Is quantitative easing trying to raise or lower interest rates?

The Bank of England has switched to quantitative easing. It is buying long bonds (gilts). What would count as a signal of success? We could argue that falling yields would signal success, because it is trying to reduce long interest rates to stimulate investment. But we could equally argue that rising yields would signal success, because higher expected growth in real output and inflation would increase long interest rates.

What does that ambiguity imply about thinking of interest rates as the monetary policy instrument or indicator? What does it imply about the likely success of that form of quantitative easing?

David Altig, at the Atlanta Fed, raised the same questions back in November. He pointed to the empirical evidence linking a steep yield curve with expanding balance sheets and asked:

"So, if stimulative monetary policy is what we are after, should we be looking for lower long-term rates or higher long-term rates? Discuss."

It's an old monetarist observation, going back to the 1970's (at least), that high nominal interest rates could be seen as a sign of high expected inflation, and therefore loose monetary policy.

We forgot that ambiguity during the Neo-Wicksellian years of inflation-targeting via a short term interest rate instrument. We reasoned that if the central bank wanted to loosen monetary policy it should lower its very short interest rate instrument. This would cause longer rates to fall as well, and increase aggregate demand and future real growth and inflation.

While expectations of future inflation and real growth remained well-anchored, and thus independent of current monetary policy, it was probably OK for us to ignore that ambiguity. And moving short term interest rates in the "right" direction seemed to work anyway. (Though Greenspan's "conundrum" of falling long rates despite rising short rates might suggest otherwise.)

It is not OK to ignore that ambiguity now, when expectations of future inflation and growth are not well-anchored. And when short term interest rates are zero, and central banks consider operating on longer rates, we need to understand this better.

Suppose we could wave a magic wand, and restore confidence. Investment demand rises, consumption demand rises, expected inflation rises (people stop expecting deflation). The central bank would need to raise the overnight rate in response, and would be expected to keep it raised. Bond prices would fall, and yields would rise.

Now suppose the central bank bought bonds secretly, and at the same time someone waved the magic wand. What would happen to bond yields? The secret purchase might lower bond yields, but the magic wand would raise them. If the magic wand had a bigger effect than the secret bond purchase, the new demand for bonds by the central bank would be met by an even bigger supply of bonds, and the price of bonds would fall, and yields would rise.

Now suppose the central bank buys bonds openly, and the quantitative easing policy is expected to be permanent and expected to work. The effects should be exactly the same as the secret purchase plus the magic wand. The effects of the purchase itself are the same as a secret purchase. The effects on expectations of the future are the same as the effects of the magic wand.

Future monetary policy affects things today only via its effect on today's expectations of the future. When short nominal interest rates are zero, I believe that a permanent increase in the money supply has a much bigger effect than a temporary increase. Expectations of the future money supply matters more than today's money supply. If I am right, this means that the effects of the magic wand are bigger than the effects of a secret purchase of bonds.

And this means that a successful policy of quantitative easing will cause nominal interest rates to rise. If the policy lever is connected to the thing you want it to be connected to, then trying to push the lever down will cause the lever to rise.

Being fond of automotive analogies, I am trying to imagine what it would be like driving a car that worked like that. It would be like having the power-steering connected the wrong way round. I try to turn the wheel left, and the power steering turns it to the right. Even if I remembered which way to turn the wheel, it would be hard to turn it the right amount. And my passengers wouldn't know what I was trying to do, and might not have great confidence that I was turning it the right way. And if my passengers actually were the power-steering mechanism...?

It might be better to find another policy lever, instrument, or intermediate target. Find one where the power-assist moved in the same direction you were trying to move the lever. Where the secret policy, and the magic wand, would move the lever in the same direction.

Monetary medicines work best when we believe they work. But for us to believe in them, we must find a way of administering the medicine where the medicine and the belief that it works move the lever in the same direction. Push the lever and move it a little. Seeing the lever move a little bit in the direction you are trying to move it will restore a little bit of confidence, and the lever will move a little bit more, which will restore a little bit more confidence, and so on.

If we administer the monetary medicine in a way that the medicine, and the belief that it works, cause the lever to move in opposite directions, we lose this cumulative process of restored confidence. Buying bonds is a bad way to administer the monetary medicine, just because it does move the lever in the opposite direction to the belief that the medicine works. People see bond prices rise, and think that other people think the policy won't work. And if other people think it won't work, it will be less likely to work.

If the central bank secretly bought stocks, stock prices would rise. If waving a magic wand restored confidence, stock prices would rise. The medicine, and the belief it will work, move stock prices in the same direction. A cumulative process would restore confidence. Stock prices would be a good lever to pull. Central banks should buy stocks, if they want to do quantitative easing.

Would buying commercial paper work? Maybe. If the magic wand increased the liquidity and reduced the risk of commercial paper, the spreads would fall. If the effect of falling spreads were stronger than the effect of rising interest rates on safe liquid bonds, the magic wand would move the lever in the same direction as the secret policy. If not, and they move the lever in opposite directions, the magic wand won't work.

This post is my reflection on Scott Sumner's latest post.

Comments

[Dee]

Don't you have history on your side Nick? Didn't the Brits successfully sell gilts back in the 60s'?

I think confidence is the key but which ideology is correct??

http://business.timesonline.co.uk/tol/business/economics/article5908417.ece

If people are followers of the free market model then this piece of news should please.

They won't have too much left to sell will they? Like watching an bankruptcy auction of a farm that's been in the family for generations.

[JimP]

I wish you were right about stocks - but personally I think that if the govt bought them it would scare people silly.

And I am not kidding. I really do wish the stock plan would work. Sumner has convinced me of the need for a permanent increase in the money supply - and I would love it if we could do that through stocks. But I don't think we can.

What other countries have bought stocks? And did it work? I really do not know and would like to.

Thanks

[David Pearson]

Interesting piece. There is a similar conundrum for the Fed's gold holdings. If it sold its gold, it would force down the price; however, the currency would have less gold backing. Assume that gold backing has some effect on future expectations (a big assumption, granted, but then, why hold gold at all if it didn't?), then the gold price should RISE as the Fed sold gold.

Perhaps the Fed has already considered your argument. They have chosen to target not stock prices but credit spreads. The idea is that as long as (mostly short term) credit spreads behave, then the tail risk of deflation is low. As you say, they can pull the lever a little, watch it work, and then decide whether to pull more. This explains the behavior of the Fed's balance sheet over the past two months. They pulled the lever on commercial paper; spreads came in. Then they pulled back a little. They pulled it on MBS; spreads crashed, and then they decided not to buy at a pace that meets their $500b target (why bother? it worked already). Meanwhile, the overall balance sheet -- even ex-currency swaps -- has been shrinking in recognition of the broad "success" of the policy.

So the Fed is telling us they are worried only about deflation tail risk, and they have that under control. This implies that the economy, saved from a deflationary crash, will right itself on its own and eventually produce inflation expectations. Again, at least on the expectations front, this is consistent with the behavior of the gold price and TIPS over the past two months.

I don't agree with credit spread targeting, but over the past two or three months you can make the argument that it has worked. What is your counter argument?

[Scott Sumner]

Nick, This is an interesting look at the bizarre psychology involved in trying to make bond prices fall by buying bonds. I would suggest that people interested in this issue look at what happened to the bond market immediately after three Fed decisions that came as the economy was teetering on the edge of recession. Jan. 2001, Sept. 2007, and Dec. 2007. All three decisions moved long yields the "wrong way." The first two were expansionary surprises and long yields rose, the last was a contractionary surprise, and long yields fell. But in the last case even three month yields fell (which I don't recall ever seeing before.) But the 3 month market was right in retrospect, because the economy tanked so rapidly after the announcement that the Fed did another 125 basis points of cuts in a series of panicky moves just a few weeks later. Also note that in all three cases the stock market went the "right way" (and very strongly.) My prediction is that no matter what the Fed buys, if they have an expansionary enough and credible enough policy stance to send the stock market soaring, long yields will almost certainly rise. And I think a truly dramatic and credible initiative could cause an extraordinary increase in equity prices, despite the liquidity trap-types who say money no longer matters.
Having said all that, your point about psychology may be very important, and something the Fed should think about as it decides which assets would give the policy the most credibility. We are in such uncharted territory that I simply don't know how accurate the steering analogy is, but it might be very important.

[Scott Sumner]

Nick, I forgot to mention that your insight into the success of interest rate targeting seems right to me. In a paper I published a few years ago I argued that interest rate targeting did pretty well in the post-1982 period, because inflation expectations were fairly stable (i.e policy was credible.) It is interesting that whenever policy is not credible (the 1970s or today), interest rate targeting doesn't work very well, as the Fisher effect may overwhelm the liquidity effect.

[bob]

Interesting discussion. I definitely fall more into the Krugman/Keynesian side of things, but I can see how unconventional monetary policies could work, and I'm not ruling out anything at this point.

I've been working through Scott's blog from the first posts up to now. I'm still digesting a lot of your finer points, so I may have gotten some of your arguments wrong.

I think you underestimate how hard it is to break the "expectations trap". I haven't read your papers on the topic, but a policy of buying gdp futures strikes me as being very risky for some of the same reasons that Bill Woolsly brings up in a comment on your blog. You wouldn't be buying any underlying assets, just establishing zero-sum contracts with counterparties. (I think Nick's policy of buying indexes would be better, because the money is passed on and invested in the underlying assets). Also, private futures buyers have access to huge leverage and will sell in to your buying, if they are still looking at the underlying fundamentals. I can see why futures would be attractive if you are trying to target expectations directly, but I think you underestimate the push-back from the market. I could easily see something like Soros vs. BoE happening. If the central bank isn't successful in moving the futures, they stand to lose a large amount of money, for 0 benefit.

It's actually alleged that the government does already buy futures for the reasons you advocated (and keeps it secret for the same reasons):
http://www.sprott.com/pdf/TheVisibleHand.pdf

tbh, I actually don't doubt the existence of the "PPT", but I don't see it as especially powerful, for the reasons I outlined above. The futures market is still small relative to the overall market, and most futures traders are going to be looking at the trajectory of the underlying asset. If you start buying futures that are grossly out of line with this, they will sell a lot of futures. If you offer to write contracts the look likely to be unprofitable, a lot of people will be lined up ready to take the other side, and you will have to sign a ton of contracts to move the price up. I think it only works if a bear market slide has lost most of its momentum, and short-sellers are getting nervous. I suspect that if there is a PPT, they have learned this the hard way by losing a great deal of money buying futures into a strong bear market.

I think this gets back to a more central point: monetarist policies are attractive because they target financial assets that provide the policies with leverage. A relatively small purchase of treasuries, stocks, or futures sends price signals that are amplified and effect the underlying economic activity by changing expectations of economically important actors. The problem, IMO is that you are targeting the most "rational" and resourceful actors that will tend to neutralize your initiatives in the first instance, due to the expectations/liquidity trap. If Goldman Sachs traders are looking at guidelines telling them that the necessary interest rate is -6% , it's going to be difficult to convince them that there is a threat of inflation or that GDP futures should by higher. They have strong expectations of the opposite, and a large amount of money to neutralize your bets with. Futures buying would have to be truly spectacular in order to overwhelm the clear implications of soaring unemployment, falling prices etc. that inform the counter-parties' trading, and the motivate the avalanche of selling that would meet your higher bids.

Where you say that Keynesians get the causation backwards, I don't think that is true. They see causation going both ways, in a negative feedback loop, whereas you only seem to see the expectations half of it. The causation is running both ways between deteriorating fundamentals and deteriorating expectations. That's why I think monetary policies that only really target expectations are weak. The problem is the feedback loop, and I think you need to address both sides of it simultaneously. You can only change expectations if you change underlying economic activity, and you can only change the underlying economic activity if you change expectations. IMO, fiscal policy does both, while traditional monetary policy and OMO only really target expectations.

Now if you get into real helicopter drops to poor people, I think that a lot of these points no longer hold, but it doesn't seem as though the monetary policy being described here are really radical enough to overcome the same expectations trap amongst financial intermediaries that has rendered traditional monetary policy useless.

[Winslow R.]

I'll just add all these purchases of longer term securities by the government removes future aggregate demand as the interest payments that used to recycle within the private sector now are directed into public coffers.

'New' monetary policy easing will be deflationary unless it becomes incredibly reckless. Fed profitable purchases of private debt just are another way to raise taxes. The really bad purchases (the ones that go bankrupt) act like tax rebates.

[Adam]

Hi Guys, I'm reading and re-reading the post and comments and I just can't make sense of any of it. Surely the fact that higher long rates tend to be associated with high expected inflation comes from situations where the causality runs from inflation expectations to interest rates. Would it really ever be the case that higher long rates caused higher expected inflation?

Furthermore, if higher long rates don't cause higher inflation expectations then the BoE driving long rates higher would be equivalent to the BoE raising long term real rates and this would be unambiguously contractionary.

Seems to me that the BoE can only have one goal in mind, lower long term REAL rates. That's all. If that can be accomplished with higher nominal long rates (because inflation expectations went up even more) then fine but that does not mean they are "trying" to raise long term nominal rates. Real rates are what matters here and lower real rates is what they are trying to get.

[Nick Rowe]

Hi Adam:

1. Higher long rates might come from higher expected inflation, but they might also come from higher expected future real rates, that would occur if monetary policy succeeded in changing expectations from "don't invest, since we are in a recession" to "invest, since we will recover from the recession". Am I implicitly assuming multiple equilibria? Probably.

2. When I stop thinking about monetary policy as changing the current and expected future quantity of money, and think of it is changing interest rates, I find it hard to make sense of it either, just like you. By thinking of monetary policy as changing the current and expected future supplies of money, and looking at the two effects separately, and noticing that long interest rates go in different directions for each of the two effects, it helps me to try to get my head clear. Which suggests that thinking of monetary policy in terms of an interest rate channel may be problematic, and leading to the paradoxes.

But I wish my head was clearer on it.

[Nick Rowe]

Dee: I don't know if the BoE was buying long gilts back in the 60's. It wouldn't surprise me. What surprises me is that OMOs in longer bonds are seen as unconventional monetary policy. I used to think that OMOs were conventional, and couldn't see much difference between long and short bond OMOs.

This is a bad time for governments to be selling real assets; they should if anything be buying, especially if it will be at least partly money-financed. Unless it's a "cunning plan" (as Baldrick would say) to sell it to Zimbabwe?

JimP: I think the Hong Kong government bought stocks to (successfully) fight a financial crisis. I don't know of any other examples.

David: Thanks. I don't really have a counter-argument against credit-spread targeting. I don't think I'm against it, either, it's just that I'm not clear on the mechanics of it. Does the Fed buy the higher interest rate asset, and sell the lower, or what? And what is happening to the money supply?

Scott: Thanks. And those three examples do buttress the argument against long bond yields being a good indicator of monetary policy. Funnily enough, when the financial crisis hit, and I got distracted by blogging, I was working on a paper on exchange rates under inflation targeting. You can make very similar arguments about how the exchange rate will respond to an increase in the overnight rate target. If market participants view it as a signal that the BoC thinks the economy is stronger, and if they believe the BoC has superior information, the CA$ will appreciate. But if the market participants think that the BoC is mistaken, and so will need to reverse course in future, the CA$ will likely depreciate.

Interesting to note that the recent (last few days) increase in stock prices seems to be coinciding with a fall in long bond prices. I wonder if this is a sign that QE (or something) is working? (I have no idea if this is the beginnings of recovery, or just a sucker's rally.)

bob: I've been trying to get my head around targeting futures prices. Still not sure. I can't help thinking (another automobile analogy coming up) that it might be like trying to make the car go faster by moving the speedometer needle. But also thinking through the paradoxes of targeting a private forecast. Will probably save my thoughts for a post.

Winslow: Most would argue the other way. People lose $100 in bonds, but they get $100 in money instead, so the immediate effect on wealth is $0. And if bonds imply a future tax liability, and there's even a partial "Ricardian effect" (people foreseeing that liability), wealth will rise.

[Scott Sumner]

Nick, Based on my study of the Great Depression, I think one tends to find certain correlations much stronger when there is great uncertainty about the future course of AD. So during normal times bond prices and stock prices may not be all that closely correlated (or maybe even positively correlated), but when there is great uncertainty about the future course of AD, then stock prices are much more volatile, and also more strongly correlated with long bond yields (and maybe even commodity prices--although that correlation was stronger in the interwar period than today.) But last fall when a correlation between commodities and stocks seemed to develop (just eyeballing the data), it reminded me a lot of the Depression.

[Adam]

Well, maybe it's only a matter of your word choice but I don't think there's any case where they are directly trying to raise long rates. It may well be that higher long rates are indicitive of success but that doesn't mean they were trying raise long rates, it would just be a case that higher long rates are a side effect of success.

[Nick Rowe]

Scott: just to be clear, is this what you are saying:
In normal times, when expectations of future AD are well-anchored, we tend to see a small *positive* correlation (if any) between stock prices and long bond prices?
In the Great Depression, and recently, we tend to see a stronger, *negative* correlation between stock prices and long bond prices?

Adam: thinking about word choice does seem to help me think more clearly about it, and your example of word choice helps. It's a bit like the saying in the 70's and 80's: "The way to lower (nominal) interest rates is to get the BoC to raise interest rates".

Here's another way of thinking about it: if the BoE tries to buy a certain quantity of bonds, and the announcement that it is doing so restores confidence, the increased demand for bonds from the BoE will be met with an even bigger reduction in demand for bonds from the public. The BoE tries to buy L75 billion, and when it goes to the market it finds that L150 billion are offered for sale (at existing yields). Now this is roughly what happened (the offer to buy was over-subscribed), but that might have been a sign of success, or just a sign that it offered too high a price.

Bob: I just read the Sprott paper you linked to (see bob's comment above). It is fascinating reading.

[bob]

so, any thoughts on the Fed announcement?

[Adam]

Good, corrrect and about time.

[Nick Rowe]

Well, the stock market seemed to like it, which is good news, and so did the bond market, which is ambiguous news!

Was it the announcement about buying bonds, or the announcement that they were thinking about buying other stuff as well?

I don't know, but something seems to have been working this last few days.

[Nick Rowe]

And the US$ fell on the news, also a good sign.

[Nick Rowe]

And the price of oil, and gold, increased. But arguably the biggest news is that expected inflation increased, judging from the TIPS spread: http://acrossthecurve.com/?p=3964

All of this assumes that the Fed's announcement was bigger than expected. But my reading of the news is that this is a reasonable assumption. Since the markets are only responding to the unexpected component of the Fed's announcement, we can be reasonably confident that the total effect was bigger than what we observed, because the expected component was already priced in.

[bob]

Yep. I'm trying to figure out the expected vs. unexpected aspects of that announcement. I think the treasury buying was the main unexpected market mover. I think additional Agency MBS and ABS purchases were probably a given. It's too bad that they announced all the measures at once, and not just the treasury purchases. It would have been nice to see what effect the news would have in isolation.

I was actually expecting long treasury purchases after reading accrued interest yesterday morning (and this blog for the past month):
http://accruedint.blogspot.com/2009/03/i-suppose-i-could-hot-wire-this-thing.html

I was surprised by the sudden jump in equities, but that seems to have cooled off. We've had 200 point rallies on no news in recent months, so I didn't see the equities reaction as being especially strong, just sudden.

I'm still trying to make sense of what this all means, and how much more QE the Fed is going to pursue. I have a feeling that we are in for a repeat of Spring-Summer 2008. Bernanke succeeded in stoking inflation fears amongst investors, but with the underlying problems in the economy (unemployment, demand) persisting, I think most of the money will flow in to commodities rather than equities or corporate bonds, only to get obliterated when demand drops further. I may be wrong, but QE still seems like pushing on a string to me.

[bob]

Basic Materials +3.82%
Energy +3.66%
Financial -1.11%

From todays movement, it looks like the long commodities/short financials trade from last spring may be coming back into effect.

[bob]

a couple good articles on the Fed move yesterday:

Don't Get Your Hopes Up - Scott Summer
http://blogsandwikis.bentley.edu/themoneyillusion/?p=671

Quantitative Easing - Jim Hamilton
http://www.econbrowser.com/archives/2009/03/quantitative_ea_1.html

this one is a Montreal economist who contributes to Naked Capitalism. It's in reference to the British QE & pensions, but still well worth a look:

Quantitative Easing Bringing Pensions to the Brink? - Leo Kolivakis
http://pensionpulse.blogspot.com/2009/03/quantitative-easing-bringing-pensions.html