Believe those who are seeking the truth. Doubt those who find it. Andre Gide


Wednesday, August 31, 2011

Krugman on Gross

We have to give Bill Gross some credit for admitting his mistaken call on the U.S. Treasury market; see here.

Gross's call was based largely on the fact that the Fed's Treasury purchase program was about to expire. The idea is that with a large component of the demand for Treasury debt removed from the market, bond prices should fall once the program is terminated.

This is what you get when you apply the static Marshallian scissors toward understanding an economic phenomenon. It often works well. But not always. And in any case, we know that forecasting the path of security prices is a tricky business. This is what the efficient markets hypothesis tells us.

Gross's intuition was shared by a few of the macroeconomists I bumped into earlier this year at a macro conference in Konstanz. I remember one of my Fed colleagues and myself suggesting to a sceptical audience that one should not expect any dramatic bond price movement once the Fed's purchases ended, largely because the market had already discounted that event. I don't think we were very successful at persuading people. Well, we turned out to be right about that (maybe we got lucky). [Actually, I did give my view of Gross' predictions here in March 2011].

Subsequently, of course, the price of US Treasuries have risen substantially. We did not make that prediction (nor did we try). But evidently, Paul Krugman seems to think he did; see Who You Gonna Bet On?

Yep. All we need is the Econ 101 macro model to understand this crisis and predict bond prices. (Oh, and gold prices have shot up too...where do those appear in that model?)

Do we really need an IS-LM model to tell us that when two assets with similar risk properties and similar returns are swapped in the Fed's portfolio that nothing much is likely to happen? (And as I explain here, standard monetary models deliver "flight to quality" phenomena quite easily.)

Krugman's article gives the impression that simple IS-LM analysis could have predicted the recent fall in Treasury yields. Do people actually find this credible?

39 comments:

  1. I do.

    Low GDP + liquidity trap = quantity of bonds wont affect interest rates.

    Indeed, financial instability means investors will keep their bonds expecting a higher interest rate. Meanwhile the high public savings balances it out, ergo interest rates are unaffected.

    - CL

    ReplyDelete
  2. I think Krugman's argument is that IS-LM, whatever you think about it, has been a rather useful model that both explains and predicted what's going on, and should give pause to those who are re-jigging their own models ex post.

    ReplyDelete
  3. @Anonymous: Your comment would make sense except for the fact that interest rates fell. Please re-think your position, or set me straight.

    @jesse: Please tell me how the IS-LM model predicted the recent fall in bond yields. It did no such thing.

    ReplyDelete
  4. @David: Perhaps it might have something to do with the euro & dollar stability the past months. Although I'm not sure Mundell–Fleming model is applicable in this case. - CL

    ReplyDelete
  5. Presumably Krugman will be retiring from the NYT to start his own hedge fund.

    ReplyDelete
  6. David,

    Okay, I will take a stab at your question. Because expectations of future output and wealth have declined more (due to ongoing bad economic news), the IS curve has shifted left. This should lower the real interest rate.

    On the other hand, the LM curve should have shifted left too as the increased uncertainty should have raised money demand. This should raise real interest rates.

    The story must be, then, that the IS curve shifted farther left than the LM curve. (Maybe the increased money demand leads to even lower expectations of future output and further shifts the IS curve.) In other words, the pressure on interest rates from the goods market dominates the pressure from the asset market in the Krugman story.

    And if we want to go all out, throw in the Full Employment (FE) line that the Abel, Bernanke, and Croushore text use and you could have the natural interest rate (the intersection of the IS and FE)being lower than the prevailing interest rate (intersection of LM and IS). In this case, actual interest rates are too high relative to the natural rate and thus are further dragging down the economy. This too would then provide another reason to further lower output expectations. (See the first figure in Nick Rowe's post here for a figure that show this: http://worthwhile.typepad.com/worthwhile_canadian_initi/2011/04/reverse-engineering-the-mmt-model/comments/page/2/)

    One problem with the above explanation is that the interest rates that fell were long-term while the IS-LM just has one interest rate that usually is viewed as a short-term one, no room for the term structure. One has to hand wave between short-term interest rates and long-term interest rates.

    ReplyDelete
  7. Reading the Krugman links that you give, I don't see him claiming to predict a fall in interest rates; he claims to have correctly predicted that they would not increase sharply at the end of QE2. (Perhaps you have a different Krugman citation in mind that you failed to give.) I also see him claiming that this is consistent with standard Hicksian IS-LM where the economy is depressed.

    I missed him claiming that he can predict gold prices; do you think that is important for a macro model?

    My reading of your argument is that "predicting security prices is a tricky business." I'm sure your preferred macro models incorporate this view. Krugman here seems to saying that, on this point and recently, his models perform as well as yours on that score. I'm glad that you and he were both trying hard to convince the "crowding out" alarmists that a spike in interest rates was unlikely. Thanks for fighting the good fight, David.

    ReplyDelete
  8. Simon,

    I guess there are two forecasts here. One is that bond prices would not fall; and one is that bond prices would rise.

    PK turned out to be correct (like many dark age economists) that the termination of the bond purchase program would not lead bond prices to fall. Heck, simple expectations theory delivers this, depressed economy or not.

    But Gross was also apologizing for failing to forecast the rise in bond prices. This apology was what motivated PK's column. But PK did not predict the rise in bond prices either. My reading of his article is that he suggests that Gross would have done well to organize his investment portfolio on the IS-LM model.

    You say that you also see him claiming that "this is consistent with standard IS-LM.."

    Seriously? Do you really make modest claims of this sort? Read carefully. He is saying that IS-LM is THE model; and that people like me are living in the Dark Ages.

    As for your comment on gold prices, no, he made no claims. But my point would be the following.

    Gold prices and bond prices have recently moved in the same direction. This seems rather puzzling and important (see my previous Wonderland post). Seems like we, as a profession, might want to try to figure things out. And the IS-LM model is just not going to cut it. And PK's promotion of that model as "all we really need" is really rather misleading.

    Anyway, just my 2 cents worth!

    ReplyDelete
  9. @David Beckworth,

    Well, my point was not so much that one could not use that theory to provide an interpretation of events. I was more criticizing the idea that the theory (and that theory only) could have been used by Gross to predict events.

    We are in the business of making conditional forecasts. PK should not mislead the public into thinking otherwise.

    ReplyDelete
  10. David, my interpretation of the Business Week article is that Krugman was predicting lower output growth rates as government spending declined as the stimulus funds dissipated. Given that he has consistently argued that bond yields are primarily tracking market expectations of future output conditions (and hence future interest rates), it's easy to see how he would have forecasted falling rates. Whether he did or not I'm not sure, but the bigger point is that his conceptual framework would, annoyingly, again have been right. So yes, had Gross used this theory to predict future interest rate movements and bond prices, he would have gotten it right.

    - C

    ReplyDelete
  11. I've been following Prof. Krugman's comments about Bill Gross's bond predictions for a while and most of his comments revolve around how Gross's predictions of higher bond yields (vigilantes) are unlikely (using mostly Hicks/IS-LM models, but not exclusively). And the case he makes seems fairly strong. Interest rates are low and seem likely to remain low until employment rises, raising demand and increasing output to meet rising demand. Inflation is low and seems likely to remain low until employment rises, raising demand and putting upward pressure on prices.

    ReplyDelete
  12. David;

    You're better at language than that; predicting "not a rise" is not the same as predicting a fall. Hating PK's point of view doesn't justify distorting his words. (Do you think you need to in order to make your point?) You and I agree that many models were delivering that forecast (including models that you agree with.)

    But Gross did not go long on bonds, or neutral on bonds; he went short. PK says that was dumb. I assume that you would have (at the time) agreed. I think it is reasonable for PK (if he wants) to say "told you so." If you feel the need to, go right ahead.

    Sorry, I don't think gold price movements are important and I don't think that they have been for a very long time. Remind me, how many macro models used by central banks have gold prices as an important variable? Which applied forecasting models do? How much time did you spend on gold price determination when you taught macro? Seriously, what am I missing here?

    I agree that Krugman finds IS-LM very useful. If you read him, you'll know that he does *not* think it is all that you need. Seriously, can you come up with a citation where he really says that?

    I agree that he clearly thinks (or would if he thought about you) that you are an exemplar of a "Dark Age" economist (for lack of a better term.) Are you surprised that macroeconomists do not agree? Do you think that you should have not have to suffer criticism from people of his caliber?

    ReplyDelete
  13. Simon,

    Perhaps this post was not one of my best efforts. You do not have to apologize for your views on gold or other asset prices. I do not think I can come up with a citation where he says IS-LM is all you really need. It is not an issue of macroeconomists disagreeing. I disagree with Steve Williamson all the time, for example. But I respect Steve because he does not bullshit.

    ReplyDelete
  14. dog waste bag The stylish dog poop waste bags that make cleaning up after your dog a little bit better. Buy Biodegradable dog waste bags on wholesale price.

    ReplyDelete
  15. Normally, I remove posts like the previous one. But as I mentioned to Simon, not one of my better efforts; so it seems apropos.

    ReplyDelete
  16. "But Gross did not go long on bonds, or neutral on bonds; he went short. PK says that was dumb. I assume that you would have (at the time) agreed."

    No no no. Gross sold out of his long position in US Treasuries. He did not go short treasuries, but as he points out he was short of treasuries come the rally. There is a difference.

    Furthermore, he moved into foreign government debt, probably Canadian gov bonds, Brazilian gov bonds, and others. And there were large rallies in all these issues coincident with the US treasury rally. He also probably upped his corporate debt exposure, which rallied but less so than treasuries.

    Simon, you're dodging David's point about the coincidental rise of gold and bonds. I for one would like to know how economic models can (or can't) explain this.

    I do find it amusing to see Krugman gloating about his predicting ability. The only evidence of good prediction is a steadily rising account balance, and until he shows me how his futures account has done over the last few years, I'm not going to give him any marks. Keynes for one could put up the green.

    ReplyDelete
  17. JP Koning: My apologies -- given the Gross quotations offered by PK, I assumed that he had gone short. I should have gone to the source.

    You're right that I'm dodging the point about the coincidental rise of gold and bond prices. For now, that's all that it seems to be: coincidental. Frankly, most economists haven't worried much about the end of Breton Woods, when modern currencies moved off anything resembling a gold standard. Neither David nor I can come up so far with a good reason why macroeconomists should have been worrying more about modeling gold prices. You?

    I don't think PK is interested in running a hedge fund; if you're looking for investment advice, I think David and I and PK would all agree that you should look elsewhere.

    "The only evidence of good prediction is a steadily rising account balance...." Think about that, JP. Didn't Madoff have a steadily rising account balance? More seriously, read Andy Lo's work on hedge funds, particularly his "Capital Decimation Partners" examples. You might be surprised.

    ReplyDelete
  18. (So now I've followed JP's link back to his home page, seen his gold chart, and realized I've probably waved a red flag in front of a gold/monetarist bull by asking why we should care about gold prices. Not what I'd intended, but should be interesting nonetheless.)

    ReplyDelete
  19. @Simon:

    You're right that I'm dodging the point about the coincidental rise of gold and bond prices. For now, that's all that it seems to be: coincidental. Frankly, most economists haven't worried much about the end of Breton Woods, when modern currencies moved off anything resembling a gold standard. Neither David nor I can come up so far with a good reason why macroeconomists should have been worrying more about modeling gold prices.

    I think you jumped the gun a bit by including me in that last sentence.

    I think that you are correct to question why a particular economic variable deserves interest from the perspective of macroeconomic theory.

    My perspective is from the view of someone who believes that money and banking (broadly defined) plays an important, possibly independent, role in business cycles.

    Gold (and other commodities) has traditionally played an important role in the monetary system. Today, gold is not money in the way it once was. But gold is still an important store of value (a form of capital, and capital is something that macroeconomists do worry about).

    Moreover, even gold is not now money, it may have implications for the conduct of monetary policy. This is because even the threat of using gold as money potentially has real and contemporaneous consequences. A competing currency is something that central bankers should worry about.

    You want to attribute the recent price rise in gold and bond prices as a coincidence. It may very well be that. But it may not. And if it is not, then that is something interesting. But we won't know unless we pursue this line of enquiry.

    And that is why I think macroeconomists, or monetary theorists at least, should pay attention to gold prices.

    ReplyDelete
  20. ....sorry, that should have been

    "...haven't worried much about gold prices since the end of Breton Woods..."

    ReplyDelete
  21. My apologies, David. I should have not spoken for your opinion on Gold.

    ReplyDelete
  22. Simon, no need to apologize! I enjoy the challenges you bring forth here. It's a question that many of us who focus on this or that commodity price, or on this or that nominal interest rate, should be asked to defend more frequently.

    ReplyDelete
  23. David: "But gold is still an important store of value."

    Could you please back up that assertion with facts? I'd be curious to know what fraction of total human wealth is represented by gold. How does it compare, for example to the value of all publicly traded shares? bonds? real estate? Derivatives? While I think your assertion about the store of value is a statement about the value of the stocks, it might also be interesting to know something about the volume of transactions. How does the liquidity in the gold market compare to that of global trading volumes in stocks? bonds? currency?

    David, the rest of your argument is entirely based on "it may", "potentially", "may not be coincidence." There's a looooong list of things that can qualify on that basis (perhaps including Hicksian IS-LM.) So I'm asking, again, is that the best case that anyone here can make? I'm obviously skeptical, but I'm listening and can be convinced.

    P.S. I've apologized for what I said about your opinion on Gold. Let me briefly explain why I jumped the gun. In my first and second comments, I asked for justification of why this was an important subject. You did not reply to that part of my comment. So I wrote "Neither David nor I can come up so far ...etc." While I think this was strictly correct, I'm glad that you clarified where you stand.

    ReplyDelete
  24. Simon, I'm neither a gold bug nor a monetarist so I'm thinking, why the hell is that guy over there waving a red flag at me? I make my charts because I'm interested in how historic ideas and events shape the present.

    You say that the macro models used by central banks don't have gold prices as an important variable. Surely that can't be entirely correct, since central banks have been the biggest buyers of gold over the last three years. Ipso facto, their models must include gold. Now maybe their models are wrong, but you need a model to beat their model, and you say you don't have one. Note that central banks were net gold buyers last year for the first time since the late 1960s.

    Not only might central bankers like models with gold, so might money managers.

    The whole point of the Gross/Krugman debate seems to be that there exists some off-the-shelf model that Gross should have used to better manage his clients' money. As someone in the industry, bring it on! I'd sure like an off-the-shelf model to help me understand rising bond and gold prices. With the world's gold stock worth about $11 trillion, more than the US national debt, it's becoming increasingly important to include gold in investing models. I'm not being facetious here. I am genuinely curious what economic models have to say.

    ReplyDelete
  25. JP:

    I don't know of any central bank whose macro model includes gold prices. I don't think the IMF or OECD models include them either. Please correct me if you can cite a specific exception.
    Note that the macro models are usually designed to give macro policy advice. Central banks might buy or sell gold reserves for other reasons, but I was asking whether macroeconomists should worry more about gold prices and I think the macro models usually reflect whatever they think is important.

    You're right that portfolio managers might want to model gold prices. or copper. or spring wheat, or orange juice, or any other commodity with a liquid derivatives market. I think Krugman has been saying repeatedly that those fund managers who believed that either (a) big US deficits or (b) the expansion in the fed's balance sheet would cause bond prices to fall lost their clients a lot of money. Yet those arguments were frequently used in the political arena to stymie the kind of stimulative macroeconomic policy Krugman has repeatedly called for over the past year or two. I think his line of argument concerning Gross is designed to appeal to people in the industry who think that "the only evidence of good prediction is a steadily rising bank balance."

    In a nutshell, if you put stock in the WSJ's bond vigilantes, you're probably regretting it now. David pointed out that not just the IS-LM, but several other models also indicated that the WSJ's editorial writers were full of . As were (are?) many other political actors with similar advice. David says he was pointing out at the time that the "bond vigilante" view looked unreasonable. I think that was an important thing to do.

    I'm also confused by what you wrote; that central banks were net buyers of gold last year for the first time in long while, yet they were the biggest buyers over the last three years? I'm not sure that you meant both of those statements. I'd also be curious to know how much of that "central bank" gold purchasing just reflects the management of China's foreign exchange reserves. What exactly is a reliable source of statistics for who is buying gold?

    ReplyDelete
  26. You can get central bank gold reserve information from the World Gold Council and the IMF. Western central banks (North American and Europe) have been big sellers for the last 25 years. About three years ago eastern central banks started to buy in large amounts. Last year western central banks stopped selling, the net effect being the first increase in world central bank gold holdings since 1960 something. Eastern central bank purchases come from a very wide selection of banks, not just China.

    "You're right that portfolio managers might want to model gold prices. or copper. or spring wheat, or orange juice, or any other commodity with a liquid derivatives market."

    No, I was pretty specific about that. Bonds and gold are moving together, not commodities like copper. Why?

    "Please correct me if you can cite a specific exception."

    All I know is that many central bankers now hold certain ideas that are leading them to accumulate gold. The composition of a central bank's balance sheet surely falls under the rubric of macro policy, no?

    ReplyDelete
  27. You can get central bank gold reserve information from the World Gold Council and the IMF. Western central banks (North American and Europe) have been big sellers for the last 25 years. About three years ago eastern central banks started to buy in large amounts. Last year western central banks stopped selling, the net effect being the first increase in world central bank gold holdings since 1960 something. Eastern central bank purchases come from a very wide selection of banks, not just China.

    "You're right that portfolio managers might want to model gold prices. or copper. or spring wheat, or orange juice, or any other commodity with a liquid derivatives market."

    No, I was pretty specific about that. Bonds and gold are moving together, not commodities like copper. Why?

    "Please correct me if you can cite a specific exception."

    All I know is that many central bankers now hold certain ideas that are leading them to accumulate gold. The composition of a central bank's balance sheet surely falls under the rubric of macro policy, no?

    ReplyDelete
  28. JP: I was aware of the IMF data on central bank reserves and their compositions. (Just curious: does the WGC just use their stats or do they have their own sources?) But I also asked about figures on who was buying gold, largely because of your claim that central banks were "the biggest buyers of gold." Given the decentralized nature of the gold market (and the, um, discretion that many private participants may desire), I didn't think any reliable statistics on the overall market existed.

    Yes, you were pretty specific about what you were interested in, but I still don't really know why, beyond the notion that portfolio managers may be interested in any type of investment. As for comovement between gold and bond prices, the fact that this doesn't look like a stable long-term relationship makes many think "coincidence" is a good-enough explanation.

    As for whether the gold accumulation by central banks is considered macro, I suspect many macro economists don't really think so. Correct me if I'm wrong, but aren't most of those Eastern central banks who are accumulating gold trying to resist currency appreciation against a weak USD? And doesn't that force them to buy USD and sell their own currency? And hasn't that caused them in recent years to accumulate foreign exchange reserves in quantities such as the world has never before seen? (Okay....literally trillions of USD.) I'll agree that that behaviour is something macroeconomists do (or should) think about.

    But in that context, buying gold is an afterthought. The central banks are faced with the problem of managing a huge war chest, which starts out in USD. And then the interest rates on the USD fall and the USD looks set to weaken against other currencies (which is what they were trying to prevent in the first place.) Which gives the foreign exchange reserve fund capital losses in the tens or hundreds of billions of dollars. (Think; if you're holding $2 Trillion of USD denominated assets and the dollar depreciates by 10%..... Which is just one reason why Chinese officials have periodically called for the US Treasury to do more to defend the dollar.)

    So Asian central banks swap USD for gold to hedge some of the FX risk in holding dollars. That's not really a macro decision so much as a portfolio management decision. I don't think macroeconomists really care about the difference between holding foreign currency or gold as part of the central bank's balance sheet. That's not to say that it doesn't affect the value of the USD (and therefore it has macroeconomic consequences.) But I think macroeconomists (a) too often ignore the impact of the exchange rate on the US economy, and (b) despair at trying to model the decisions of communist party officials responsible for managing multi-trillion dollar financial portfolios.

    Or do you think that something else is behind the central bank movements towards gold?

    ReplyDelete
  29. "Given the decentralized nature of the gold market (and the, um, discretion that many private participants may desire), I didn't think any reliable statistics on the overall market existed."

    ETF data can give you a feel for private buying. You can roughly build up central bank information from IMF data. Obviously if you are close to the market you can get better information.

    "As for comovement between gold and bond prices, the fact that this doesn't look like a stable long-term relationship makes many think "coincidence" is a good-enough explanation."

    Fair enough, I'll have to just shop around for some sort of model that can give me an answer.

    "Or do you think that something else is behind the central bank movements towards gold?"

    Those are probably some of the reasons. But it can't be purely about offloading USD exposure, since central banks could be buying euro, yen, or other denominated assets but are instead choosing gold. Unlike ten years ago, gold markets are now very big and very liquid. Liquidity attracts liquidity. There are many political reasons too.

    ReplyDelete
  30. JP: "But it can't be purely about offloading USD..."

    Well, you work in the industry, so what advice would you give to a client who has a position or tens or hundreds of billions that they want to invest and yet keep as liquid as possible. I'm just an academic; I don't think that there are many markets globally with liquidity on that scale. But go ahead and make your list.

    Now cross USD-denominated assets off the list, because you're worried about the bearish prospects for the USD exchange rate. Then cross the EUR-denominated assets off the list, because
    of the ongoing nature of the Euro-zone govt-debt/banking crisis and its exchange rate impact. Then cross the JPY off your list, because your second-biggest client is the Bank of Japan and the whole point of their FX intervention is to sell (not buy) yen. What's left?

    Of course, I'm oversimplifying. But all of these give reasons to diversify away from these three very liquid asset classes. Everything next on the list (CHF, CAD, AUD, XAU) benefits as a result.

    ReplyDelete
  31. Which is one of the reasons why the simultaneous rise in gold and bonds are so odd. If everyone is trying to get out of bunds, treasuries, and Japanese government bonds and into gold, then why are the prices of all these assets rising together?

    ReplyDelete
  32. Simon: "You're right that I'm dodging the point about the coincidental rise of gold and bond prices. For now, that's all that it seems to be: coincidental."

    Krugman isn't dodging; he seems to think it's no coincidence at all.

    http://krugman.blogs.nytimes.com/2011/09/06/treasuries-tips-and-gold-wonkish/#more-24069

    ReplyDelete
  33. JP: You're right that it's a good example of a macroeconomist thinking about the relationships between bond prices and gold prices (and I just finished arguing that they don't. Doh!)

    I don't think Krugman is ready to say that this view is the right one, however. Read his penultimate paragraph

    "OK, none of this necessarily rejects other hypotheses about gold; in particular, there could be a bubble over and above the Hotelling aspect. But the crucial message is, I think, right: If you believe that gold prices are signaling an inflationary threat, I have to tell you, I do not think that price means what you think it means."

    ReplyDelete
  34. This really great read i thinks so.i am blogger and expat SEO.The web site really needed for every IT user.Please correct me if you can cite a specific exception.Thanks for info.

    ReplyDelete
  35. Thanks .. this is a really great post and and none of this necessarily rejects other hypotheses about gold

    ReplyDelete
  36. thanks very much.. very informative post.

    ReplyDelete
  37. The time has only for live stream sports hd online ! We always try our best to get you the best possible HD streams to watch live sports online.

    ReplyDelete
  38. Free sex videos free porn aways way. Home you teen sex pornhub hd free videos. Smart porn tv free sex watch nowe !

    ReplyDelete
  39. I remember one of my Fed colleagues and myself suggesting to a sceptical audience that one should not expect any dramatic bond price movement once the Fed's purchases ended, largely because the market had already discounted that event. China outsourcing

    ReplyDelete