaK Signals | Historinhas
    19. September 2013 at 17:14

    […] were never “driven down” and actually mostly went up (see this post by Scott Sumner on “Interest rates and the face/vase problem“)). Stocks which had already been trending up continued to do so. But note that the “taper […]

  27. Gravatar of lxdr1f7 lxdr1f7
    19. September 2013 at 18:24

    What is the liquidity effect?

    Fed policy is half the equation, expectations are the other half. So therefore any policy stance can result in alot of variability because the policy stance is not the only factor determining yields.

    Brendan

    “Yes, EMH violations. But the EMH regularly is violated by reality in these ways, across time, and across asset classes.”

    Also there may be trading on inside info going on so the EMF might not be violated.

  28. Gravatar of ssumner ssumner
    20. September 2013 at 18:39

    Matt, First of all, you didn’t even mention the Fed in the comment I was responding to. But even if it’s the Fed doing the bond buying, it can easily reduce bond prices. Suppose Fed bond buying created inflation fears. That would reduce bond prices.

    Morgan, Yes, in theory that would happen. I’m not sure how powerful the effect would be, unless it was a big government pay raise.

    dlr, I don’t see how you can accuse me of changing position. I agree that the evidence suggests that tapering rumors raised bond yields. So I was using the exact same procedure as when I looked at earlier periods to support the MM position. You object to my claim that only 20 of the 100 bp rise in yields is tighter money, and yet I never did any breakdown like that in 2011. So I see no contradiction at all.

    The micro (daily) data suggests tapering has raised rates. But no matter how many examples you find of daily movements where yields rise and stocks drop, those correlations CANNOT POSSIBLE BE REPRESENTATIVE. And that’s because over the past year stocks have soared as yields have soared. So it seems that some other mysterious factor is affecting markets in a much more powerful way that tapering.

    As far as the “mysteries,” let me just say that among economists I’m among the most likely to say that we don’t know what interest rates mean. “Never reason from an interest rate change.” All I claim is that the really really big swings in rates, from 20% to zero percent, usually are dominated by the income and inflation effects, not the liquidity effect. And nothing in this 100 basis point (one percentage point) change has made me change my mind.

    However I do plead guilty in one respect. I think in my eagerness to show “contrarian examples” where monetary policy made rates move in an unexpected direction, I left the impression that MMs believe that rates always move in the contrary direction. Well clearly that’s wrong, so I should have made that clearer.

Leave a Reply