Wednesday, October 30, 2013

Of Course Monetary Policy is an Asset Swap -- But that Doesn't Make it Any Less Useful


Nobel Laureate Eugene Fama made waves in the past few days when he called QE a "neutral event". Fama argued that because QE was just the exchange of one kind of interest bearing asset (money that collects IOER) for another (long term treasuries and agency MBS), the policy could have no effect. In my view, his comments were mistaken because they focused on microeconomic intuition and thus ignored the macroeconomic effects of monetary policy.

Fama's main argument was that monetary policy changed character at the zero lower bound. Starting in 2008, the Federal Reserve started paying interest on excess reserves. Once the Fed started to pay IOER, excess reserves held by banks became interest bearing assets. As a result, now when the Fed conducts QE, all it is really doing is taking the private sector's long term bonds and assets and exchanging them for short term (interest bearing) cash. So no new currency makes it into the economy, and as such QE can have no effect.

The set up is correct, but not the conclusion. By ignoring the role of expectations at the zero lower bound, Fama glosses over the real reason why QE matters. Scott Sumner explained this a few weeks ago:
So QE works for very simple reasons. Permanent monetary injections are effective even at the zero bound. QE programs are a signal that central banks would prefer at least slightly faster nominal GDP growth. Slightly faster nominal GDP growth requires that at least a small portion of the currency injection be permanent. So by signaling a preference for slightly faster nominal GDP growth, central banks are implicitly signaling a preference to have at least a small portion of the QE program be permanent (for any given IOR rate). Markets believe the central banks (and why shouldn’t they?) And hence asset prices react to the QE program.
In short, QE is effective because it changes expectations about the future monetary base. Since QE signals an increase in the monetary base in this period and all future periods, it raises expectations about future nominal income and therefore boosts the economy now. Since the short term rate will not be below the interest paid on reserves forever, then the injection of currency has a positive effect on the future price level. This is not a story of wealth effects from appreciating financial assets or a reach for yield due to lower average bond duration. It's not a story of people borrowing as the result of lower interest rates. It's just a simple tale of price expectations and forward looking monetary policy.

Another peculiar argument Fama made was that the Fed's current policy should actually be raising short term interest rates. Because the Fed is introducing more short term debt (in the form of cash) on the market, then short term rates should rise. Analogously, because the Fed is buying up so much long term debt, then the long term rates should fall.

But this misses the underlying macro context. Because the purchase of assets represents a signal about a desired monetary outcome, the result of this "asset swap" is not as neutral as Fama would believe. Moreover, because the Fed has made a commitment to keeping the Fed Funds rate low for an extended period of time, the short rate will not rise. Instead, by expanding the money supply, the Fed keeps short rate low.

Now, if I were a single market participant, the situation would be different. If I individually decided to sell short term debt to buy long term debt, then with sufficiently large quantities I could raise the short term rate and lower the long term rate. But since the Fed controls the printing presses and has the power to pin down the short term rate, this logic does not apply. Instead, if rates did rise, the Fed could just purchase more T-Bills. Even though QE introduces more short term assets into the system, it does not raise the short end of the yield curve.

Furthermore, Fed's purchases of long term treasuries are supposed to raise long rates, not lower them. Since the the long term rate goes up with higher NGDP expectations, then Fed purchases of long term bonds should raise long term rates. And as Michael Darda has repeatedly shown, rates rose during every period of QE. This is just another example of how microeconomic intuition can fail catastrophically in the world of macro. Because Fed purchases have macro level effects on inflation and economic growth, the Fed can actually buy more of an asset and have the price of it go down.

Source: FRED, MKM Partners

If all this seems counter-intuitive, do not blame yourself. Instead, the blame should go to our obsession with interest rates in models of monetary policy. Because monetary policy has historically been implemented through changes in the Fed Funds rate, people have equated monetary policy with interest rate policy. But in truth, interest rates should be seen as reflections of the money supply. So when we say that the Fed is cutting short term rates, what we really mean is that the Fed is expanding the money supply, and as a result, short term rates are falling.

Thinking in these terms will make sure you don't forget about the macroeconomic side of monetary policy. If Fama had said, "IOER means the Fed's printing of money to buy long term bonds has no effect on long term rates and actually raises short term rates", it would have been immediately clear something was off. Why would printing money have an effect on short term rates given that new mass of money can be used to buy T-Bills? Money immediately evokes macro intuition, whereas interest rates focus on micro intuition. As such, focusing on money allows you to guard against simple mistakes.

Thinking in terms of money also makes sure you don't mix causality. Ronald McKinnon argued in a recent WSJ opinion piece that the Fed should raise interest rates will stimulate banks to start lending. If you translate his statement about interest rates into money, it would read "contract the money supply so banks start lending". The first statement appeals to microeconomics, and seems sensible. But even a little thought about the second statement in terms of money immediately reveals the mistake. To stimulate lending, rates will need to fall as the money supply expands. But over time, when interest rates rise, it will be because of the monetary expansion boosting inflation.

Monetary theory is peculiar because it contradicts a lot of basic microeconomic and intuitions. As such, you often see very smart people (Nobel prize winners included) make smart-sounding arguments that are ultimately false. So for as much as I respect the work Professor Fama has done in the field of empirical finance, I disagree with his description of QE. It's not some neutral event, and to think so distracts from the urgent task of monetary reform.
---
DeLong offers his comments here.

Edit: 10/31 -- Added QE and long term rates graph from Michael Darda

57 comments:

  1. Nice post. I'm not sure I get your distinction between interest rates as a micro concept and money as a macro concept, can you clarify? Also, by what channels does raising expectations of future nominal income boost the economy now (if not borrowing related?)

    "Thinking in terms of money also makes sure you don't mix causality," might be a risky thing to say.

    Looking forward to the comments/discussion on this post!

    ReplyDelete
    Replies
    1. Grr. I typed up a response then pressed the wrong button. Let's try again.

      My underlying point was that the interest rate can be thought of as the price of certain assets, and that evokes a certain microeconomic intuition about asset pricing. On the other hand, the money supply is decidedly a macro concept, and, at least personally, is more reflective of general equilibrium concepts. As such, by thinking in terms of M and not r (at least for these kinds of casual conversations), makes sure you don't mix up the domains.

      The McKinnon example really stuck out to me. What he said in terms of interest rates can sound somewhat meaningful, but if you translate it in terms of M it is absolutely ludicrous.

      I also think framing monetary policy in this light clears up the "save the savers" issue. If you think of what they're saying as "the economy is bad, but we should contract the money supply and hurt others so the money I hold is worth more", then their position seems much less defensible than "we have a right to positive interest rates"

      As to how M affects P, I think of it as a Hume/specie flow issue. And even if the effect goes through interest rates, that's a side effect. Focus on the M, and r will follow. But looking at it the other way around leads to mistaken analysis.

      Delete
    2. "But in truth, interest rates should be seen as reflections of the money supply. So when we say that the Fed is cutting short term rates, what we really mean is that the Fed is expanding the money supply, and as a result, short term rates are falling"

      If the short term rate is determined by the rate of interest the central bank pays on reserves, then doesn't raising or lowering the IOR rate change the short term rate without changing the base money supply?

      Also, if the central bank were to set its discount rate equal to its target short term rate, then could it not similarly change the short term rate without changing the base money supply?

      Delete
    3. Anonymous12:22 PM

      Once again you demonstrate a remarkable lack of understanding about a topic you insist on publicly commenting on, namely macroeconomics. Case in point:

      "My underlying point was that the interest rate can be thought of as the price of certain assets, and that evokes a certain microeconomic intuition about asset pricing. On the other hand, the money supply is decidedly a macro concept, and, at least personally, is more reflective of general equilibrium concepts. As such, by thinking in terms of M and not r (at least for these kinds of casual conversations), makes sure you don't mix up the domains."

      This paragraph doesn't make a bit of sense. Please go to graduate school, learn some economics, then come back.

      Delete
  2. I have a problem with your title, Yichun. Monetary policy "is not" an asset swap. LSAP is an asset swap, but that is not the sum total of monetary policy. Explicit setting of interest rates as the Bank of England does is also monetary policy, but it isn't an asset swap.

    I also think you glide too lightly over collateral effects. LSAP both increases the monetary base and reduces the quantity of safe assets. Its effects are not achieved purely because of future price expectations arising from the increase in the monetary base, They are also achieved through price expectations arising from reduced quantity of (non-monetary) safe assets. Cash is not always a substitute, even if it is interest-bearing (because of IOER). Portfolio effects matter.

    Regarding long-term rates - I presume you are talking about conventional QE, not Operation Twist- type measures? Operation Twist was designed to depress long-term rates, not raise them.

    ReplyDelete
    Replies
    1. I do make this mistake a lot, don't I? I guess I should have been clear that what I meant was open market operations in general, for which QE is a subset.

      I actually did take a look at the collateral issue in the past, and wrote about it in the context of IOER here: http://synthenomics.blogspot.com/2012/08/interest-on-excess-reserves-illustrated.html. But since then, I'm not sure if I'm finding much convincing evidence that the safe asset issue has a large enough of an empirical effects. We have seen a dramatic collapse in shadow banking credit, but that hasn't stopped QE from offsetting our negative fiscal policy shocks. And given the way equity, TIPS, and gold markets reacted to QE announcements, I'm not convinced of the empirical applicability of this theory (even though it sounds like a fun thing to model).

      And perhaps that was the intention on Twist, but I'm not sure if was successful at all, and if so that would support my model. Because the Fed didn't actually make a commitment to changes in M, that policy had minimal effect.

      Delete
  3. "In short, QE is effective because it changes expectations about the future monetary base. Since QE signals an increase in the monetary base in this period and all future periods, it raises expectations about future nominal income and therefore boosts the economy now. Since the short term rate will not be below the interest paid on reserves forever, then the injection of currency has a positive effect on the future price level. This is not a story of wealth effects from appreciating financial assets or a reach for yield due to lower average bond duration. It's not a story of people borrowing as the result of lower interest rates. It's just a simple tale of price expectations and forward looking monetary policy."

    But surely expectations of future nominal income have to based on something. Nominal income expectations depend on expectations of higher wealth and lending. Just permanently increasing base may not make expected income greater. There has to be some realistic mechanism in place whereby people actually believe it will happen otherwise the fed will just create excess reserves. The fed has to be capable of actually attaining its goals otherwise expectations won’t change.

    The fed needs to incorporate a more effective mechanism whereby it deals directly with the public who have a higher MPC which will increase velocity. The current asset holders have a low average MPC which is making MP very ineffective.

    ReplyDelete
  4. Anonymous1:12 AM

    Good Lord. If people are making important decisions based on their "expectations about the future monetary base", they are blooming idiots, since the size of the monetary base has hardly any causal relevance to anything.

    Of course, I can't rule out that part of Fed policy is to massage the expectations of idiots who are mentally encumbered with extinct dinosaur theories about the significance of the monetary base that the Fed's own researchers do not themselves believe. I would like to see some empirical evidence, however, that counts how many of these folks there are.

    Also, it seem to me that a hypothesis that posits an important role for expectations of the future size of the monetary base, without any direct empirical evidence that such expectations are playing a role, is not an alternative to microeconomic intuition, it is itself just another intuition, and it is equally microeconomic: it is a hypothesis about the psychology and behavior of market participants.

    BTW, while Fama is right about the asset swap concept, the fact that the Fed is now paying IOR is irrelevant to that point. Any way you slice it, it's an asset swap. If the rate of interest on reserves were 0%, that would just mean that it was swapping a non-maturing, 0% interest asset for a maturing, interest-bearing asset. Dollars are a kind of asset. The point is that QE consists of market transactions. The Fed doesn't stuff dollars into people's pockets. It offers dollars for treasuries and agency securities; and the exchange occurs because the sellers of the assets accept the offered dollars. These sales are competitive, which means that the assets surrendered are very close in value, from the seller's point of view, to the dollars obtained in their stead. So as Fama said, one kind of asset goes off the seller's balance sheet, and another asset of roughly equal value goes onto it. That's why the impact is negligible. The rate of IOR just determines what kind of swap it is. If the IOR rate was lower, the seller's would demand more dollars.

    Also, Fama's point has nothing to do with "no new currency making it into the economy". This way of putting it seems to presuppose that if reserve balances were not earning interest on reserves, they would be fleeing bank reserve accounts and going "out" into the economy as currency. This picture shows a misunderstanding of the role of reserve balances in the nation's payment system. There is no "out".

    ReplyDelete
    Replies
    1. Anonymous7:11 AM

      You can try Yahoo's Breakout videos and CNBC for talking heads that believe an increase in the monetary base increases expected future nominal income. People that believe it are probably more apt to say it publicly than the people that don't believe. On the other hand, Bernanke said in 2012 that recent research showed asset purchases did cause downward pressure on yields, so if that is the cause of higher nominal income expectations, then in an indirect way, the "idiots" have a point.

      Delete
    2. Anonymous9:32 AM

      I'm not denying that Fed asset purchases exert downward pressure on yields for whatever class of assets the Fed might purchase. Whenever a new, large customer aggressively enters the market for any kind of asset, the price of the asset will go up and the yields will go down. These are interest rate channel effects. I'm just extremely skeptical that there is any efficacious monetary policy channel that depends on the monetary base.

      The number of people who listen to the CNBC talking heads must be limited, because despite a high level of voluble inflation hysteria coming from that direction during QE1, actual inflation was not noticeably affected.

      Delete
    3. I would argue that asset price responses to QE suggest that expectations do matter. These equities can't go up on the backs of empty air, and even when the discount rate rose after the initial taper announcements, equity prices stayed strong. Moreover, the breakdown of the Philips curve in the 1970's does suggest that expectations matter, and that's why we generally accept that principle when doing this kind of modeling.

      And there's nothing wrong with microeconomics -- you can model everything I'm saying in a complicated microeconomic framework. It's just that you can't use your first instincts (more demand = higher price), because there are other things going on (more fed demand -> less private demand b/c of monetary expansion -> lower price).

      This theory isn't precisely about the exact size of the monetary base -- saying so is shorthand for saying "expected future monetary policy". But given I wanted to emphasize the importance of M, I phrased it in that way.

      For more on the yield/price issue, I addressed that in the post. The idea here is that monetary effects make your initial supply/demand analysis incorrect.

      Delete
    4. Anonymous1:02 PM

      I think you meant to type something different than "discount rate"?

      A problem with the expectations-management paradigm is that it really has nothing to do with monetary policy per se. For example, if enough people came to believe that polyester fabrics emit green kryptonite vapors that destroy healthy animal spirits, then I suppose that if the Fed announced it was trading a million Ben Bernanke bumper stickers each month for large quantities of polyester shirts to take some of those shirts off of household balance sheets, that might have some role in moving expectations about the future state of the economy. And the improvement in expectations might be self-reinforcing. You might call these expectations about the future path of "monetary policy". But it really has nothing to do with monetary policy apart from the fact that the Fed is the one dispensing the bumper stickers.

      Is the expectations channel for the Fed really based on any well-founded fundamentals having to do with monetary policy, or is the monetary dimension of these policy announcements merely accidental and arbitrary - given that (i) the Fed happens to be an authority center, (ii) money, rather than bumper stickers, just happens to be the thing the Fed manufactures, (iii) by convention and tradition, the Fed does its direct business with financial firms, and financial securities just happen to be the things these firms can swap for the stuff the Fed manufactures?

      Also, there are many different centers of economic power and authority in the country, and so the statements of all of them influence expectations in various degree, and sometimes in competing ways. Are the Fed's statements that much more important than the statements of other authorities?

      These days, a certain number of people have become convinced - mainly due to the efforts of a persistent and garrulous flock of bloggers, financial industry gurus and press, and monetarist economists - that QE programs, though some mysterious process that they don't really understand, is "keeping the economy afloat" by "pumping money into the economy". These slogans are repeated uncomprehendingly and ad nauseum by some economists and folks in the financial press. Naturally, then, if the Fed signals that it might taper the asset programs, that causes a degree of panic and brearishness: "My God, the Fed is tapering! Tapering I say! We're doomed!"

      But if the Fed spent a month releasing and publicizing the increasing barrage of research arguing that there is little fundamental basis for these kinds of expectations apart from their own self-perpetuating, self-fulfilling nature, it could announce a plan to taper QE without producing any harmful effects. At this point, the bulk of the impact is probably psychological.

      This is no way to manage an economy. Techniques based on the management of public expectations grounded in intellectual fads are inherently temporary, variable and unpredictable - and can even reverse themselves suddenly and without warning. And right now, for every person who thinks that QE is "keeping the economy afloat", there is probably someone else who thinks it is is burying the economy in the graveyard of history. Which way is the expectations game actually cutting?

      The QE mania seems like a further syndrome in the decadent postmodern devolution of our economy into a an airy confection made up of worthless financial fluff, gumming up a financial industry that is far larger than historical experience would suggest is appropriate. The sooner people realize that not much hangs on whether QE comes or goes, the better off we will be.

      We need to start investing our attention and policy resources in the the creation of more tangible assets of durable and proven human value, things whose usefulness to wealth creation doesn't come and go with the winds of popular fads and ivory tower dreamwork.

      Delete
  5. Carola, you definitely should read this post: http://www.themoneyillusion.com/?p=24433 So if you're a person of the concrete steppes, then sure you can think about asset prices, wealth effects, and excess money-balances, but it's really about the economy snapping to a new nominal (and real) equilibrium.

    Frances, shouldn't you think about the supply and the demand for safe assets? With that in mind, David Beckworth's posts may be helpful: http://macromarketmusings.blogspot.com.au/2011/12/why-global-shortage-of-safe-assets.html
    http://macromarketmusings.blogspot.com.au/2013/01/resolving-safe-asset-shortage-problem.html

    "contract the money supply so banks start lending" does not seem sensible to me at all. "But in truth, interest rates should be seen as reflections of the money supply." - a more Sumnerian thing to say would be to acknowledge that the myriad factors reflected in interest rates make it a very weak signal for the workings of monetary policy, and should be discarded as a Keynesian obsession.

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
  6. Sorry, these are even better Beckworth posts, read these instead: http://macromarketmusings.blogspot.com.au/2013/06/is-fed-squeezing-shadow-banking-system.html
    http://macromarketmusings.blogspot.com.au/2013/10/what-george-bailey-can-teach-us-about-qe.html

    He also has a new paper out on the subject: http://economics.olemiss.edu/files/2013/09/MonetaryTransmission0913.pdf

    ReplyDelete
  7. Good post.

    Paying interest on reserves (raising it from zero) will tend to raise short term interest rates. And paying higher interest on reserves will raise short term interest rates more.

    Open market purchases of long term assets with interest bearing reserves won't raise short term interest rates. It might lower long term rates, but as you correctly explain, by impacting expectations of future spending on output, the result could be for long term rates to rise. It requires that private investors sell more long term securities than the Fed is buying. And if those selling the long term securities purchase capital goods (or consumer goods,) then the higher interest rates are associated with more spending on output, not less.

    As for some of the comments. When firms hold cash (piles of shot term securities,) they are lending. If they sell those short term securities and buy capital goods, they are lending less and investing more. This will tend to raise interest rates on short term securities.

    ReplyDelete
  8. Anonymous7:17 AM

    Bernanke said in 2012 that recent research showed asset purchases did cause downward pressure on rates. It's not the same as saying rates went down, and in fact, they tended to go up.

    ReplyDelete
    Replies
    1. Well, I don't think Bernanke actually believes that an expansionary monetary policy would lower rates in the medium run. He is well aware that interest rates are a misleading guide to monetary policy. But of course, he cannot say that because he doesn't have the time to explain money theory to the monkey circus that is Congress.

      Delete
    2. Anonymous12:23 PM

      Maybe he has time to explain it to you, because you clearly don't understand it.

      Delete
  9. The role of expectations to make QE work undoubtedly has some positive-towards-increased-activity effect.

    The counter-activity reason that QE works slowly, if at all, is mentioned in your post but is so important, that it is worthy of restating..

    The Treasuries and MBS that-are-purchased-by-the-Fed are held by decision makers who are long term investors or represent long term investors. These investors would not trade interest bearing assets for cash unless they perceived the buyer was over-paying. Once the trade was made, the decision makers (now holding cash) have the opportunity to re-invest cash into other assets.

    The banks holding accounts belonging to decision makers have more reserves and can make more loans. The current lending climate is hostile with low interest rates and extraordinarily turbulent forward expectations.

    No doubt, decision makers evaluate both micro-economic and macroeconomic factors to make reallocation decisions.

    ReplyDelete
    Replies
    1. Anonymous1:23 PM

      This comment has been removed by the author.

      Delete
    2. Anonymous1:34 PM

      The banks holding accounts belonging to decision makers have more reserves and can make more loans.

      Reserve accounts are already so chock-full of excess reserves, it is hard to believe that the addition of more can be be having any impact on the marginal willingness to make loans. There are no liquidity constraints holding back lending. Plus firms are already sitting on plenty of their own cash.

      And are the people selling assets to the Fed really long-term investors in those assets. Or is the Fed's aggressive participation in those markets creating opportunities for short-term, but very modest, arbitrage profits? The net effect here seems to be that the Fed is dribbling small net amounts of cash income into sectors that don't really need it, and that only have the ability to drive demand for other financial products, while the sectors that do need income and that drive demand on the ground are being income starved.

      Delete
  10. Good post.

    "Since the short term rate will not be below the interest paid on reserves forever, then the injection of currency has a positive effect on the future price level."

    Do we know this for sure? If the market expects the Fed to use interest on reserves (IOR) as its main policy instrument for the next few decades (ie. if the short term rate hugs IOR and moves up or down with changes to IOR), then currency injections now will be irrelevant. The important lever in this case is open mouth operations -- telling the market at which level IOR will be in the future.

    Even if the market expects short term rates to rise a fraction above IOR in the coming decades, then permanent open market operations in the present can only operate by narrowing the expected value of this rather skinny fraction. Yes, open market purchases will have *some* effect on the future price level, but much less than if IOR didn't exist at all (as was the case pre-2008). If IOR was 0% forever, permanent open market purchases would operate on much wider fraction, and therefore have a much larger effect on the future price level. (I think this is why Scott Sumner doesn't like IOR -- it throws a wrench in the effectiveness of permanent open market purchases)

    ReplyDelete
    Replies
    1. In reality, I believe QE is more of a commitment mechanism to easy money, rather than about the precise size of the monetary base -- I abstracted that away in this post, however. I would generally agree that IOER is contractionary, and that with a lower IOER, QE would be more expansionary. However, I'm not sure if you would want to get rid of IOER, because it does have other financial stability benefits. In that case, I would want the T-Bill rate to be above the IOER rate, ideally with both of them positive.

      Delete
  11. Diego Espinosa11:31 AM

    Injections of reserves are never permanent in an inflation targeting regime. Reserves will grow or shrink as needed to hit the policy rate needed to, in turn, hit the Fed's target.

    Yichuan, please give us a concrete example of how a "permanent injection" works. We have $2.3tr in ER's today. Let's say LSAP's communicate some "permanence" in the future. How much will the Fed retain in ER's in what year? Pick any year. How will it know that this is the right number to commit to? How will it control inflation once that number is known to be "permanent"?

    With regard to IOR-paying reserves, they render the notion of "permanence" irrelevant. There is no difference between shrinking the base by raising the FFR and increasing the opportunity cost of existing Fed deposits by raising the IOR.

    ReplyDelete
  12. Fama never ruled out the expectations channel. From econtalk, Fama, "[On controlling the interest rate] Maybe they can tweak it a bit; they can do a lot with inflationary expectations."

    It's a strange dance the Fed does to change expectations. And while I agree that that is the method that QE works primarily through, I have no confidence that everyone on the Fed board agrees with that. Fama is criticizing the rationale they actually give for their policy.

    ReplyDelete
    Replies
    1. That's reassuring. If so, I would absolutely agree with Fama's true belief that lowering rates is a bad justification for QE.

      Delete
  13. The Market Fiscalist10:44 AM

    "But in truth, interest rates should be seen as reflections of the money supply"

    I don't understand this. Perhaps if the monetary base was fixed this may be true but not in today's economy.

    When a CB targets interest rates it is in effect committing to meet the economy's demand for the mix bonds and money at that interest rate (via OMO). If it targeting inflation it will adjust the interest rate up or down in response to changes in inflation trends. This change of interest rate will cause the public's mix of bonds and money to change and this will have an effect on spending (and inflation). This causality appears to run from interest rates to the money supply and not from the money supply to interest rates.





    ReplyDelete
  14. "it’s quite straightforward to demonstrate that there is virtually zero relationship between changes in the monetary base and subsequent job growth, nor is there any inverse relationship between inflation and unemployment"

    ouch!!! this is absolutely devastating to the Keynesians.

    ReplyDelete
  15. Yichuan Wang:
    "Furthermore, Fed's purchases of long term treasuries are supposed to raise long rates, not lower them. Since the long term rate goes up with higher NGDP expectations, then Fed purchases of long term bonds should raise long term rates. And as Michael Darda has repeatedly shown, rates rose during every period of QE."

    It seems obvious to me, and to many other people, but where are the academic studies confirming that this is true?

    Most highly publicized academic studies on QE seem to come in four flavors: 1) event studies on changes in security yields on the days of announcement (e.g. Krishnamurthy and Vissing-Jorgensen, 2011), 2) panel data studies on flow and stock effects of QE on daily security yields during the programs (e.g. D'Amico and King, 2010) , 3) times series studies on the effect of open market operations during normal times (e.g. Hamilton and Wu, 2011) and 4) studies on the macroeconomic effects of QE using major models calibrated to normal times (e.g. Fuhrer and Olivei 2011). In short there's nothing in the way of empirical studies on the macroeconomic effects of QE, and the underlying assumption of nearly all these studies is that the primary Monetary Transmission Mechanism (MTM) channel is the Traditional Real Interest Rate Channel, which is almost certainly not the case at the zero lower bound (ZLB).

    That's not to say there are no academic studies confirming what is obvious to the eye. There’s “An Injection of Base Money at Zero Interest Rates: Empirical Evidence from the Japanese Experience 2001–2006" by Yuzo Honda, Yoshihiro Kuroki, and Minoru Tachibana (March 2007):

    http://www2.econ.osaka-u.ac.jp/library/global/dp/0708.pdf

    Figure 3 shows that Honda et al finds the original Japanese QE raised the yields of 5, 7 and 10 year bonds by a statistically significant amount. But to my knowledge this is the only academic study with this result.

    Honda et al is a Vector Autoregression (VAR) study that uses reserve balances as a variable. By my reckoning there are only a dozen VAR studies that use either reserve balances or the monetary base as a variable. Of those, only two focus on a period of time when the policy rate was at the ZLB. (The other study is "Quantitative easing works: Lessons from the unique experience in Japan 2001-2006 by Eric Girardin and Zakaria Moussa (January 2010). Using rather more complicated methods it does not seem to find a statistically significant effect of QE on bond rates.)

    VAR studies on monetary policy used to be quite common in the 1980s and 1990s (e.g. Sims, Bernanke/Blinder, Christiano/Eichenbaum/Evans etc.). One would think with three recent reasonably long zero lower bound episodes with significant amounts of QE (Japan 2001-06, the US 2008-present and the UK 2009-present) that more VAR studies would have been done on these incidents. I've estimated a VAR similar to Honda et al for the US over the period since December 2008 using industrial production, the PCEPI, the monetary base and the 10-year T-Note as variables, and find QE increases 10-year T-Note yields by a statistically significant amount, but then our eyes already told us this was the case.

    P.S. I've also run Granger causality tests using the Toda and Yamamato technique on the US monetary base since December 2008, and I find the monetary base Granger causes 10-year T-Note yields, the real broad dollar index, the S&P 500, the DJIA, the PCEPI, 5-year inflation expectations using TIPS, commercial bank deposits and commercial bank loans and leases. The impulse responses are all consistent with what theory predicts. With the exception of the 10-year T-Note, these variables are important to the Exchange Rate Channel, the Tobin Q Channel, the Wealth Effects Channel, the Bank Lending Channel, the Balance Sheet Channel, the Unanticipated Price Level Channel and the Household Liquidity Effects Channel of the MTM. So speaking strictly empirically, QE is definitely not a "neutral event".

    ReplyDelete
    Replies
    1. Mark,

      That sounds like fantastic work. Do you think you could email me what results you've found? I'm doing some time series research with Miles Kimball right now and those kinds of materials might be helpful in clarifying some of the statistical issues in my mind.

      Delete
  16. "Furthermore, Fed's purchases of long term treasuries are supposed to raise long rates, not lower them." Are you sure you meant to say that? Why would higher nominal income with a smaller stock of LT assets make their prices go *down*? Are LT bonds inferior assets and have a negative income elasticity of demand? Here's what the Board of Governors' website says about their LSAP program: "...The Fed's purchases reduce the available supply of securities in the market, leading to an increase in the prices of those securities and a reduction in their yields." Are you suggesting the Fed is trying to trick markets by saying this, and that their intent all along is to drive mortgage rates higher? Clearly I'm thinking in micro terms, but I'm also trying to apply common sense. The graph of 10-year T-note rates you posted demonstrates that when the Fed buys LT assets, investors start dumping them. But as the stock of these assets decreases their price goes up (rates come down). I hope I didn't totally misunderstand what you were saying. But the discussion here appears to be falling prey to over complicating the Fed's intent and missing the forest for the trees.

    ReplyDelete
    Replies
    1. There are multiple theories about the term structure of interest rates:

      http://en.wikipedia.org/wiki/Yield_curve

      The Liquidity Premium and Preferred Habitat theories are the most widely accepted theories of term structure of interest rates because they explain the major empirical facts about the term structure so well. They also combine the features of both the Expectations Theory and the Segmented Markets Theory by asserting that a long term interest rate will be the sum of a liquidity premium and the average of the short term interest rates that are expected to occur over the life of the bond.

      So, for a given liquidity premium, if long term interest rates change this means that the average of the sum of expected inflation and expected real interest rates have changed. In turn, short term real rates reflect monetary stance and are likely to be low when real GDP is growing too slowly and high when real GDP is growing too fast.

      So theoretically long term rates should correlate well with the sum of expected inflation and the expected real GDP growth rate, which is of course the expected NGDP growth growth rate. Do they?

      Naturally:

      http://3.bp.blogspot.com/-fQOh2jjNS_8/T2kdMyGHxAI/AAAAAAAACXI/y5k6QONVFGM/s1600/treasyield_NGDP.jpg

      In fact this has been corroborated by research:

      http://papers.ssrn.com/sol3/papers.cfm?abstract_id=270936

      Delete
    2. Mark, Thank you for replying to my comment.

      Yes, but the liquidity premium is not a constant. It varies somewhat with the stock of the asset. These are assets with prices that are in part subject to supply and demand because part of the demand is for reasons other than yield: many institutions are required to hold AAA-rated assets; and some risk averse investors are willing to pay a premium for these assets. Even if the Fed is increasing M when they purchase them, they are still reducing the stock of these assets.

      Perhaps I misunderstood Yichuan's post on this point, but I thought part of what he was saying is that LSAP *raises* LT rates, which I disagree with. The Fed said they intended to lower LT rates by reducing the stock of MBS and LT treasuries, and that is what happened. Look at mortgage rates for example. I realize there are people on this blog who believe these rates went down for other reasons, but I haven’t seen any evidence to convince me this is so.

      Clearly there are other channels involved influencing expectations, but these channels don't appear (to me at least) to fundamentally change the relationship between the supply and demand for these assets.

      Please reply if you find fault in anything I said. Thanks again, -Mark`

      Delete
    3. "Yes, but the liquidity premium is not a constant. It varies somewhat with the stock of the asset...Even if the Fed is increasing M when they purchase them, they are still reducing the stock of these assets."

      The nominal income effect is far more important than the liquidity premium effect. Even now the Fed is a relatively minor holder of Treasuries (16.3% of all outstanding in 2013Q2) and Agencies (about 16.7% of all outstanding in 2013Q2) but it is responsible for creating 100% of all U.S. base money.

      "Perhaps I misunderstood Yichuan's post on this point, but I thought part of what he was saying is that LSAP *raises* LT rates, which I disagree with."

      To be absolutely clear, Yichuan is saying QE *raises* longer-term interest rates.

      "The Fed said they intended to lower LT rates by reducing the stock of MBS and LT treasuries, and that is what happened. Look at mortgage rates for example. I realize there are people on this blog who believe these rates went down for other reasons, but I haven’t seen any evidence to convince me this is so."

      QE1 was announced on November 25, 2008 and concluded on March 31, 2010. The 10-year T-Note closed at 3.35% on November 24, 2008 and closed at 3.84% on March 31, 2010, an *increase* of 49 basis points.

      QE2 was hinted at in Jackson Hole on August 27, 2010 and concluded on June 30, 2011. The 10-year T-Note closed at 2.50% on August 26, 2010 and closed at 3.18% on June 30, 2011, an *increase* of 68 basis points.

      QE3 was hinted at in Jackson Hole on August 31, 2012. The 10-year T-Note closed at 1.63% on August 30, 2012 and 2.69% on November 5, 2013, an *increase* of 106 basis points.

      http://research.stlouisfed.org/fred2/series/DGS10

      30-year mortgage rates averaged 3.59% the week ending August 30, 2012 and 4.10% the week ending October 31, 2013, an *increase* of 51 basis points.

      http://research.stlouisfed.org/fred2/series/WRMORTG

      Furthermore, as I mentioned above, I've estimated a VAR similar to Honda et al for the US over the period since December 2008 using industrial production, the PCEPI, the monetary base and the 10-year T-Note as variables, and find QE increases 10-year T-Note yields by a statistically significant amount.

      This phenomenon has also been repeatedly observed by the popular media:

      http://www.theatlantic.com/business/archive/2013/01/the-1-thing-the-worlds-smartest-people-dont-get-about-the-fed/266812/

      I can't explain why the FOMC QE announcements say things like "these actions should maintain downward pressure on longer-term interest rates." They clearly don't, and having read most of Bernanke's research I assure you he is well aware that if QE is effective, it should raise longer- term interest rates.

      "Clearly there are other channels involved influencing expectations, but these channels don't appear (to me at least) to fundamentally change the relationship between the supply and demand for these assets."

      The macroeconomic effects of QE far outweigh any microeconomic effects it may have.

      Delete
    4. Thank you again Mark for replying to my comment. Clearly you have thought a lot about this (certainly more than I have). Could you point me to the research by Ben B. on this subject? I don't doubt that you are correct but other readers of this blog have referenced a statement he made in 2012 saying that recent research showed asset purchases did cause downward pressure on yields.

      Why did yields come down so much after QE 1 and 2 ended? Did income expectations drop that much? But then why did the stock market continue to rise while yields continued to fall? Why did the stock market drop so much at the hint of the taper (but rates jumped up)?

      Why is the long-term composite treasury rate (for durations > 10 years) today 100 basis points lower than it was at the end of QE1 (3.41 today vs. 4.41 on 3/31/2010), and 57 basis points lower than at the end of QE2 (3.98 on 6/30/2011)? Are income expectations lower now than then?

      Didn’t Japanese government bond rates go down with the advent of Abe Shinzo?

      It would be interesting to see a large-scale survey of home buyers to find out what drove their purchase decisions during this period. I would think it was lower mortgage rates rather than higher income expectations based on some amount of permanent increase in the monetary base. This is totally anecdotal, but the people I know who purchased homes all had saved for many years and because of low rates were able to buy a house they wanted. Mortgage rates make a *huge* difference for home buyers.

      In December 2008 the monthly average commitment rate on 30-year fixed-rate mortgages was 5.29%. In October 2013 it was 4.19% (it was 3.54% in May before the talk of tapering). Does this reflect lower income expectations now than in December 2008?

      Do you believe that home buyers purchase homes because they *consciously* believe QE will permanently raise their nominal income? And do their income expectations jump and fall when the Fed starts and stops its asset purchases?

      These are real questions, not rhetorical ones. I hope I do not sound like I’m being sarcastic.

      Thank you again Mark, I do appreciate your clearly expert opinion and I *am* trying to understand this all better. But I am having trouble meshing your view with individuals' decisions. If you feel like replying I would enjoy hearing from you.
      -the other Mark


      Delete
    5. "Could you point me to the research by Ben B. on this subject?"

      Rather than a paper, how about a recent speech on the subject?

      "Long-Term Interest Rates"
      March 1, 2013

      http://www.federalreserve.gov/newsevents/speech/bernanke20130301a.htm

      "I don't doubt that you are correct but other readers of this blog have referenced a statement he made in 2012 saying that recent research showed asset purchases did cause downward pressure on yields."

      Event studies show statistically significant declines in longer-term yields on the *days of QE announcements*. This probably reflects the liquidity premium effect.

      "Why did yields come down so much after QE 1 and 2 ended? Did income expectations drop that much?"

      Yes.

      "But then why did the stock market continue to rise while yields continued to fall?"

      QE1 concluded on March 31, 2010 and QE2 was hinted at in Jackson Hole on August 27, 2010. The S&P 500 closed at 1169 on March 31, 2010 and 1047 on August 26, 2010, a *decrease* of 122 points.

      QE2 concluded on June 30, 2011 and QE3 was hinted at in Jackson Hole on August 31, 2012. The S&P 500 closed at 1321 on June 30, 2011 and 1433 on August 30, 2012, an increase of 112 points.

      http://research.stlouisfed.org/fred2/series/SP500/

      So, on average, 100% of the increase in the stock market has occured during QE.

      "Why did the stock market drop so much at the hint of the taper (but rates jumped up)?"

      Did it? Ten-year T-Note yields surged in May and June. With the exception of a few days in late June the S&P 500 has consistently closed *higher* than it was in early May:

      http://research.stlouisfed.org/fred2/graph/?graph_id=144974&category_id=0

      So it's not clear to me that the "taper talk" has had much effect one way or another.

      "Why is the long-term composite treasury rate (for durations > 10 years) today 100 basis points lower than it was at the end of QE1 (3.41 today vs. 4.41 on 3/31/2010), and 57 basis points lower than at the end of QE2 (3.98 on 6/30/2011)? Are income expectations lower now than then? "

      Yes.

      "Didn’t Japanese government bond rates go down with the advent of Abe Shinzo?"

      The BOJ announced its QE program on April 4, 2013:

      http://www.boj.or.jp/en/announcements/release_2013/k130404a.pdf

      The 10-year JGB yield closed at 0.56% on April 3, 2013. It dropped to 0.45% on the day of the announcement. But then it rose to 0.93% on May 29. It's fallen since then but it closed at 0.60% on November 7, an *increase* of 4 basis points since before QE was announced:

      http://www.bloomberg.com/quote/GJGB10:IND

      "Mortgage rates make a *huge* difference for home buyers."

      That's a *huge* myth. If anything there's been a positive correlation between mortgage rates and housing construction since 2009:

      http://research.stlouisfed.org/fred2/graph/?graph_id=129829&category_id=0

      Usually there's no correlation at all between mortgage rates and housing construction. By far the most important factors affecting residential investment are household incomes and financial wealth.

      "In December 2008 the monthly average commitment rate on 30-year fixed-rate mortgages was 5.29%. In October 2013 it was 4.19% (it was 3.54% in May before the talk of tapering). Does this reflect lower income expectations now than in December 2008?"

      Yes.

      "Do you believe that home buyers purchase homes because they *consciously* believe QE will permanently raise their nominal income?"

      Not necessarily "consciously", but yes.

      "And do their income expectations jump and fall when the Fed starts and stops its asset purchases?"

      Yes.

      "These are real questions, not rhetorical ones. I hope I do not sound like I’m being sarcastic."

      You don't, but you do seem to know an awful lot of stuff that "just ain't so".

      Delete
    6. This comment has been removed by the author.

      Delete
    7. “...but you do seem to know an awful lot of stuff that "just ain't so".”

      -Thank you for your vote of confidence… :)

      I’m not sure how we can be looking at the same data:

      Income expectations are not lower now according Michigan survey data: (http://www.federalreserve.gov/econresdata/notes/feds/2013/why-have-americans-income-expectations-declined-so-sharply/). Granted, they’re nowhere near where they were before the recession.

      The S&P 500 rose about 6% between the end of QE2 and the hint of QE3. During this same time the monthly average 30-year fixed mortgage rate went from 4.51 to 3.60 (June 2011 vs. August 2012).

      Permits for new residential housing units increased by about 30% during this period. The trend growth flattened out during QE3 as rates went up.

      If you plot newly permitted residential development during this period you see that it took off after QE2 while mortgage rates were diving. Once QE3 started, the trend growth for housing permits slowed as mortgage rates went up. It’s quite negatively correlated with mortgage rates during that past couple of years. Since rates went up in June housing starts slowed.

      Gross fixed capital formation went down during QE1 until long-term rates started falling at which point it started to rise. It went down during QE2 as rates rose, but then started growing when rates began falling precipitously. It grew steadily and significantly between the end of QE2 and the beginning of QE3 while rates were diving. It faltered in the beginning of QE3 as rates went up, but then continued to grow. From the beginning of QE1 until around January 2013 it is negatively correlated with interest rates.

      Gross private domestic investment was similar. It too went down until rates started falling at the end of QE1 at which point it turns upward until rates started going up at the beginning of QE2 at which point it started to fall. It started going up again when rates started dropping at the end of QE2. In particular it jumped up significantly when QE2 ended while rates fell dramatically. It did, however, start growing even faster during QE3 even while rates went up. But between the beginning of QE1 and the beginning of QE3 it is negatively correlated with interest rates.

      Japanese government bond yields most certainly came down in the run up to Shinzo’s election and continued to fall. To be fair, they appear to have been falling before it became apparent he would be elected. But they are lower now than when he was elected.

      The stock market dropped more than 4% in the days following the taper discussion and didn’t fully recover until the Fed decided to postpone the taper.

      I’m not saying the income expectations channel isn’t there, but it doesn’t appear to have changed the fact that investment and house purchases have been negatively correlated with rates for the past few years and that investment recovered in large part because rates went down (and according to the Michigan survey income expectations have risen).

      Gross fixed investment appears to be becoming positively correlated with rates (which I presume is what we want), but housing isn’t quite there yet.

      Thank you Mark for putting up with my insistent questions and thanks for the BB speech reference.
      -Mark

      Delete
    8. This comment has been removed by the author.

      Delete
    9. "Income expectations are not lower now according Michigan survey data:"

      Actually they are. The survey can be found here:

      http://press.sca.isr.umich.edu/

      I did extensive econometric work on this very data in April. The survey data comes in monthly frequency but the following quarterly averages should give you the idea: The following are expected percent change in nominal household income one year from the survey:

      2006Q1-3.8
      2006Q2-4.0
      2006Q3-3.6
      2006Q4-4.6
      2007Q1-4.1
      2007Q2-3.7
      2007Q3-3.8
      2007Q4-3.2
      2008Q1-2.9
      2008Q2-1.8
      2008Q3-2.7
      2008Q4-1.0
      2009Q1-(-0.6)
      2009Q2-0.6
      2009Q3-0.5
      2009Q4-0.9
      2010Q1-0.7
      2010Q2-1.4
      2010Q3-1.5
      2010Q4-1.7
      2011Q1-1.0
      2011Q2-0.9
      2011Q3-0.9
      2011Q4-1.2
      2012Q1-1.6
      2012Q2-1.1
      2012Q3-1.8

      Here are a couple of good blog posts on the Survey:

      http://macromarketmusings.blogspot.com/2013/04/the-ongoing-dereliction-of-duty.html

      http://macromarketmusings.blogspot.com/2013/05/balance-sheet-recessions-are-really.html

      And here's another good paper on the Survey:

      http://www.chicagofed.org/digital_assets/publications/economic_perspectives/2013/1Q2013_part2_french_kelley_qi.pdf

      I suggest you reread the post you linked to (I've seen it before). It does not say what you think it says. Household nominal income growth expectations fell off a cliff and have never recovered.

      Delete
    10. "The S&P 500 rose about 6% between the end of QE2 and the hint of QE3. During this same time the monthly average 30-year fixed mortgage rate went from 4.51 to 3.60 (June 2011 vs. August 2012)."

      The S&P 500 has increased by over 50% since late November 2008. During the 41 months there has been QE it has increased at an average annual rate of nearly 13%. That 6% increase only amounts to a 5% annual rate.

      "Permits for new residential housing units increased by about 30% during this period. The trend growth flattened out during QE3 as rates went up.

      If you plot newly permitted residential development during this period you see that it took off after QE2 while mortgage rates were diving. Once QE3 started, the trend growth for housing permits slowed as mortgage rates went up. It’s quite negatively correlated with mortgage rates during that past couple of years. Since rates went up in June housing starts slowed."

      Just out of curiousity I did Granger causality tests on 1-unit permits, 1-unit construction, nominal 30-year mortgage rates and real 30-year mortgage rates (adjusted by yoy PCEPI) over the period from December 2008 to present. I find no Granger causality in any of these permutations except that I find that nominal 30-year mortgage rates Granger causes 1-unit permits. The impulse response results show a significant negative effect in months 8-19 following an increase in mortgage rates. So this confirms what you are saying but this is unusual. Over longer periods of time no such correlations exist.

      Delete
    11. "Gross fixed capital formation went down during QE1 until long-term rates started falling at which point it started to rise. It went down during QE2 as rates rose, but then started growing when rates began falling precipitously. It grew steadily and significantly between the end of QE2 and the beginning of QE3 while rates were diving. It faltered in the beginning of QE3 as rates went up, but then continued to grow. From the beginning of QE1 until around January 2013 it is negatively correlated with interest rates."

      Both real private fixed investment and longer-term rates fell until 2009Q1. From 2009Q1 to 2010Q1 longer term rates rose. Real private fixed investment continued to fall until 2009Q4. Since then private real fixed investment has risen more or less consistently while longer term rates have done their QE dance. I certainly don't see a correlation and there isn't enough data to do Granger causality tests.

      http://research.stlouisfed.org/fred2/graph/?graph_id=145064&category_id=0

      "Gross private domestic investment was similar. It too went down until rates started falling at the end of QE1 at which point it turns upward until rates started going up at the beginning of QE2 at which point it started to fall. It started going up again when rates started dropping at the end of QE2. In particular it jumped up significantly when QE2 ended while rates fell dramatically. It did, however, start growing even faster during QE3 even while rates went up. But between the beginning of QE1 and the beginning of QE3 it is negatively correlated with interest rates."

      Real private domestic investment fell until 2009Q3 and with the exception of back to back declines in 2010Q4 and 2011Q1 it has risen more or less consistently since. I don't think what happened at that particular time is significant enough to draw the conclusions you're drawing. In any case there isn't enough data to do Granger causality tests.

      I've previously done Granger causality tests on nominal 10-year rates, real 10-year rates, private nonresidential investment as a percent of GDP and private residential investment as a percent of GDP from 1993Q1 to present. What I find is that both kinds of interest rates Granger cause both kinds of investment and that residential investment Granger causes nominal interest rates. Impulse response results show significant effect for both kinds of interest on nonresidential investment, but the effects are positive in both cases which is the opposite of what naive theory predicts.

      Delete
    12. "Japanese government bond yields most certainly came down in the run up to Shinzo’s election and continued to fall. To be fair, they appear to have been falling before it became apparent he would be elected. But they are lower now than when he was elected."

      Abe doesn't directly control monetary policy. There was a lot of spectulation before the BOJ made its announcement but nobody really knew if they were going to do what they did and that's evident by the response of the markets.

      "The stock market dropped more than 4% in the days following the taper discussion and didn’t fully recover until the Fed decided to postpone the taper."

      If you say so, but I don't see much evidence for a negative response.

      "I’m not saying the income expectations channel isn’t there, but it doesn’t appear to have changed the fact that investment and house purchases have been negatively correlated with rates for the past few years and that investment recovered in large part because rates went down (and according to the Michigan survey income expectations have risen)."

      Any negative correlations you think you have found between interest rates and real economic activity are at best ephemeral, and the Michigan survey clearly shows the opposite of what you claim.

      I'm sorry if I come off as dismissive but I really believe you are subject to a lot of misconceptions concerning monetary policy and interest rates.

      Delete
    13. “I'm sorry if I come off as dismissive…”

      Not at all. I’ve found this discussion quite enlightening and I have learned a lot.

      “…but I really believe you are subject to a lot of misconceptions concerning monetary policy and interest rates.”

      Perhaps, but actually I think we are in agreement except on one point. Under normal circumstances everything you’ve said is clear in the data. But since some time in 2009 until around the beginning of QE3, things were different because of severe current income constraints. Investment was negatively correlated with interest rates and housing starts were negatively correlated with mortgage rates. The data you posted from the Michigan survey is also negatively correlated with rates during this period. In almost every instance that rates rose during QE, the Michigan data reflected a drop in confidence; and in almost every instance that rates dropped, confidence rose. Plot it and you'll see what I'm talking about

      In fact if you plot inflation expectations, capital investment, consumer confidence, housing starts, and various LT rates, what you see is that during QE 1 and 2 inflation expectations rose leading to higher nominal rates and subsequent lower investment and lower confidence. When LT rates fall (I believe because of the reduced stock of LT assets), consumer confidence went up, business investment went up, and housing starts went up. I think this reflects a *current-income- constraint* for which higher inflation expectations are not as effective.

      What this suggests to me is that during this period the Fed used QE (at least for 1 and 2) to keep inflation expectations from falling too low. But relied on what we have loosely called the “microeconomic effects” of reducing the stock of LT assets to bring down the term premium on LT rates. This interplay of maintaining expectations and reducing the term premium brought real *and* nominal borrowing rates down significantly.

      All that said, I do acknowledge that the monetary transmission mechanism (I’ll call it inflation expectations) played a much bigger role during this period than I had previously thought. This discussion has forced me to examine the data more closely and to rethink what I thought I knew. Perhaps I was “subject to a lot of misconceptions concerning monetary policy and interest rates” as you said.

      But I have to say that I *still* believe the monetary channel (during this period following the financial crisis) was mostly effective by not allowing inflation expectation to fall too low. And that for current income constrained businesses and consumers it required LT rates to come down via the liquidity premium. In essence I think the Fed has been targeting and real and nominal LT rates.

      Would a more extreme QE have been better? Would a different communication strategy from the Fed have been more effective? Would businesses and consumers have started borrowing more due to rising inflation expectations despite higher nominal rates, or would current income constraints have held back a recovery? I believe the latter, but we'll never really know. Maybe targeting a NGDP futures market would have worked, maybe it wouldn’t have.

      One thing I think is clear: that the Fed’s communication strategy for QE3 was very different than previously and that the results were also different. Given the tax increases, sequester, government shutdown, etc. it is somewhat surprising to see that the economy didn’t tank as a result. I think this can be largely attributed to QE3. Residential development, however, still appears to be constrained by current household income and would likely benefit from lower mortgage rates until AD recovers to some degree.

      Are we still in disagreement or are we on the same page now?

      Delete
    14. This is probably unnecessary but I want to be more clear on the point of current income constraints. Basically, if you can't afford a mortgage now it doesn't matter whether you believe the price level will be higher in the future. The same applies for small business loans given current AD. This is why I think the Fed has been targeting nominal LT rates to some extent. And to be clear, this isn't the norm--as you've pointed out--but I think times were different after Lehman.

      Now to be fair, I didn't enter this discussion having a clearly thought out view like this. You may or may not agree with what I've said, but nevertheless I want to thank you for forcing me to reexamine my own 'taken for granted' views about monetary policy at the ZLB.

      Delete
    15. "Investment was negatively correlated with interest rates and housing starts were negatively correlated with mortgage rates."

      The second statement is supported by the evidence. With respect to the first statement I've run additional tests.

      The investment data we discussed previously is not available in monthly frequency. However, Nondefense Capital Goods Excluding Aircraft Industries (ANXAVS) correlates very well with the equipment portion of private nonresidential investment and Private Construction Spending: Nonresidential (PNRESCONS) correlates very well with the structures component. I ran Granger causality tests on both of these data series and the nominal and real 10-year T-Note over the period since December 2008 and find no signs at all of Granger causality.

      "The data you posted from the Michigan survey is also negatively correlated with rates during this period. In almost every instance that rates rose during QE, the Michigan data reflected a drop in confidence; and in almost every instance that rates dropped, confidence rose. Plot it and you'll see what I'm talking about."

      I ran a Granger causality test using data in monthly frequency from December 2008 to September 2012 and found that household nominal income expectations Granger cause 10-year T-Note yields at the 10% significance level. Furthermore the impulse response results show that household nominal income expectations have a significant *positive* effect on 10-year T-Note yields which is exactly what term structure theory predicts and which is the opposite of what you are claiming.

      "In fact if you plot inflation expectations, capital investment, consumer confidence, housing starts, and various LT rates, what you see is that during QE 1 and 2 inflation expectations rose leading to higher nominal rates and subsequent lower investment and lower confidence."

      I ran Granger causality tests on inflation expectations as estimated with 5-year TIPs with respect to ANXAVS, PNRESCONS, household nominal income expectations, 1-unit permits and nominal and real 10-year T-Notes. The only sign of correlation I found was with ANXAVS. Inflation expectations Granger causes capital goods shipments at the 5% significance level. The impulse response results show a significant *positive* effect. This is consistent with the Unanticipated Price Level Channel of the Monetary Transmission Mechanism, which essentially argues higher inflation improves the balance sheets of firms and leads to increased lending and investment. .

      "When LT rates fall (I believe because of the reduced stock of LT assets), consumer confidence went up, business investment went up, and housing starts went up. I think this reflects a *current-income- constraint* for which higher inflation expectations are not as effective."

      The above results suggest that this is not true for households and that this is the complete opposite of what is true for firms.

      "What this suggests to me is that during this period the Fed used QE (at least for 1 and 2) to keep inflation expectations from falling too low. But relied on what we have loosely called the “microeconomic effects” of reducing the stock of LT assets to bring down the term premium on LT rates. This interplay of maintaining expectations and reducing the term premium brought real *and* nominal borrowing rates down significantly."

      The only sign that lower interest rates may have served as a channel for higher real economic activity has been through residential construction (1-unit permits). All of the other above evidence suggests that the effect of QE on nominal income and wealth expectations, and on inflation expectations, have been far more important than its effects on nominal or real interest rates, and which in fact have been to raise nominal interest rates.

      Delete
    16. "All that said, I do acknowledge that the monetary transmission mechanism (I’ll call it inflation expectations) played a much bigger role during this period than I had previously thought. This discussion has forced me to examine the data more closely and to rethink what I thought I knew. Perhaps I was “subject to a lot of misconceptions concerning monetary policy and interest rates” as you said."

      This discussion has forced me complete even more of the Granger causality tests that I had previously contemplated, and to do some which I had not thought of doing (e.g. on 1-unit permits).

      "But I have to say that I *still* believe the monetary channel (during this period following the financial crisis) was mostly effective by not allowing inflation expectation to fall too low. And that for current income constrained businesses and consumers it required LT rates to come down via the liquidity premium. In essence I think the Fed has been targeting and real and nominal LT rates."

      Again, the evidence shows QE has caused longer-term nominal interest rates to go up. The evidence also shows that the only type of real economic activity affected by interest rates is housing starts, and that correlation suggests that if anything QE may inhibit housing starts. But since the main effects of QE are through financial assets, inflation expectations and the exchange rate, and these are unambiguously positive, I do not think that is anything worth worrying about.

      "Maybe targeting a NGDP futures market would have worked, maybe it wouldn’t have."

      I'm convinced NGDP Level Targeting would be a huge improvement over our current way of doing things for a variety of reasons not just due to the fact we are still recovering from a serious recession.

      "One thing I think is clear: that the Fed’s communication strategy for QE3 was very different than previously and that the results were also different. Given the tax increases, sequester, government shutdown, etc. it is somewhat surprising to see that the economy didn’t tank as a result. I think this can be largely attributed to QE3."

      The CBO's original estimates suggest that payroll employment should have been depressed by about 1.14 million jobs by 2013Q3 by the tax increases and the sequestor. Payroll employment rose 1.7 million between 2012Q4 and 2013Q3, up from the 1.4 million increase in the previous three quarters. This suggests that the fiscal policy tightening was completely offset by QE3 with room to spare.

      "Residential development, however, still appears to be constrained by current household income and would likely benefit from lower mortgage rates until AD recovers to some degree."

      I think the most beneficial thing that could be done for residential investment would be to increase the household sector's nominal income and wealth.

      "Are we still in disagreement or are we on the same page now?"

      Delete
    17. "Are we still in disagreement or are we on the same page now?"

      I still think there may be a huge gulf between us.

      Delete
    18. Nice work. However, I think ANXAVS will be more influenced by exchange rates. So yes, ANXAVS does continue to expand rather steadily from its low point around August 2009. But it continues to grow while LT rates go up and down in large swings. And it continues to grow when LT rates dive down in the latter part of QE2. When inflation expectations drop substantially after QE 1 and 2 end, ANXAVS keeps growing, and it doesn’t appear to have responded to QE3. It leveled off roughly at the pre-recession level and is relatively flat from around March 2012 until now. I don’t think it is impertinent to this discussion, but I also don’t think it tells us much about the points we disagree on.

      It’s great that you are running Granger causality tests but I don’t have a lot of confidence in them for a relatively short duration using monthly data.

      With respect to the investment data we discussed previously, Gross Private Domestic Investment (GPDI) and Gross Fixed Capital Formation (GFCF), we do only have quarterly data. But I still think these are good measures as are 1-unit residential construction permits (though I’ve been including multi-unit permits as well).

      I can’t claim causality, but there is blatant negative correlation with LT rates from about the end of QE 1 to the beginning of QE3. The same is true with the Michigan survey data you posted. It’s unmistakable if you plot them; from the end of QE1 to the beginning of QE3 (or the end of the series) it moves opposite to the 10-year T-Note rate just like the investment data does, and just like housing permits do relative to the 30-fixed rate.

      If you look at 10-year T-Note rates and 10-year expected inflation, what you see is that they move in step (with an additional term premium on the T-Note rate). But toward the end of QE2, and continuing until recently, the non-inflation related term premium dropped way down. Prior to this there was something close to a 130 basis point spread and it dropped to something like a 4 basis point spread. In other words the T-Note rate fell much further than 10-year inflation expectations, so the non-inflation-expectations-related portion of the term premium fell dramatically. Also at this time, mortgage rates dropped along with inflation expectations but continued to fall dramatically well after inflation expectations had leveled off.

      The point is that when this happened, residential construction permits shot up dramatically as did GPDI and GFCF. Consumer confidence went up at this point too. The S&P 500 fell with inflation expectations but then shot up and continued more or less the same trend growth into QE3. ANXAVS also continued to expand during this period until it leveled off near the level it was before the recession.

      So while I can’t claim causality, there is undeniable correlation and this says to me that the Fed was able to drive down the term premium on LT rates while maintaining more or less stable inflation expectations, and that doing this coincided with a strong recovery in investment and particularly residential construction.

      I think the reasons we are able to look at the same data and see different things is that you are approaching it from causality and I am looking at graphs and correlations.

      Delete
    19. I've been trying to find a way to create a link to some graphs but am having technical difficulties. I hope you don't mind I'm sending you an email with some graphs and an Excel spreadsheet with the data I've been using.

      Delete
    20. Sure, send me an email.

      Delete
    21. Mark,

      My contact info is here [http://synthenomics.blogspot.com/p/about-me.html]. I can't actually find your email.

      Delete
    22. Yichuan,
      That comment was directed at Mark Summers. I can be reached at msadowski01@comcast.net.

      Delete
    23. I sent it to you both. I'll keep trying to find a way to post a link on the off chance anyone else is paying attention to this conversation.

      Delete
  17. Hello Everybody,

    My name is Mrs. Monica Roland. I live in UK
    London and i am a happy woman today? and i
    told my self that any lender that rescue my
    family from our poor situation, i will refer
    any person that is looking for loan to him,
    he gave me happiness to me and my family, i
    was in need of a loan of $250,000.00 to
    start my life all over as i am a single
    mother with 3 kids I met this honest and GOD
    fearing man loan lender that help me with a
    loan of $250,000.00 U.S. Dollar, he is a GOD
    fearing man, if you are in need of loan and
    you will pay back the loan please contact
    him tell him that is Mrs. Monica Roland that
    refer you to him. contact Mr. Anthony Hampton
    via email: (easyloans03@gmail.com)

    ReplyDelete
  18. Venezuela looks like the model for Abenomics.

    http://www.ft.com/intl/cms/s/0/b41fbcb4-d0f1-11e2-be7b-00144feab7de.html

    ReplyDelete