The Empire’s last stand: Real interest rates

You may recall that a week ago I made a sort of “Emperor has no clothes” accusation against the economics profession.  I claimed that economists are always talking about “easy money” and “tight money” without have any coherent definition, indeed not even knowing that they have no coherent definition.  In this post Bob Murphy challenges my dismissal of real interest rates as the proper indicator.  He makes some good arguments, which I will address as well as I can, but in the end I am still left with 4 reasons for dismissing the view that real interest rates provide a useful indicator of the stance of monetary policy.  Furthermore, I think that any one of these four arguments would be sufficient to prove my point:

1.  The theoretical argument: 

Contrary to widespread impression, there is no theoretical presumption in a IS-LM model that easy money lowers real interest rates.  As Robert King pointed out in the JEP (pp. 77-78, 1993):

If changes in the money stock are persistent, then they lead to persistent changes in aggregate demand. With a rational expectations investment function of the neoclassical form, persistent changes in demand for final output lead to quantitatively major shifts in the investment demand schedule at given real interest rates. These effects are generally sufficiently important that real interest rates actually rise with a monetary expansion rather than fall: the IS curve effect outweighs the direct LM curve effect. Further, if persistent changes in the money stock are only gradually translated into price or wage increases, then there are very large additional expected inflation effects on nominal interest rates.

In December 2007 I noticed that a contractionary surprise by the Fed actually reduced 3-month T-bill yields for exactly the reasons mentioned King.  I don’t have any data on 3-month real rates, but it wouldn’t surprise me if they fell as well.  Indeed the reason most economists assume that Fed easing lowers real rates is simply that they assume easy money lowers nominal rates, and it stands to reason that easy money couldn’t lower inflation expectations.  Ergo, real rates must have fallen.  But once you realize that easy money doesn’t always reduce nominal rates then it becomes clear that there is no reason why it should necessarily reduce real rates either.

The implication of King’s observation is that even if my next three arguments are all wrong, if real rates did not in fact rise late last year, it would still not make real interest rates a good indicator of monetary policy.

2.  I showed you mine, now you show me yours:

When I mention to other economists that real rates rose dramatically between July and November 2008, the reaction is usually: “Oh my gosh, I never noticed that.”  Then they scramble around for some way to discredit my argument.  Perhaps the real interest rate in the TIPS market is not the “correct” real interest, despite the fact that it is the only direct, objective, measure of real interest rates that we have.  If you were looking for a risk-free real interest rate, what better place is there to look than the rate on indexed, risk-free, government bonds?  Aren’t these bonds virtually the textbook definition of what economists mean by “the” real interest rate?

Now if you look hard enough it is possible to find flaws in any measure of real interest rates, but let’s go back and deconstruct what is really going on here.  Why don’t the economists I talk to react to my statement about TIPS yields by saying: “Oh, am I glad to hear that!  I was having trouble understanding how AD could have been falling so fast last year during a period when money was easy.  Now that you’ve explained money was tight, the whole picture makes sense; the falling commodity, real estate and equity prices, falling industrial production, as well as the soaring value of the dollar against the euro.”

But guess what, not one economist has reacted that way.  Instead they all look for ways to dismiss the TIPS yields.  This tells me two things.  First, that they have already made up their minds, they have already decided that money was easy, and now they are simply looking for some way of validating their prior beliefs.  And second, the fact that the TIPS yields were news to most economists that I talk to means that they really don’t pay any attention to real rates at all.  Most probably just look at nominal rates, and assume that that’s all they need to know.  Or perhaps nominal rates adjusted by backward-looking inflation, which is almost as bad.  Only when it is pointed out that nominal rates are extremely unreliable do they claim that “economists use real interest rates as the indicator of monetary ease and tightness.”  OK, then where is your real interest rate indicator?  I showed you my TIPS yields, what are you using?  In fact, if economists really were interested in real rates they would at least know the stylized facts of what happened to the TIPS yields last year, even if only to reject them and look elsewhere for real rates.  But many don’t even pay enough attention to real rates to know that the TIPS yields soared, and yet they’re quite confident that they “know easy money when they see it.”

Bob Murphy does have some good arguments.  He points out that nominal rates on Treasury bonds fell significantly in the last half of 2008, and that it seems improbable that inflation expectations were falling much more sharply.  It seems improbable to me as well, but it seems to have occurred.

3.  Inflation expectations did plummet:

In this section I will explain why I think inflation expectations did plummet last July through November, despite the many attempts to cast doubt on this sea change in expectations.  For those of you having trouble visualizing how expectations could change so fast, recall that in early July oil was $147 a barrel, and people were worried about inflation.  By November commodity prices and economic activity were in free fall.  Real estate prices were also plunging.  Wages cuts were on the horizon.  T-bill yields were near zero, and the specter of Japan was increasingly being mentioned.  That’s a pretty big change in 4 months.

Yes, there are reasons to question the relevance of the TIPS spreads.  Bob mentions the fact that TIPS yields should never be higher than non-indexed Treasury yields, because the bonds can only be indexed upward at maturity, not downward (with deflation.)  Nevertheless, if you use existing “off the run” TIPS, that have already had some accrued gains due to past inflation, the value may be expected to fall from current levels as future deflation undoes some of the positive indexing from past inflation.

Bob also raised the issue of liquidity.  Bob (and others) alleged that there was a mad rush for liquidity last fall, and that it drove T-bond prices much higher than they otherwise would have been.  Because TIPS are less liquid, their prices fell relative to T-bonds, and thus the yields rose.  And (so the story goes) this has nothing to do with falling inflation expectations.  Sorry, I just don’t buy this explanation.  It certainly contains a grain of truth, but nowhere near enough grains to explain the bond market movements in the second graph in Bob’s post.

And BTW, here is what Bob wrote to explain his state of mind on seeing what actually happened to real interest rates last year:

Here’s the chart that Scott refers to, and I must confess at first it shocked me:

Shocked?!?  I’m shocked that he’s shocked, as I indicated in the previous section.  (But in his defense, Bob does pay attention to the monetary base, so he is not one of those economists that I accused of just relying on nominal rates.)  Here’s my take on the graph, and I freely admit to being no bond market expert, so please take this as just a hypothesis.  The rise is real rates in the 5-year TIPS market was exaggerated by two factors:

1.  There’s a small lag in indexing to the CPI.  Thus when commodity prices move dramatically a near-term CPI adjustment in the TIPS is predictable, but doesn’t truly represent forward-looking “expected inflation.”  In the last half of 2008 this biased upward the rise in real rates.  Fortunately, on a 5-year bond this bias should be no more than a few dozen basis points.  And rates rose more than 350 basis points.

2.  The liquidity issue mentioned by Bob may have also biased upward the rise in real rates.  But here’s why I don’t think it was a big factor.  Bob focuses on the difference between TIPS and non-indexed T-bonds.  And also the fact that real yields required by investors in safe government bonds soared during a period of turmoil:

I have also included the yield on AAA bonds to show that they spiked too, though not as much, as TIPS yields did, precisely at the time that the nominal Treasury yields collapsed. Again, I’m not saying that I know exactly what was going on for these critical months, but what I am saying is that the above chart is not what you’d expect from a straightforward collapse in the expectations of real GDP growth and CPI growth.

I agree the pattern looks kind of odd.  I admit the fall in inflation expectations and the rise in real rates seem a bit steep.  What troubles me the most is that for a brief period we had negative inflation expectations over 5 years.  I find that implausible (although let’s not lose sight of the fact that Japan experienced some persistent mild deflation, we were in the midst of one year’s worth of mild deflation, and also that things seemed pretty scary around that time.)

But here’s what I think is more interesting.  The TIPS yields (which after all are real interest rates) soared dramatically relative to AAA bond yields.  It looks to me like the AAA bond yields increased about 1%, whereas the TIPS yields rose about 3.5%.  This implies a fall of inflation expectations of 2.5% over just 4 months.  That’s pretty impressive.  Now you might be thinking that AAA and TIPS are not perfect substitutes.  True, but to the extent they differ it understates the fall in inflation expectations.  Why?  Because the AAAs were increasingly seen as being risky during that 4 month period.  Recall that this was when the bond ratings were found to be unreliable, even for AAA bonds.  So even if there was no change in inflation expectations, you would have expected AAA yields to have risen significantly relative to TIPS yields.  This implies that the actual decrease in inflation expectations was considerably larger than 2.5%.

What about the earlier liquidity argument?  I don’t see how that could apply to the TIPS/AAA spreads.  If anything I would consider AAA bonds to be less liquid than TIPS.  Yes, TIPS may seem illiquid when compared to regular Treasury bonds, but compared to anything else in the universe they are pretty liquid assets.  Their market now runs into the $100s of billions.  Unlike AAA bonds, all TIPS have the same default risk (virtually zero.)  The bid-asked spread in the TIPS market is very small.  And yet unless AAA bonds were significantly more liquid than TIPS, it seems to me that it is pretty hard to deny that inflation expectations fell by significantly more than 2.5% in late 2008.

So I end up remaining convinced that our best estimates show sharply rising real interest rates during the key July to November 2008 period, albeit the increase in real rates may have been less than implied by the TIPS markets.   And even if some bond expert comes in here and blows this argument out of the water, I still think the TIPS yields are the “correct” real interest rate if we are trying to estimate the stance of monetary policy.  The next section will explain why.

4.  Two pragmatic arguments:

Even if inflation expectations did not fall sharply in late 2008, and the real interest rate on T-bonds did not rise sharply, I would continue to argue for the relevance of TIPS yields on two pragmatic grounds.  (Here I’m adopting my normal “‘wildly throw enough darts at the target and hope at least one hits” approach):

a.  Let’s think about what we look for in the stance of a monetary policy indicator.  I picture one of those robots from a 50s sci-fi movie.  When policy is going off course you want the indicator to bark out “Warning, warning, danger Will Robinson, err, I mean danger Ben Bernanke.”  Red lights should be flashing in all of the halls in the Fed DC headquarters.  Now think about this.  If TIPS, which are perfectly good bonds, are seeing their yields soar through the roof in September 2008 because the financial system is imploding and there is such a mad scramble for liquidity that T-bonds are the only thing anyone is willing to hold, umm, perhaps might it be time to consider cutting the Fed funds rate from say 2%, to 1.75%?  I guess not, as they didn’t take that radical step.  Yes, I suppose that would have been reckless inflationism.

b.  Where is it carved in stone that T-bonds are the right asset for our model?  Yes, they are risk free , but so are TIPS.  Couldn’t you argue that if there is a wild scramble for T-bonds then they are the outliers, those are the yields that are unreliable, and TIPS are more like the broader class of “normal bonds?”   I often call myself a methodological pragmatist.  There is no theoretically pure answer to economic questions, it all boils down to what works best.  So consider the following argument.  A banker is considering making a loan, or considering the purchase of a risky corporate bond.  What sort of rate of return is required to make the banker take action?  What is the opportunity cost of funds than the banker is contemplating using in this investment?  Is it the T-bond yield of 2%, minus some unknown inflation risk, or the TIPS yield of 4%, which is guaranteed in real terms with no default risk?  The answer seems obvious to me.  No banker in her right mind would lend out money unless the real rate of return was expected to be considerably higher than 4%.  After all, regardless of how one feels about the liquidity argument discussed earlier, one can’t dodge the fact that anyone could walk into the TIPS market last November and buy a risk free asset with a very high real rate of return.  So how does that not become a minimum estimate of the real opportunity cost of funds?  Especially for bankers contemplating using the funds in very high-risk investments in a rapidly deteriorating economy.  I don’t know about you, but those sorts of risk-free real rates don’t seem appropriate in a severe deflationary recession, regardless of the liquidity characteristics of indexed bonds.

Of course as the economy began to decline sharply, real rates came down rapidly.  It should be mentioned however, that some graphs exaggerate the speed of the decline by shifting to a “on the run” bond around December 1st, 2008.  Mankiw discussed this problem in his blog around that time.

One other brief comment.  Bob takes me to task for harping on market expectations as being the gold standard of forecasts, but quietly ignoring the fact that the gold market itself is screaming high inflation.  I view the gold market as a world market affected by real factors like rising wealth in India and China (where gold is popular) and not being as specifically focused on US inflation expectations as the TIPS markets.  But yes, I’ve also been surprised by the strength of gold.  If we have high inflation over the next couple years (beyond just an oil blip) it would undercut my model.  My last stand will be core inflation, if that shoots up I’ll just quietly slip away.

PS.    Bob Murphy seems to think I’ve gone over to the dark side, so I used a vague allusion to Star Wars in the title.  I guess that makes him Darth Vader in my eyes.  However, remember that in those bland later Star Wars films we found out that Darth was well-intentioned but had been through some “traumatic events.”  Bob’s seen the evil way that governments can debase currencies and thus is understandably worried that we’re being set up again.

PPS.  If Bob’s Darth Vader then who am I?  Luke Skywalker?  I’d like to think so.  I think of my Joan Robinson quotation as being like the missile that Luke lobbed into the center of the Death Star.  They never knew what hit them.

Or perhaps Yoda:    When 5% NGDP target path you reach, evil recession you will face no longer.



45 Responses to “The Empire’s last stand: Real interest rates”

  1. Gravatar of David Stinson David Stinson
    10. November 2009 at 19:09

    I think when investors bought regular treasuries last fall, they weren’t motivated by the yield but rather the lack of downside nominal risk. They were looking for somewhere to park money until the chaos subsided somewhat. In those conditions, capital protection is enough. A key part of capital protection is the ability to exit with minimal loss, which is where the liquidity comes in, particularly if many investors were exiting at once.

    On the gold market vs. bond market issue, perhaps both could be right. Based on the TIPS spread, the bond market is forecasting lowish inflation and a weakish recovery. Many commentators have inferred from the gold price rise that the gold market is forecasting high inflation. I think the recognition of gold’s potential role as a monetary asset increases during times of economic stress. Thus, it seems logical to view the recent US$ gold price essentially as a currency exchange rate. A rising gold price implies a net excess supply of US$ relative to gold. Now gold has risen a great deal not just since last fall but over the last decade when inflation in the US was notionally under control. During that period, however, we have had two asset price bubbles. If one holds the view that excess money balances can be spent on existing assets (as well as on components of GDP), then a narrow TIPS spread could be consistent with rising gold prices and a rising stock market.

    On the question of whether real interest rates can be used an indicator of monetary tightness or looseness, as someone with fledgling Austrian leanings, I might have thought that the proper indicator from an Austrian perspective was not the level of real rates per se but their relationship to the natural rate.

  2. Gravatar of ssumner ssumner
    10. November 2009 at 19:59

    David, I agree about the T-bonds. If people were just looking for safety, the T-bond yields may have been misleading. Other rates were more indicative of the economic situation, and they were rising last fall. That’s why I focus on TIPS yields

    I agree about the natural interest rate, but it is not observable. We need something in the real world that we can look to in order to judge whether money is easy or tight. I think it is expected NGDP growth.

    Gold is a mystery to me.

  3. Gravatar of Bob Murphy Bob Murphy
    10. November 2009 at 20:14

    Scott I appreciate the reply. Presumably I will return the volley on my own blog in a few days, preferably when I have a work deadline that I want to ignore. Just a few quick reactions:

    * It is crucial for us to nail down the calculation of the “constant maturity” TIPS number. I will make some calls (maybe tomorrow) and try to pin that down. I still maintain that if money were neutral, TIPS yields would be completely unaffected by monetary policy.

    * Of course money isn’t neutral, that’s the point of this blog. Yet I claim that your theory requires TIPS yields to collapse once the Fed “tightened.” I.e. you are saying the economy was chugging along nicely, in fact we had $147 oil because Asia was booming, and then Bernanke turns the screws and causes real returns to go up by a factor of 8 (or whatever) in two months? That makes no sense at all.

    * If you think Luke Skywalker lines up with the guy defending the central bank, whereas Darth Vader is the guy who opposes a cartel of banks from getting injections of trillions of dollars in order to sop up US federal government debt, then I think you need to re-watch the trilogy.

  4. Gravatar of Bob Murphy Bob Murphy
    10. November 2009 at 20:27

    Oh another two clarifications:

    * I am not saying that real interest rates are the right indicator of Fed policy. As David Stinson suggested, I think the real test is whether the Fed is making markets move away from the direction they’d be if there were no Fed. Obviously that is a difficult thing to ponder. Scott didn’t misquote me or anything, but someone reading his post would probably come away thinking that I was saying “real rates are the right way to gauge monetary policy,” when if anything that’s what Scott saying. My point on the TIPS stuff was that Scott was wrong to point to the skyrocketing TIPS yield and say, “See? Clearly Bernanke tightened.”

    * To underscore why we need to figure out how those off-the-run bonds are included etc.: Scott seems to be saying that a collapse in inflation expectations could partially explain the rise in the TIPS yield as indicated in the chart. I am thinking that no, that’s not possible. As Scott himself says, the TIPS yield in theory is supposed to show what the real interest rate is, so inflation expectations shouldn’t affect it at all (except indirectly, if inflation expectations end up influencing the equilibrium real interest rate because money’s not neutral). I am not convinced that the off-the-run stuff should matter; I think if anything, older bonds would experience a one-shot change in their market values, so that their yield-to-maturities (I think that’s the term I want) would be identical to freshly issued TIPS. And clearly inflation expectations should affect the yield on a freshly issued TIPS.

  5. Gravatar of Bob Murphy Bob Murphy
    10. November 2009 at 20:30

    Sorry last sentence should be: “And clearly inflation expectations should NOT affect the yield on a freshly issued TIPS.”

    The standard way you derive inflation expectations from TIPS yields is to look at the spread between the nominal Treasury and the TIPS of the same maturity. So holding all else equal, if inflation expectations collapsed then the nominal yield should drop while the TIPS yield stays the same.

    Then if you augment that by saying that inflation expectations collapse at the same time everyone thinks it might be another depression, then you’d see TIPS yields fall and nominal yields fall even more. You would not see the TIPS yield skyrocket; that doesn’t make any sense.

  6. Gravatar of StatsGuy StatsGuy
    10. November 2009 at 21:13


    “I think of my Joan Robinson quotation as being like the missile that Luke lobbed into the center of the Death Star. They never knew what hit them.”

    First, it was a proton torpedo.

    Second, he didn’t lob it into the center of the death star. He fired it into a thermal exhaust port, where it initiated a chain reaction that destroyed the Death Star.

    Third, the Imperials _absolutely_ knew what hit them. If you recall:

    Moradmin Bast: “We’ve analyzed their attack, sir, and there is a danger. Should I have your ship standing by?”

    Governor Wilhuff Tarkin: “Evacuate in our moment of triumph? I think you overestimate their chances.”

    How many overconfident faces can we put in Governor Tarkin’s uniform? Hmmm…


    Re Bob’s comments… “He points out that nominal rates on Treasury bonds fell significantly in the last half of 2008, and that it seems improbable that inflation expectations were falling much more sharply.”

    Ditto on ssumner’s responses to the liquidity argument.

    One could make an argument about polarization of market participations (which is the argument I would make about gold) – some people expected instant hyperinflation what with Tarp and stimulus and (most feared of all) Obama winning an election – you can search various blog posts around November 08 and see all sorts of hyperventilating hyperinflationist warnings that never materialized. So we could have had a bi-furcating set of market participants, with one set of paranoid folks rushing into TIPS (and gold) to seek inflation protection even as another set was fleeing from assets (like stocks) into “safety” of Tbills.

    But, why then no arbitrage? And why would the latter set of “Safety” seeking investors avoid TIPS? Makes no sense.

  7. Gravatar of rob rob
    10. November 2009 at 21:46

    StatsGuy, “why no arbitrage?” Exactly. Sure would like to know how a Master of the Universe type arbitrage trader would explain things. I’m tempted to arbitrage it myself except there must be a good reason why others aren’t, as obvious as the inefficiency seems on the surface. Of course this gets me back to one of my anti-EMH theories, which is that most market participants buy into a weak-EMH to some degree, creating a conundrum. It’s the “seems to good to be true” problem. Say a glaring inefficiency seems to exist. No one will trust it because it is lying too much in plain sight. It doesn’t fit the mental models. The market must be at least somewhat right. And if the inefficiency was going to correct quickly, it already would have, so one must assume that it could last a long time, making for a scary arbitrage trade. Sort of like a bubble that you really, really think is a bubble, like Nasdaq in the late 90’s — but who has the balls to short it? The market can stay “wrong” longer than you can stay solvent, is I think the expression. But maybe it is, if I understand what Scott is saying –and I often don’t– a case of worldwide inflationary pressures for commodities relative to the dollar vs. domestic expected inflation. For example, and I hope this isn’t a bad one: it wouldn’t surprise us, would it, if commodity inflation takes off relative to the Yen but without the expectation of inflation in Japan’s domestic economy? Does that make sense to anyone?

  8. Gravatar of Jon Jon
    10. November 2009 at 23:25

    As I showed in our last exchange, PPI dipped severely negative during the summer of ’08. By this account CPI, NGDP, TIPS, etc were all lagging indicators.

    TIPS, TIPS, TIPS… *I* believe policy was tight but I don’t believe that your TIPS data is meaningful. TIPSs are not the ‘real-rate’–they sell at a discount relative to the expected value of a BLS report that has nothing to do with ‘inflation’ in the operative sense of macro economic models other than containing the word ‘inflation’ in the title.

    Lets pose a thought experiment. Suppose banks only funded home mortgages. They do nothing else. Now someone hands you some data on the nominal-rate charged in the loan market and the BLS CPI and asks what the real-rate was. This isn’t answerable with the data provided–surely the rate of home-price inflation matters more.

    The problem here is that if start thinking of inflation as neutral (effecting all goods equally), then we’re already assuming a regime were the real-rate is neither tight nor loose, and is rather unchanged. This pulls the cart before the horse when the expected regime and subject under discussion IS a shift in the real-rate.

    It is precisely when the real-rate varies from the natural-rate that you cannot use the CPI/TIPs to deduce the real-rate.

  9. Gravatar of malavel malavel
    11. November 2009 at 02:33


    I have some amazingly good news for you. Yesterday I championed your NGDP futures targeting scheme for the Swedish Pirate Party. Please stop laughing now! We actually have two pirates in the European parliament. So it wont take long before the ECB is in our hands.

    May the force be with you!

  10. Gravatar of Dioktos Dioktos
    11. November 2009 at 04:49

    “This implies that the actual increase in inflation expectations was considerably larger than 2.5%.”

    … Do you mean “decrease”?

  11. Gravatar of Scott Sumner Scott Sumner
    11. November 2009 at 08:22

    Bob, Bernanke doesn’t have to turn any screws to tighten policy. because the of the Fed’s flawed interest rate targeting policy, if the fed merely keeps the rate fixed at 2% then money can become much tighter if the Wicksellian natural rate falls. And it did fall in 2008.

    In my view the proper way of thinking about money being loose or tight is in terms of whether there is enough money to hit the policymakers nominal target (inflation of NGDP.) By that definition money became very tight in the second half of 2008. Now some people say “no, it’s interest rates” Then I show that doesn’t work. Then they say “I mean real interest rates” OK, but that confirms my NGDP target approach–real interest rates also signaled tight money. Don’t get me wrong, I oppose using real rates as policy indicators. I am skeptical about real rates for some of the reasons you mentioned. All I’m saying is that if we are going to use real rates, well then money WAS tight late last year.

    Right wing inflation hawks at the Fed are keeping money too tight, and turning a blind eye to the suffering of 10.2% of the workforce. I am convinced Luke would be with me, fighting the good battle against the forces of reaction. Want proof? Don’t all Hollywood films contain hidden “liberal” messages? There must be a hidden allusion to NGDP targeting somewhere in the Star Wars trilogy. . . .

    Bob#2, Thanks for clarifying that. And as my previous response indicated, I also think real interest rates are not a reliable indicator. So we agree there. I hope other commenters can weigh in on the TIPS issue.

    Bob#3, I agree that the sharp rise in TIPS during a recession is a bit of a puzzle. Of cours elater they did fall. But my pointn is that it happened, so people should go around saying:

    1. real rates are the indicator of fed policy
    2. Fed policy was really easy in the fall of 2008.

    Where is the evidence for these two assertions? I now understand that you don’t agree with point one, but plenty of people do. Perhaps my reply was addressing a broader issue than the specific points raised in your post, but I don’t have a good explanation for some of the puzzles you raise. I just wanted to emphasize that those puzzles don’t save the establishment economists who keep talking about easy money on the basis of interest rates. They lack evidence.

    Statsguy, I don’t know whether I should be impressed or concerned that you know so much technical info about the Death Star. 🙂 Oddly, I kind of knew ‘missile’ was the wrong term, and expected someone would correct me.

    You raise some good issues. But I keep coming back to the emperor has no clothes metaphor. When economists blandly assert that Fed policy was easy last year, have they delved into the intricacies of these TIPS puzzles? I see no evidence.

    ron, I wish I had bought TIPS last fall, I had a hunch the yield was too high. I have no explanation.

    I agree that it is very possible that you could have commodity inflation in yen terms, with little or no domestic inflation within Japan.

    jon, I agree that the TIPS are flawed to the extent that the CPI was flawed. Elsewhere I argued that actual deflation was more rapid than the deflation measured by the CPI. This means actual real interest rates rose by more than TIPS real interest rates.

    And I also agree that it is for other reasons that we can conclude that money was tight.

    malavel, The Swedish Pirate Party? That’s the breakthrough I have been waiting for.

    Dioktos, Thanks, I fixed it.

  12. Gravatar of Doc Merlin Doc Merlin
    11. November 2009 at 09:12

    First of all wrt gold and TIPS. TIPS do not index inflation, they index CPI. There is a very big difference. If countries we import from peg to the dollar and the FED chooses to inflate the currency, inflation will be high but CPI will be small. Like much of macroeconomics, CPI only works as a proper index of inflation when currencies are pegged to commodities.

    Imports to the US by %: China 16.5%, Canada 15.7%, Mexico 10.1%, Japan 6.6%, Germany 4.6%

    Buying TIPS is betting China, Canada, Mex, Japan, as an aggregate are going to inflate their currencies less than the US is, and that technological advancement doesn’t lower prices enough to counteract this.

    Inflation In your currency – Inflation in world currencies (not including yours)-production efficiency growth=what TIPS protect against

    Buying gold is betting that the aggregate of wold currencies will inflate faster than your own currency can deflate, versus increases in gold production due to new technologies and mining sources.

    Inflation world currencies (not including yours) + inflation in your own currency – production efficiency growth (in gold)= what gold protects against

    I hope this helps explain the Gold/TIPS discrepancy.

  13. Gravatar of Ryan Vann Ryan Vann
    11. November 2009 at 11:38

    As David mentioned, there is a risk aversion element to TIPS. For me, using a aggregate demand model to explain the recession means that TIPS aren’t a very good monetary measure of monetary policy because they are almost entirely an asset protection policy, and don’t really correlate to lending levels.

    With that said, I still agree with the tight monetary policy story, but I rely on experience as a loan officer at my previous employer. Our rates were basically through the roof, and our lending standards were incredibly high during the 3,4th qs of 2008, and in the 1,2 qs of 2009. Many other financial firms were doing the same. I remember seeing pretty terrible business and consumer loan numbers last year industry wide.

  14. Gravatar of RebelEconomist RebelEconomist
    11. November 2009 at 14:58


    The AAA yield series is derived from twenty to thirty year bonds (read the description on FRED), and so is not really comparable to five year treasury yields. TIPS auctions are relatively infrequent, so the “constant maturity” series may well undergo some relatively large step changes as the bond taken to represent the five year maturity point changes, especially if the real yield curve is steep.

  15. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    11. November 2009 at 15:34

    ‘ Gold is a mystery to me.’

    I don’t know much about it either, but have read that supply is an issue as new production has been in decline for awhile. Sort of a ‘Peak Gold’ thing.

    BTW, just how big is the gold market?

  16. Gravatar of Doc Merlin Doc Merlin
    11. November 2009 at 16:17

    Patrick, yah the gold bullion market is pretty small, roughly on the order of a billion or two dollars a day. Its also nowhere near the all time peak in price however. Also, I get wary when people talk of “peak commodities” to me, that is bubble thinking.

  17. Gravatar of Ash Ash
    11. November 2009 at 16:29

    Hi Scott – my first comment here and I hope its useful!

    Regarding TIPS, the liquidity issue during Q4 2008 was very real. In fact, the Cleveland Fed which used to publish inflation expectations from TIPS after adjusting for liquidity stopped doing so on 31-Oct ( ).

    Another option which has its own flaws is to use the inflation zero coupon swap quotes which quote as USSWIT5 on Bloomberg ( ). A reasonable explanation of the difference between the two and the zero coupon swap contract itself can be found at .

    As you can see, the zero coupon swap went from a 5Y expected inflation of 3% to -0.5%. This would have to be compared to the 5Y vanilla swap rate to get an indication of the real rate. I don’t have the numbers in front of me but stripping out some crazy days in November, we probably saw this real rate measure move from 1/1.5% to around 3%. Again, its worth reiterating that none of these measures are perfect in such chaotic market conditions but it’s almost certain that real rates did go up by atleast 1.5% on a 5Y tenor during Q4. So you are right!

    Having said all that, I’m not sure much would have been achieved by keeping real rates from spiking up. The real problem in such a massive deleveraging is the dramatic increase in credit spreads and collateral requirements i.e. the credit terms. The increase in the “risk-free” real rate doesn’t help but it’s dwarfed by the deterioration in credit terms for almost every borrower in the market. And there’s nothing irrational in this deterioration in credit terms given increased volatility, macroeconomic risk, risk of asset price collapse etc.

  18. Gravatar of ssumner ssumner
    11. November 2009 at 17:36

    Doc Merlin, I have no idea how inflation could be high but not the CPI. The CPI is somewhat inaccurate, but it’s not that far off.

    For which country does gold predict inflation? The US? Japan? China? It can’t predict inflation in all of them, as gold prices differ between countries by just the exchange rate.

    Ryan, I agree with what you are saying about TIPS. I don’t think TIPS (or any other interest rates) are good ways of judging monetary policy.

    Rebeleconomist, Both good points. But I wager if someone finds me the right data, my conclusions will still hold. Notice how hard it is to find out what real interest rates are? And yet somehow economists are convinced the Fed had an easy money policy last fall. On what basis?

    Patrick, I have a new post on peak gold.

    Ash, Yes, I’d say it is one of the most useful comments I have ever received. I knew about the CPI futures, but wasn’t sure where to get the historical data. Check out the 2 year CPI futures, there was severe deflation expected in the fall of 2008, more than 2%.

    So now we have a completely independent source showing rising real interest rates in late 2008. Your 5 year numbers support both of my “hunches”:

    1. Real rates did rise
    2. The TIPS somewhat exaggerated the rise.

    But I also think the 2 year rate is highly relevant, as monetary policy is usually assumed to mostly impact shorter term rates. So for those who care about interest rates, it is clear that 2 year real rates rose very sharply.

    I had read the BOE article already, and was aware that CPI futures usually showed higher inflation expectations than TIPS spreads. I can’t recall whether they mentioned a reason. I assumed it was because those buying CPI futures would typically be people worried about inflation. So to get someone to “insure” against the risk, they had to pay a risk premium. Does that make sense?

    So to conclude, if economists are going to use interest rates as the measure of tight money, and if real interest rates are the “right” measure of interest rates, I think we can pretty definitely conclude that money was tight last fall.

    The only area where I slightly disagree is your final comment about what could have been achieved by keeping real rates from spiking. Of course I don’t favor targeting real rates, but I do favor having a much more expansionary policy when NGDP expectations are falling. Had they done so, then the “macroeconomic risk” and “asset price collapse” that you refer to would have been far less severe. So although I agree that modestly lower real rates wouldn’t have helped much had the macroeconomy and asset prices still gone down sharply, I would emphasize that easier money could have greatly reduced the size of that crash. Had it done so, the banking system would have been in much better shape. I also think real rates might have spiked less, but I see that as a side effect of a policy actually aimed at asset prices not interest rates.

    I have also followed the Cleveland Fed, but find their stuff perplexing. They just put out a graph showing 4% inflation expectations over the next year. That seems crazy–I believe TIPS spreads and CPI futures will be far more accurate over the next 12 months. Perhaps I misunderstood their graph—did anyone else see it?

  19. Gravatar of Jon Jon
    11. November 2009 at 18:35

    Scott: Inflation in “Final Goods” averaged 5% for most of the past decade. I don’t want to belabor that point; I don’t want to discuss gold. Just saying.

    I wrote up a nice long response for the other thread but I lost track of it. The short-version: I think you’re pigeonholing my arguments on PPI. I was attempting to argue not that the Fed should suppress PPI inflation; my argument was that the Fed should react to PPI deflation because PPI most directly attaches to the real-rate as it prevails in the short-dated loan market that the Fed uses to gear up changes in policy.

    And thus it is fluctuations in PPI that cause unintentional swings in the policy stance of the Fed.

  20. Gravatar of Bob Murphy Bob Murphy
    11. November 2009 at 18:48

    Just wanted to say this is quite honestly the most useful batch of blog comments I have ever encountered.

    And I waste a good two hours a day reading blogs.

  21. Gravatar of Doc Merlin Doc Merlin
    11. November 2009 at 22:49

    “Doc Merlin, I have no idea how inflation could be high but not the CPI. The CPI is somewhat inaccurate, but it’s not that far off.”

    I discuss my arguments in this in a newer thread. I will post the summary here, part of which is that you are using two different definitions for inflation. Gold price does not give predict expectations in prices for consumer goods, which CPI does give us. In the current floating monetary exchange system the CPI, in large part, gives us the effect of relative monetary expansion.

    Country A prints more money to lower their exchange rate in an attempt to make their consumer goods more competitive in country B. The CPI in county B measures this as deflation, but B’s money supply hasn’t changed.

    “For which country does gold predict inflation? The US? Japan? China? It can’t predict inflation in all of them, as gold prices differ between countries by just the exchange rate.”

    I think we have a difference in the definition of inflation, and are talking past each other. You are using inflation meaning “prices in a country increasing”, and I am using it meaning “money supply is expanding faster than some money substitute.” Gold price and other investment commodities don’t predict CPI increasing. They predict investor belief that currencies’ monetary base are expanding too fast, relative to the expansion of that investment commodity.

    Gold gives us the effect of aggregate monetary expansion. Ask any gold investor, why do they buy gold? They say its because they think the subjective value of their currency is falling faster than gold is. If they expected that a different currency would retain its value better than gold they would buy that currency instead.

    I am not a gold bug, I don’t think gold has “inherent value” whatever that means. Gold isn’t magical, gold isn’t somehow special, its just another currency (yes it is still used that way) that currently expands in quantity very slowly relative to other currencies.

  22. Gravatar of ssumner ssumner
    12. November 2009 at 06:33

    Jon, I agree that the PPI is often a more accurate indicator of monetary conditions than the CPI. As I said, in the Great Depression I preferred to use the PPI. But again, if we had monthly NGDP I’d like that even better.

    But I strongly disagree with your argument that we’ve had 5% inflation. The inflation rate over the past 10 years is closer to two percent for final goods. Where did you get your data?

    Thanks Bob, I have also learned a lot. In the past I’ve said that I view the blog as a sort of collaborative effort, which is why I try to answer all the comments (I know some people think that’s a waste of time.) But what starts in the comment section often ends up as posts. This post started in the comment section for a previous post.

    Of course my views are always evolving as more new information comes in. Obviously your post played a big role in this one, as I was forced to think about the graphs you provided.

    Doc merlin, I did address your point in the newest post. But let me add here that I do consider inflation to be a rise in prices, which is also pretty much the official definition used by economists. Of course I actually focus on NGDP, but if inflation was my target, then I would use stable prices as a benchmark for monetary policy. I’m not concerned by what is happening to the money supply, as long as prices are stable (or plus 2% if that’s the target.)

  23. Gravatar of Jon Jon
    12. November 2009 at 07:56

    Scott: I’m looking at the fred PPIFCG time series “Price Index of Finished Consumer Goods” over the period 2001-present.

  24. Gravatar of John Hall John Hall
    12. November 2009 at 10:44

    Scott (or Ash),
    Do we have any idea on how the swaps or CPI futures are hedged? For instance, if I were a market-makers in any of these markets, I could create a synthetic TIPs bond out of a nominal bond and an inflation swap or CPI futures. I could create a synthetic TIPs bond out of these instruments and then trade that instrument against an actual bond (the way people make basis trades in the CDS market). I would think that this kind of behavior would link these markets together so that a shock to one would propagate, but I don’t know anyone who makes markets in them so I really don’t know for sure. For instance, in bloomberg the two year breakeven inflation rate is USGGBE02 and it is closely correlated with USSWIT2. However, the actual spread between the two widens to a record 3.65% in October 2008 when its normally like 50bps. I would think that a market-maker would be happy to step in and try to arb that difference. It does normalize, but not until April when the recovering was beginning to take hold. I take this as evidence of illiquidity in these markets though it wouldn’t by itself explain why breakeven inflation fell so sharply in October 2008.

    At the time most explaining it as illiquidity because no inflation forecasts I saw (for two years or five years) were predicting those kinds of inflation rates. I still think it’s unfair to just quote 2 year inflation breakevens (which would make sense to correct) without noting that the 5 and 10 year ones sharply declined as well. Since 10 year inflation expectations did not change meaningfully, it’s much easier to conclude that liquidity was a factor that extended to the whole TIPs market and explains the move in breakevens.

    Economists in the financial industry were well aware of the shocks to real interest rates. I think it was well-followed at the time. The reason why I maintained the liquidity story in the comments to your last post was because when it happened that’s what most were saying then.
    Damodoran wrote about it here:
    Krugman here:

  25. Gravatar of david glasner david glasner
    12. November 2009 at 14:27

    Scott, Let me take issue with your identification of the TIPS yield with the real rate of interest. The TIPS yield may be an estimate of the real rate as long as expected inflation is positive (for purposes of this discussion let’s assume that the CPI is the relevant measure of the price level). However, if deflation is expected, the return to holding a TIPS bond is the nominal yield (whatever that happens to be) minus the expected reduction in the nominal value of the bond to reflect the expected reduction in the CPI. Now if I can hold a regular treasury bond without having its principal adjusted for deflation and can earn whatever the nominal yield the Treasury bond is now generating, for anyone to hold a TIPS bond, it must offer a nominal yield sufficient to compensate for the expected loss of nominal principal associated with the expected CPI adjustment in its nominal value. So the yield on the TIPS bond, when deflation is expected does not in any way correspond to the real rate of interest, it actually is more closely related to expected deflation which is return one foregoes by holding a TIPS bond instead of cash or a Treasury bond. The real rate of interest in the fall of 2008 was undoubtedly negative, so the whole discussion in this thread seems slightly askew to me.

    If you respond that I am confusing the real rate with the natural rate, I guess I wouldn’t disagree too violently, but then I would ask you why you think that there is any difference between the real rate and the natural rate when the expected rate of deflation is equal to or greater than the real rate.

  26. Gravatar of Ash Ash
    12. November 2009 at 16:51

    Scott – first a minor point, USSWIT5 is an OTC zero-coupon swap rate, not a futures contract traded on an exchange.

    I think the argument in the BOE paper as to why the swaps-implied inflation rate is higher is just a market flow argument. There is strong demand for receiving inflation in the swaps market which can’t be met in the TIPS market because of the inability to customize flows and thin liquidity.

    The new approach by the Cleveland Fed ( )
    has a few problems. For one, it also incorporates economists’ forecasts apart from market data. From a quick glance at the underlying paper ( ), it also seems that their model throws out a negative inflation risk premium at the short end. I suspect that these two factors are responsible for the unusually high inflation expectations thrown out by their model in the short end. However I’ve gathered this from a 20 minute read of a rather long-winded and technical paper so I could be wrong!

    On the liquidity issue, check out pg 20 of this paper by Campbell, Shiller and Viceira ( ) which quotes from another paper by two Pimco bond traders Hu and Worah. There is no doubt that the TIPS market in Q4 2008 was distorted.

    Also, I would not put much faith in the short-end implied inflation rates at that time, whether they be the swap rates or the TIPS. We would all love to have some reliable short-end inflation forecasts for macroeconomic purposes but the market in the short-end was distorted by even more technical factors related to the structured product market in inflation. Essentially, common structured products in inflation involved banks selling inflation floors with strikes of 0% or -1% to their clients which were assumed to be worth very little. When implied inflation in the short end started falling, banks being short these floors had to sell more inflation to dynamically hedge themselves driving the market even lower. If you have access to Euromoney, the RBS inflation head discussed this in the 2nd para of this article – . If you don’t have access, I’ll just quote it here:
    ” Inflation markets say little about where we think inflation is going and all about supply and demand for the product. Over the last few months, particularly in the US and Europe, at the front end of the curve you have the fallout of commodity prices and the carry impact. If we jump forward 12 months, where will the year-on-year prints come out as a consequence of the commodity highs and what impact will that have on the three-month, linked bonds with their carry adjustment? At the same time people are jumping out of Tips (Treasury inflation-protected securities). The banks’ policy might be to hold treasuries, but Tips have suffered because of perceived illiquidity. As implied breakevens push lower and tend towards zero and through zero, people are becoming longer inflation and needing to sell it. That was the reason, in September, that one-year inflation in the US dropped 500 basis points in two weeks. There are people who are short of, not just 0% floors, but -1% to -2% floors. Whoever considered -2% deflation? Suddenly I have a massive intrinsic in my option price because the one year is trading at minus four! What commodity prices are doing is relevant. But curves have steepened at the front more because of embedded option situations than commodity prices.”

    So conclusion being that 2y and below breakevens from that period are unreliable, 5y and 10y are much better.

    John – the 10y breakevens also tumbled in that period from 3% to around 1.2% . I agree that the dislocations in the TIPS did partially feed through to the swaps market as well. I’m only arguing that the longer tenor swaps are the best we’ve got and that the argument that real rates went up by a nontrivial amount in Q4 is pretty hard to refute.

  27. Gravatar of Scott Sumner Scott Sumner
    12. November 2009 at 19:09

    John Hall, I am still trying to figure all this out. And I see that Ash has another post which will be more informative than anything I can say.

    But what I can do better than some of these technical experts is supply some perspective. Let’s take the Damodaran paper, which seems to argue against my hypothesis about what was going on with real rates and inflation expectations. Maybe he is right, but consider the assumption he needs to make his argument. He argues that it can’t be deflation, because a deflationary monetary policy should have lowered both nominal and real interest rates. Strangely enough, I agree. I am surprised that both nominal and real rates didn’t fall last fall. But now let’s think about what that means. If he expects a deflationary monetary policy to reduce real rates, doesn’t that mean that real rates are a very poor indicator of monetary policy. Low real rates do not imply easy money.

    On the other hand if the conventional view that high rates meant tight money was correct, then this would mean that a deflationary monetary policy would not reduce real rates. And that would refute the assumption he uses in arguing against my interpretation. Either way I win. And that’s really the issue I care about here—the issue that motivated this whole discussion. Are real interest rates a reliable indicator of monetary policy? And every day I become more convinced they are not.

    David, I’m not sure I follow your argument.

    The reported yields on TIPS are real yields. The nominal yield is the real yield plus inflation. Suppose you have a TIPS yielding 3.5%, and there is 2% expected deflation. Then:

    Real yield = 3.5% = expected nominal yield of 1.5% minus expected inflation, which is negative 2%

    3.5% = 1.5% – (-2%)

    Where am I wrong?

    The natural rate plays no role here. It is unobservable, even ex post.

    Ash, Again very helpful. A few comments:

    I just couldn’t wade through the Cleveland Fed paper. Common sense saves one a lot of time in economics. And when someone starts out saying they can estimate expected inflation, and they produce a figure of 4% over the next year, I tune out. Maybe that’s unfair, but I find that forecast so bizarre it makes me doubt their procedure. I’m not saying inflation can’t be 4% next year, but it certainly isn’t the market forecast, nor is it the consensus forecast of economists. It’s like saying the S&P500 could be 800 next year, or 1600 next year. Either number could be right, but neither are market expectations of next years S&P.

    I agree the TIPS markets was distorted in Q4 2008, but wasn’t the Treasury market much more distorted? Everyone says there was a rush for liquidity. Wouldn’t that distort the price of T-bonds (in an upward direction, i.e. reduce yields?) So weren’t T-bond yields also unreliable? If so, where are the reliable interest rates that we can use to determine whether money was easy or tight? (That’s a rhetorical question.)

    BTW, some say “well the TIPS market is smaller.” But the Non-Treasury bond market in total isn’t smaller. And all those bonds are less liquid than T-bonds. So aren’t T-bonds the outlier?

    Regarding the paragraph you quote. They say:

    “That was the reason, in September, that one-year inflation in the US dropped 500 basis points in two weeks. There are people who are short of, not just 0% floors, but -1% to -2% floors. Whoever considered -2% deflation?”

    I think they are probably partly right, but the way they present the argument is a bit misleading. Thay act like 500 basis points drops in one year inflation expectations are preposterous. In fact, the only thing I find odd is that it happened so quickly. But if they had said 2 months rather than 2 weeks there would have been nothing at all strange. Remember that in the 12 months after July 2008 the ACTUAL inflation rate fell roughly 700 basis points. Furthermore, the turnaround in both the commodity markets and the overall economy was so dramatic that inflation expectations were almost certainly falling fast. In the September 16th meeting the Fed was actually more concerned that inflation would be TOO HIGH, rather than too low. By mid-October the stock market crashed and anyone with two eyes could see we were heading into a deflationary recession.

    So while I agree that the markets were all distorted at that time, let’s not overlook the fact that it would have been quite rational for there to have been a very steep drop in inflation expectations.

    In the end I buy your conclusion, real rates almost certainly rose, but less than reported. But I think some readers might have gotten a misleading impression from that paragraph you quote.

    Thanks again for all the info.

  28. Gravatar of Scott Sumner Scott Sumner
    12. November 2009 at 19:19

    Jon, I thought you meant the CPI. Why didn’t you use the PPI for all goods? How does that differ from the PPI for final goods? The term “inflation” usually refers to all goods, not just a subset. PPI final consumer goods excludes both non-consumer goods, and also the retail markup part of consumer goods.

  29. Gravatar of Jon Jon
    12. November 2009 at 21:42

    The assumption that inflation refers to all goods implies simple cross validation test should be possible. If the common conception is correct, I should be able to select a subset of the prices at random and compute an accurate estimate. From the perspective of the CPI this is not so. Most ensembles of prices differ from the CPI. This occurs because the CPI weights are not randomly drawn.

    Regardless, I happened to look at the time series for consumer goods only by random draw. The time series for finished goods generally is essentially the same.

  30. Gravatar of davglasner davglasner
    13. November 2009 at 05:59

    Scott, You are not wrong, you just didn’t get my point (for which surely I am responsible). The yield on a TIPS bond is certainly a “real” yield. However, if expected inflation is positive, that real yield does in fact correspond to the conceptual ex ante real rate of interest (defined as the nominal rate that would prevail if expected inflation were zero). So when expected inflation is positive, you can infer the (ex ante) real rate of interest by looking at the yield on a TIPS bond. I am saying that when expected inflation is negative, you can’t infer the unobservable ex ante real rate by looking at the yield on the TIPS bond, because the yield on the TIPS bond must be sufficient to induce people to hold a TIPS bond rather than Treasury bonds or cash that earn real yields from deflation. The yield on cash in this circumstance exceeds the ex ante real rate which is exactly why the economy is tanking when you are in this situation. You and Bob Murpshy agree that it was anomalous for the yield on TIPS bonds to have spiked in fall 2008, because the real rate should not have been rising as the economy was tanking. And I am saying that it only seems anomalous because you are carelessly identifying the yield on the TIPS bond with the ex ante real rate. But when expected inflation is negative, the yield on the TIPS bond is disconnected from the ex ante real rate, it is governed instead by the ex ante yield on holding cash i.e., by expected deflation. So the fact that the yield on the TIPS bond spiked as the economy tanked was not all anomalous. That’s all I’m saying.

  31. Gravatar of ssumner ssumner
    13. November 2009 at 08:00

    Jon, I guess I have forgotten the context of the debate. Subsets of the CPI differ from the overall CPI. I agree, indeed that is what I said. But this has nothing to do with the statement that “inflation refers to all goods.” Inflation is defined as the price change in all goods. One may not like that definition, but it is what it is. If you have some other concept in mind, better to use another term than ‘inflation.’
    I use NGDP growth, because I find that concept more useful than inflation.

    David, I don’t agree, for two reasons:

    The yield on TIPS is definitely the ex ante real rate, as a buyer of TIPS is guaranteed exactly that real rate of return. It is both the ex ante and ex post real return on TIPS.

    The real yield on cash would only be equal to the real interest rate if nominal rates were zero. But nominal rates on 5 year bonds weren’t zero last year.

    (Most probably we are still talking past each other.)

  32. Gravatar of David Stinson David Stinson
    13. November 2009 at 08:17


    “Coxe says that there are a host of reasons for the recent outperformance of gold, the least of which is the crumbling dollar. As long as interest rates are at zero, the carry trade in the US dollar will continue to flush liquidity into the markets and into gold, and as that involves shorting the dollar the dollar will continue to slump.”

  33. Gravatar of david glasner david glasner
    13. November 2009 at 08:29

    Yes we are, the ex ante real rate is the yield from investing in real assets. The yield on holding TIPS when deflation is expected is the yield from holding cash. That’s a bid difference. I am using the case where nominal rates are zero as a limiting case, but the same general point applies even with a positive nominal yield on cash which simply reflects the Keynesian liquidity premium.

  34. Gravatar of Scott Sumner Scott Sumner
    13. November 2009 at 13:54

    David Stinson, Just because the dollar stays at zero, doesn’t mean the dollar will continue to depreciate. The dollar rose strongly between July and November last year, despite near zero interest rates on T-bills in September-November 2008. The fact is that no one can predict where the dollar will go. If they could, they’d be rich. I believe that investors expect the dollar to appreciate against the euro over the next year. Perhaps someone has forward market data.

    David Glasner, Consider the following data:

    TIPS: real yield = 5%

    T-bonds: nominal yield = 4%

    Cash: nominal yield = 0%

    Now assume that expected inflation is negative one percent. I claim that:

    TIPS have a an ex ante real rate of 5%
    T-bonds have an ex ante real rate of 5%
    Cash has an ex ante real rate of 1%

    I still don’t see how you can say the real interest rate in that economy is 1%.

    Every definition of real interest rates that I have ever seen has used bond yields, and the Fisher equation. That is all that I am doing. Can you provide me with a source that claims real interest rates during deflation are the real return on holding cash? In general, the claim I see is that during deflation the market real interest rates cannot fall below the real yield on cash

    BTW, I think this statement may have thrown me off. this is from your first response:

    “The real rate of interest in the fall of 2008 was undoubtedly negative, so the whole discussion in this thread seems slightly askew to me.”

    Did you mean to say positive? If cash has a zero yield, and deflation is expected, why wouldn’t the real return be positive? (Even if you defined the real interest rate as the rate of return on cash.)

    I also see one additional problem with your argument, it creates a puzzling discontinuity. You said at positive expected inflation the real rate is the TIPS yield, but at negative expected inflation rates it is the real return from holding cash. That means if inflation went from plus .001% to minus .001%, the real interest rate would plummet from the TIPS yield to roughly zero.

  35. Gravatar of John Hall John Hall
    13. November 2009 at 14:21

    I agree that interest rates (nominal or real) need not be indicative of loose or tight monetary policy. However, I still think that levels of and changes in some interest rates can convey information about the policy stance.

    My point has centered on your explanation for the spike in real interest rates. You argued in the past that it meant the markets were forecasting deflation. I disagreed and said it was a liquidity issue. If these markets were distorted b/c of liquidity, that means they conveyed less information about monetary policy. If anything, I agree with you on this (though for slightly different reasons). I just disagree that inflation expectations plummeted as severely, which is what made real interest rates a lousy indicator.

    Scott, I’m sure you’ve addressed this (and please direct me to the post if you have), but if you’re trying to have forward-looking monetary policy, then why should you care about the output gaps or having monetary policy return to a 5-6% nominal GDP growth path. It seems like what should matter is the forecasts for growth for the year ahead, not the forecasts for output gaps. In other words, why should monetary policy want enough growth to return to trend, rather than just growth.

    I explain two rules below, but the second (using paths) comes closer to your thinking about monetary policy (closer but not exact, as I’m aware). From a theoretical perspective, why would you say the second is better?

    I find it helps to view monetary policy as tight or easy based on year ahead inflation and GDP forecasts. Ideally market-based or from a set of sources, but I settle for the SPF survey from the Philly Fed since it has a long history. If a central bank were to follow a rule that sets Fed funds equal to core PCE plus 2% plus forecast CPI minus 2% plus forecast GDP minus 3% (or similarly 5% nominal GDP growth forecast), then you would say monetary policy was tight from Q11995-Q12001, easy from then to Q12006, tight from then to Q42007 then easy since then. If anything, the SPF forecasts are substantially below the consensus forecasts now and monetary policy would be too easy on that basis.

    To approximate a nominal GDP path rule (or at least just the real GDP side of it), I took the compound growth rate of real GDP over the past two years, subtracted 3%. I then took 1 plus that rate times 1 plus the real GDP forecast minus 3% from above to get a measure of the expected output gap in a year’s time. By this standard, monetary policy was easy from Q22001-Q42005, then tight through Q42007, easy through Q32008 and then tight since then (ie this rule demands negative interest rates currently).

    Though I wouldn’t want to admit it three months ago, but both of these policy rules seem to suggest that monetary policy was pretty good in the mid-90s, though it was easy in the early 90s.

  36. Gravatar of D. Watson D. Watson
    13. November 2009 at 14:22

    Scott, you said, “But guess what, not one economist has reacted that way.” Actually, I’ve felt that way I think 3 times over the last few months reading your posts. It took a smidgeon of time to read through enough of the arguments that it all came together, but that largely sums up exactly how I felt. I may be a newly minted economist, but I doubt I’m really the only one.

    A note about TIPS: it doesn’t matter if TIPS has a discount factor as long as that discount factor is not what is wildly fluctuating. If that discount is relatively stable, its movement is still indication of the trend of real rates. We deal with proxies and instrumental variables all the time and this is little different.

  37. Gravatar of 123 123
    13. November 2009 at 14:32

    Information provided by Ash is very intersting, but Scott has a point here asking are T-bond yields reliable. The same anti-bubble that has reduced the price of TIPS has also reduced the price of private sector assets.

  38. Gravatar of Jon Jon
    14. November 2009 at 08:47

    Subsets of the CPI differ from the overall CPI. I agree, indeed that is what I said. But this has nothing to do with the statement that “inflation refers to all goods.”

    I disagree very fundamentally. If I pick sufficiently large ensembles at random, I should arrive at the ‘same’ answer. It should not be necessary to carefully construct a weighted market-basket…

  39. Gravatar of Scott Sumner Scott Sumner
    14. November 2009 at 12:32

    John Hall, I do not want to target output gaps. I don’t even want to estimate output gaps (as I don’t trust the estimates.) I prefer a 5% NGDP growth target, level targeting.

    The approach you mention sounds like a forward-looking Taylor Rule. That would be better than a backward-looking Taylor Rule. But I still prefer a simple NGDP growth target, where the market sets interest rates, not the Fed. In a financial crisis the Wicksellian equilibrium rate might depart quite far from the prediction of your policy rule.

    D. Watson, I was referring to those I spoke to in person, but I am glad to hear that I have at least some persuasive power.

    123, Thanks.

    Jon, Your argument would make sense if we merely wanted to estimate the amount of inflation caused by monetary factors. But if we want the total inflation rate, including the impact of supply shocks, then weighting does matter. If housing had the same weight as toasters, then the CPI would look much different when relative housing prices were changing.

  40. Gravatar of david glasner david glasner
    15. November 2009 at 19:09

    Scott, You are thinking of the real rate as the nominal return on any asset adjusted for the change in the price level. Given expected inflation, the ex ante yield on Treasury bonds and TIPS bonds should be equal. If expected inflation is positive, then the yield on Treasury bonds exceeds the yield on TIPS bonds and if expected inflation is negative the yield on TIPS bonds exceeds that on Treasury bonds. The confusion arises because I asserted that in October 2008, the real return on capital was negative, which it certainly was. The nominal return on Treasury bonds was positive, so you rightly ask me how I can reconcile a posited negative real rate and expected deflation with a positive nominal rate on Treasury bonds and a zero nominal yield on cash. The answer is that the positive yield on Treasury bonds and the zero yield on currency reflect a liquidity premium and the existence of that liquidity premium is a disequilibrium phenomenon that drives down the prices of real assets until the expected return to holding real assets becomes positive. Of course falling asset prices can also reinforce expectations of falling asset prices so the downward spiral can go on for a while before an equilibrium is found. On reflection, I might have been better advised to call the real rate that I was referring to a natural rate, but I wanted to underscore that I was referring to the expected yield on holding real assets. In that disequilibrium environment, the usual Fisher equation doesn’t hold because asset prices must fall for the real rate to rise enough to make the nominal rate consistent with expected deflation.

    In the case I have in mind, the nominal rate on Treasury bonds is 1 percent, expected inflation is negative 4 percent and the yield on TIPS bonds is 5 percent. You can call the return on TIPS bonds a real return, which it is, but it reflects expected deflation and the liquidity premium on holding Treasury bonds and the downward adjustment in the principal of the TIPS bond to eliminate the appreciation in the value of money associated with deflation. There is nothing pathological about your case with a 4 percent real rate and a deflation rate less than the real rate. You get into trouble when the rate of deflation exceeds the real rate, because that is what drives a downward spiral of asset prices until the Fisher equation can be restored at a new higher real rate reflecting the reduced level of asset prices.

  41. Gravatar of Scott Sumner Scott Sumner
    16. November 2009 at 06:27

    David, Thanks, I’m getting a better picture of where you are coming from. A few comments/questions:

    1. How can you be sure the real return from capital was negative in October 2008? Any investor could buy a risk free TIPS with a strongly positive rate at that time. In that case, why would an investor pay a price for a real asset (say common stock or commercial property) at which the expected return was negative? I would think that those assets had to be priced at such a level where new investors expected a real rate of return at least as high as TIPS. (Obvious existing investors would earn accounting losses based on what they originally paid, but we are concerned here with ex ante, expected returns, measured in terms of economic profit, not accounting profit.)

    If we return to my original argument then I still think I am on solid ground. I argued that real interest rates don’t seem to be the criteria by which economists judge whether money was tight or not. I cited the fact that real interest rates were high in October/November 2008, but most economists thought money was easy. I still think that if an economist does use the real interest rate as an indicator of the stance of policy, he is usually thinking about bonds, not real assets. I have often heard people argue that expansionary Fed policy reduces interest rates on bonds. Rarely to I hear people discuss stocks. That’s not to say you are wrong about how real rates should be defined, but it does support my point that current views about the stance of monetary policy are incoherent.

  42. Gravatar of david glasner david glasner
    16. November 2009 at 09:46

    I’m not sure that the real return on capital was negative in October 2008, but I would interpret a deep and prolonged drop in all asset prices and a huge drop in investment as pretty good indications that the real return on capital was negative or at least (which is really all you need) less in absolute value than the expected rate of deflation. I agree that when expected future prices are such that assets cannot be profitably deployed, the value of the assets must fall until it becomes worthwhile to purchase those assets, but the process of revaluation is a nightmare, and a new equilibrium is not necessarily achieved quickly if future price expectations are very pessimistic. That the return to holding cash or Treasury bonds or TIPS exceeds the return on holding real capital is precisely the mechanism that causes an economic meltdown.

    I agree with you that you can’t measure monetary tightness or ease by looking at any single indicator because any indicator can take on a particular value for more than one reason. That is true of any market. Price is determined by supply and demand, so you can’t infer from a price change alone whether price rose because of a change in demand or supply. So I am indeed totally supporting your point about monetary policy. If the return to holding financial instruments (cash, Treasury bonds, TIPS) exceeds the return to holding capital, that indicates that there is a substantial liquidity premium and that monetary policy is tight, not easy. The reason that I posted was to provide an explanation for the behavior of TIPS in October 2008 which Bob Murphy seemed to think was inexplicable.

  43. Gravatar of Scott Sumner Scott Sumner
    17. November 2009 at 09:15

    Thanks David, Let me emphasize that I totally agree with your statement that sudden downward adjustments in asset prices are a “nightmare.” Indeed, I focused a lot on the fall in stocks, commodities and real estate last year. As you say, you can either think in terms of the fall in asset prices, or the fact that expected returns were negative before prices fall.

  44. Gravatar of TheMoneyIllusion » October 2008: the Taylor Rule vs. NGDP futures targeting TheMoneyIllusion » October 2008: the Taylor Rule vs. NGDP futures targeting
    20. January 2010 at 17:30

    […] the flaws with nominal rates.  My point is “OK, if real rates are the right indicator, then why no criticism of the Fed’s extraordinarily contractionary policy during July to November 2008, when 5-year real rates jumped […]

  45. Gravatar of Bitcoin Faucet Rotator Blog Tipsy TIPS spreads Bitcoin Faucet Rotator Blog Tipsy TIPS spreads
    10. April 2015 at 07:02

    […] and my guess it is the growing expectation of near-term negative inflation. Note: here is an older comment David makes on TIPS spreads and liquidity premiums on Scott Sumner’s blog. Note: In addition […]

Leave a Reply