Why nominal GDP matters.
This is a follow up to my somewhat misunderstood previous post. In the comment section Bill Woolsey made the following point:
Krugman is explicitly saying that real interest rates must be very negative to motivate an increase in real and nominal expenditure.
If the Fed promised 3% inflation, people would still not spend much, and any increase in the quantity of money (aimed at generating that inflation) would be hoarded.
The Fed’s promise of 3% inflation would have little effect, and inflation wouldn’t be 3%.
I agree with Bill that this is what Krugman has in mind. It’s hard to be sure, but he has said that “high” inflation expectations were needed, and at the same time linked to Fed studies showing the Taylor Rule implied a negative 6% interest rate was needed for a robust recovery. This of course implies that expected inflation needs to be at least 6%, as nominal rates on loans can’t be negative. But Krugman also says the SRAS curve is fairly flat right now. He frequently uses the Keynesian “slack” model of inflation, which suggests that when unemployment rates are very high a large increase in AD would initially lead to much higher output, with at best a small rise in inflation. I have some problems with the slack model of inflation, but in this case I think Krugman’s about right. If we had a 10% rise in AD or NGDP over the next 12 months, then we’d probably get around 7% or 8% RGDP growth, and around 2% or 3% inflation.
Now my readers may disagree with this assumption, but the point is that everything Krugman has written on the subject suggests to me that he think these numbers were about right, and yet they completely contradict his statements about the sort of inflation expectations that we’d need to get a robust recovery, at least if I (and Bill) are reading his views on inflation expectations correctly. To see why, consider what would happen if the Fed adopted a monetary policy expansionary enough to move inflation expectations (in the TIPS and CPI futures markets) up to around 3%, (as compared to the current figure of less than 1%.) If the SRAS curve is relatively flat, as Krugman seems to assume, then those sorts of inflation expectations would be associated with pretty high real GDP growth expectations, indeed the sort of rapid recovery he pointed to approvingly in his earlier piece on the 1976 and 1983 recoveries.
Good economists rarely fall into logical contradictions. And it’s obviously very possible that I have misconstrued Krugman’s views on either inflation expectations or the SRAS. But let’s assume I haven’t; what lessons can we draw from this contradiction? In my view it shows the importance of focusing on NGDP expectations, and the folly of focusing on interest rates. Interest rates are determined in a forward-looking way. If the nominal economy is very weak, and the weakness is expected to continue, interest rates will be low. But if the nominal economy is weak, and yet very rapid nominal growth is expected, then the equilibrium interest rate will be higher. In other words, an expansionary monetary policy that is expected to be successful might not require any reduction in interest rates at all. In contrast, a backward looking Taylor Rule that just plugs in historical data will implicitly be very pessimistic about the future as well, and will suggest that extremely low interest rates are required to get people to spend. That may be true if the policy is expected to fail, but then why would anyone want to adopt a policy expected to fail?
We can never directly estimate the “right” interest rate for a given macroeconomic situation. Instead we need to adopt an NGDP target (or at least a price level target) and let the markets decide what interest rate is appropriate when enough money is injected to move the expected NGDP (or inflation) rate to its target level. Both Bill and I have discussed how this can be done with index futures convertibility. But if we don’t have that system in place, we can’t avoid the problem by simply dodging the issue. We must infer market estimates of expected NGDP growth as best as we can. We can look at various proxies for inflation expectations and real growth expectations. There is a circularity problem that makes this approach inferior to index futures convertibility, but it’s still much better that what we are doing now.
2. MV=PY, or more precisely; M*(PY/M) = P*Y
Some people agreed with Krugman’s argument that the MV=PY equation is not helpful and indeed can be misleading, and didn’t like my criticism of his critique of the quantity equation. I also think the equation is pretty unhelpful, and can be misused. But that’s not what I was criticizing. Krugman’s post was directed against a David Beckworth post that never once even alluded to the quantity equation. Instead Beckworth pointed out that NGDP growth rates soared during the Great Inflation, fluctuated around 5% during the Great Moderation, and fell into negative territory during the recent recession. And David is quite right that these three changes in the behavior of NGDP caused each of those three historical episodes. What do I mean by “cause?” I mean they were both necessary and sufficient conditions. Given the structure of the US economy, persistently high NGDP growth rates almost inevitably lead to high inflation. And a sudden decline in NGDP is almost inevitably going to lead to a severe recession. If the Fed had kept NGDP growing at 5% a year then the current recession would have been either mild, or non-existent. And David is right that monetary policy could have prevented this from occurring. Indeed if the Fed, the ECB, and the BOJ had higher inflation targets, and promised to adhere to level targeting, then monetary policy can even be highly effective in a liquidity trap. Where did I learn this? From reading Krugman’s 1998 paper on the liquidty trap. What does MV=PY have to do with any of this? Nothing. That’s why I thought it was such a non sequitor for Krugman to start out by implying that falling NGDP isn’t the problem (of course it is precisely the problem—even Krugman doesn’t believe RBC explanations of the current recession) and then back up his views with a completely off-topic discussion of the quantity equation, something Beckworth never even mentioned.
Yes, the other paper cited by Krugman did mention velocity, but only to argue that it suggested a need for fiscal stimulus, which Krugman supports. So the MV=PY equation was misused in my view, but it was misused by a Keynesian, not a monetarist.
3. Free Exchange
The blogger for The Economist also weighed in on the issue here. I can’t really fault him/her for anything specific, but I still think that the focus on inflation, combined with a certain ambiguity in the English language, leaves a misleading impression:
Scott Sumner makes the same point:
“As a practical matter if monetary injections have no impact on the expected future inflation rate then they won’t impact current [aggregate demand].”
I suppose this is partly my fault, as my quotation suggests that I view higher inflation as an unambiguous “good thing.” What I really favor is higher expected NGDP growth. Here are my views:
1. Higher expected NGDP growth is needed right now.
2. Higher NGDP growth can be achieved through boosting AD, perhaps by monetary expansion.
3. As a practical matter, higher expected NGDP growth will raise inflation expectations somewhat.
4. So in this sense higher inflation expectations could provide an indication that the needed boost in AD is occurring.
5. However, for any given increase in NGDP (which is what really matters) it would be better to have more real growth and less inflation.
I don’t like focusing on expected inflation and real interest rates because I think these concepts cloud the picture. Yes, if nominal rates are stuck at zero, higher expected inflation does reduce real interest rates. But low real interest rates are also a symptom of a weak economy. I don’t see low interest rates as a “good thing.” Higher expected NGDP might well increase nominal rates, possible even real rates if the expected recovery is very strong. And I don’t see interest rates as being important in the transmission mechanism. I can’t really blame Free Exchange for emphasizing inflation, because most economists do prefer to view the transmission mechanism in terms of real interest rates. But I see an expansionary monetary policy as doing the following:
1. Boosting expected future NGDP through the excess cash balances mechanism (as long as the liquidity trap isn’t expected to last forever.)
2. Higher expected NGDP boosts the current price of assets such as stocks, commodities, and real estate.
3. Higher asset prices boost current AD and then NGDP.
4. If the SRAS is currently fairly flat (as we hope) this leads to a lot more real growth and just a little more inflation.
And that’s what we want, isn’t it?
Tags:
8. November 2009 at 14:20
Baloney. It’s a fundamental contradiction of logic to confuse a logic / math construct for a causal explanaion — and “good” economist do this constantly.
“Good economists rarely fall into logical contradictions.”
And note well, there is a lot of evidence on the side counting against your description of Krugman as a “good” economist — e.g. he lacks scholarly ethics, he’s badly trained in the history of economic thought, his NY Times column is routine full of bad economics, etc., etc.
Moreover, how useful a metric of “good” can it be when we can’t use that metric to distiguish better explanations from worse ones, e.g. equally “good” economists give us non-compatible and contrary economic conclusions and explanatory strategies.
Imagine engineers who each produced “good” engineering results that showed the other guys building would fall down. There’s no way we’d say they were both “good” engineers — and if there was no metric for picking one of these results as a “good” result and one as not “good”, we’d abandon the idea that we can provide any principled way of demarcating between “good” economists and bad ones.
In this zone of macroeconomics and public policy there is no ground at all for saying that Krugman is a “good” economist — he’s shown dishonesty, incomptetence and lack of professional behavior in far too many ways.
8. November 2009 at 14:24
Greg, I agree that is why we need to start using skill scores for economic predictions and have a central place to record these predictions.
8. November 2009 at 15:12
Greg, OK, Replace “good economist” with “someone I regard as a good economist.” That’s objective, isn’t it?
Doc, They’d have to be conditional predictions.
8. November 2009 at 16:03
Forgive my ignorance, but how do higher asset prices lead to higher AD?
8. November 2009 at 16:43
Sure, Scott, good point.
Conditional predictions, it is.
8. November 2009 at 17:18
The logical contradiction is in the belief the wage inflation is linked to the loan market. I simply cannot agree that this is true, short rates are associated with the financing of working capital–primarily inventory holding not wages.
Slackness in the labor market is irrelevant.
I think krugman is following in a false formalization of the excess demand for money. People need more income, wage inflation would give this, but so would any purchase of assets–again the problem is that the economy is large and the base is small. You need to disperse the funds via the loan market to gear up properly. But whether this is a problem depends on ppi.
8. November 2009 at 17:19
Interest rates need to go up for recovery. But thats one side of the balance sheet. If they go up and Gross Domestic Revenue (Think of total sales revenue for all business, including intermediate business) tanks, or does not quickly resume its previous highs, then the recovery in employment and business improvement will be delayed.
Scott your reasoning is sound if you apply it to nominal gross domestic REVENUE. Not nominal gross domestic product. While we need many metrics to appraise an economy, just as a value investor needs many ratios to appraise a stock, anything to do with GDP has to be judged as somewhat down the list.
8. November 2009 at 17:25
This is slightly off subject, but we’ve discussed on this blog that you use Mishkin’s Money & Banking text for your classes. I have switched to Laurence Ball’s new text, which has a much better in-depth treatment of a liquidity trap (Mishkin mentions it in passing on a page) and how monetary policy might address it. Using that text has allowed me to share your ideas of nominal GDP targeting with my students, and even some of your blog posts–or other posts/articles referring to your blog. I definitely recommend checking it out as you might find it easy to meld your approach into it as well.
8. November 2009 at 18:09
The problem is there is no such thing as a liquidity trap. There is only the resurgence of the demand for money and/or cash for holding, and the potential destruction of ponzi-money. The combination of cash-injection via debt retirement and the reserve asset ratio, ruthlessly and dramatically applied, can always over-ride a ponzi-money collapse, no matter how severe.
8. November 2009 at 18:37
Scott, I want to come back to something that we discussed in the comments of the last post. You have said that you use the TIPS and CPI futures as your measure of inflation expectations. Since TIPS are adjusted for CPI inflation isn’t this basically the same thing? I also seem to remember that you have acknowledged that CPI has problems as a measure of inflation, so I don’t see how you can base so much of your view about current monetary policy on this one measure of inflation expectations.
My point is that I think you may be underestimating the current NGDP expectations. There are a lot of contradictions in the market right now, but generally I see policy as more stimulative than you do. Yes, I would like to see interest rates higher to confirm that and I’d like to see a more general rise in commodity prices outside oil, gold and few other contracts, but the fall in the dollar leads me to believe that those issues will be resolved relatively soon.
8. November 2009 at 18:56
Good post Scott. I think that some people may not understand your important point about weak expected future NGDP affecting equilibrium real interest rates. So I decided to re-state it in ISLM-speak. http://worthwhile.typepad.com/worthwhile_canadian_initi/2009/11/why-current-ad-depends-on-expected-future-ad-scott-sumner-in-islm.html
I know you don’t like ISLM, but I think it needs doing.
9. November 2009 at 09:10
Krugman blogged TIPS again today, worth a read, but I think he is driving the nonsense bus off a cliff with it somewhere in there. Still amzed that what passes for goverment these days hasn’t listened to one of you on this matter. I mean if the truth was that more stimulus is needed and fiscal policy can’t be brought through due to lack of political will, what is tying the Fed’s hands?
9. November 2009 at 13:07
jsalvatier, There are a variety of mechanisms:
Higher real estate prices encourage production of new buildings
Higher stock prices encourage corporate investment
Higher commodity prices encourae more production and exploration
Higher wealth encourages more consumption
As these factors put people back to work, higher employment also encourages more consumption
Doc Merlin, Go to Mankiw’s blog and look at the graph of the Obama adminstration’s prediction of the U-rate with and without stimulus, if you want to see a conditional prediction that failed spectacularly. Indeed, even if you adjust for the fact that the economy worsened more than expected before the bill was passed, the post-passage performance is still much worse than expected.
Jon, I agree that short rates are not very important for wages.
Not sure I followed the last part about the ppi.
graeme, I will keep an open mind. I’m still haveing trouble figuring out how important the distintion is between NGDP and nominal domestic final sales (which Bill Woolsey recommends.)
JTapp, Thanks for using my blog in your classes. I actually had lunch with Larry Ball (and Mankiw) last fall. He didn’t seem very enthusiastic about the role of expectations, so I wasn’t sure whether I’d like the text. But a quick lunch conversation can be misleading. I’ll take a look at the text sometime.
Graeme, The weird thing about liquidity traps is that almost everyone agrees that at least some kinds of monetary policy are still effective. And almost everyone agrees that you can’t just lower the fed funds target any further. So the real debate is about which ways of dealing with zero rates are best. I agree with you that much of what is said about liquidity traps is nonsense. I don’t even like the term “trap.”
Joe, Good question. the CPI is flawed, but what else do we have as a measure of inflation expectations? I should clarify my views on NGDP expectations. i think 5% NGDP growth is quite likely. I also think, however, that we have fallen so far below the trend line that we need a bit faster growth in the catchup period. I can’t give you an exact figure, but when I started the blog in February I said that I would use that NGDP as a conservative starting point (it was already too low.) But we have fallen several more points behind, so even using that ultra conservative target, at least 7% NGDP growth would be helpful for the next 12 months.
more to come . . .
9. November 2009 at 14:24
Nick, Thanks, I left a comment over there.
Travis, Yes I saw it. Not much there to comment on.
9. November 2009 at 17:27
“Graeme, The weird thing about liquidity traps is that almost everyone agrees that at least some kinds of monetary policy are still effective. And almost everyone agrees that you can’t just lower the fed funds target any further. So the real debate is about which ways of dealing with zero rates are best. I agree with you that much of what is said about liquidity traps is nonsense. I don’t even like the term “trap.””
How we square that circle is that we realise that the bankers have built up a system that suits them. Instead of public choice pull factor its bankers choice pull factor. Even so far as having monetary policy that isn’t monetary policy but bank subsidies.
See how in a panic everyone is saying “We’ve got to recapitalise the banks.” No we don’t. How about recapitalise Scott. How does that grab you. Forget about the banks lets recapitalise Scott.
Low interest rates, particularly when the stock market has crashed, well they are a subsidy. How about give me zero% loans. I’ll refinance the house. Buy shares when they are down. Pay off my credit cards and so forth.
What about a TARP program for me. Old paper scribbled on going at ten bucks per sheet. Each item of clothing I no longer use that much going for $500 a hit. This is abuse what the banks have gotten away with even in the time of Keynes. We have to call it for what it is it is stealing.
And consider this liquidity traps hocus pocus. Its not valid for monetary policy. Its only valid for bank subsidies. So if we say monetary policy is monetary policy and bank subsidies are not monetary policy and not allowed under the equal protection clause, then there is no liquidity trap and there never could be one. Since direct cash injection through debt retirement and increasing the RAR is always effective. Can never be not effective. And in fact these two measures are the only actual monetary policy as differentiated from mere stomach-crawling to the banks.
10. November 2009 at 00:00
Scott, you remark: “Not sure I followed the last part about the ppi.”
1) “Efficient” fed policy (small changes in the base leading to big changes in macro-variables) is constrained by the zero-bound because the money multiplier declines. If banks don’t recycle the base, much more substantial base injections are required. This is why Krugman keeps disagreeing with you.
2) Changes in the Fed Funds rate primarily appear in the cost of borrowing of working capital, not consumer financing–because these sorts of loans constitute the bulk of short-term lending by banks. The effect of fed policy on long-dated debt is tenuous.
3) CPI is a wage proxy. Unemployment tempers wages, and so it tempers CPI, but this does not a liquidity trap make. The Fed can get traction even if the labor market is slack because the relevant inflation statistic is producer prices, e.g., the prices of finished goods or inputs.
Indeed from this perspective, its clear how tight Fed policy became late last summer: http://lostdollars.org/static/ppi.png and how quickly inflation responded to Fed easing thereafter. Nonetheless, PPI is still running below its trend in the past decade.
10. November 2009 at 17:15
Graeme, The “opportunity cost” of a bad monetary policy is that the public got railroaded into thinking we had to bail out the banks. I understand the fear people had last fall, but we were mostly looking at the wrong problem. And when we did bail out the banks we found out the real problem was falling NGDP, and that problem didn’t go away (or falling revenues in your approach.)
Jon, I understand that hoarding worry, but it’s all in the context of expectations. If you commit to do whatever is necessary to boost NGDP rapidly, then people won’t want to hoard the monetary injections. Especially with a negative rate on reserves.
I agree about the CPI and wages, especially the core CPI. And I agree that at zero rates the Fed can influence a lot of other prices (commodities, real estate, stocks) that aren’t in the CPI.
10. November 2009 at 17:48
Scott: I won’t call it hoarding at all. But my key point was that the PPI matters not some other inflation index.
10. November 2009 at 18:23
If you want NGDP at about 5%, how do you do that?
Get people to spend savings? Get people to spend in the present? Get people to spend with more currency denominated debt? Get foreigners to do the spending?
10. November 2009 at 18:29
“jsalvatier, There are a variety of mechanisms:
Higher real estate prices encourage production of new buildings
Higher stock prices encourage corporate investment
Higher commodity prices encourae more production and exploration
Higher wealth encourages more consumption
As these factors put people back to work, higher employment also encourages more consumption”
It seems to me you are ignoring whether those higher prices are from currency denominated debt. If so, are there imbalances that have built up over time?
10. November 2009 at 18:33
Graeme Bird said: “See how in a panic everyone is saying “We’ve got to recapitalise the banks.” No we don’t. How about recapitalise Scott. How does that grab you. Forget about the banks lets recapitalise Scott.”
Right, it is not the banks that need recapitalized. It is the lower and middle class in the high wage countries. That is where the break in the chain of payments is. Too bad the spoiled and the rich along with the fed do not care about the lower and middle class.
10. November 2009 at 18:37
“Graeme, The weird thing about liquidity traps is that almost everyone agrees that at least some kinds of monetary policy are still effective. And almost everyone agrees that you can’t just lower the fed funds target any further. So the real debate is about which ways of dealing with zero rates are best. I agree with you that much of what is said about liquidity traps is nonsense. I don’t even like the term “trap.””
I believe it is surprisingly easy to get out of a liquidity trap. The problem is the spoiled and the rich and the fed don’t like the solution because they like their debt slaves just the way they are.
10. November 2009 at 20:25
“And I agree that at zero rates the Fed can influence a lot of other prices (commodities, real estate, stocks) that aren’t in the CPI.”
After telling us there was not a housing bubble, why should the fed be allowed to do/influence anything???
11. November 2009 at 08:39
Jon the PPI has both advantages and disadvantages over the CPI. It is more timely, and less distorted by sticky prices. But it reflects both monetary and supply shocks. Which is why the NGDP is better than either. I used the PPI (WPI) in my depression research.
Too much Fed, I am not saying the Fed should try to target asset prices like stocks, commodities and real estate. I just meant that those variables are impacted by Fed policy; hence even in a liquidity trap when rates can’t fall further, Fed policy affects other variables in the economy.
Last spring I had a number of posts explaining how I think Fed policy affects spending. Basically it works through two factors. First, the monetarist excess cash balance mechanism explains the long run effect of more money on NGDP. Second, the expectation of the long run effect impacts current asset prices, which impacts current AD.
11. November 2009 at 11:41
Got milk?
Oops! Got links to the posts?
12. November 2009 at 06:36
Too Much Fed, Try this one.
http://blogsandwikis.bentley.edu/themoneyillusion/?p=461