My Doppelganger

Here is a very interesting WSJ article on Lars Svensson.

STOCKHOLM — One of the architects of inflation targeting — now a widely used central-bank policy — says central banks should encourage expectations that they will let prices overshoot their target, and then do so.
That would help combat rising unemployment and aid economic activity around the world, Lars E.O. Svensson, deputy governor of the Swedish Riksbank, said in an interview.

At first glance, the overshooting would seem to violate my proposal to “target the forecast.”  But on closer examination it is exactly the same idea.  We both support the concept of “level targeting” which would make up for any shortfall in one period, by having the target variable grow extra fast the next period:

Mr. Svensson stressed that underlying price levels were a more effective target than inflation rates for a central bank in the long run.

You may recall that although Svensson is the least known of the “big four” Princeton professors I mentioned in an earlier post here, he is my favorite.  It’s good to hear that he is so influential:

“He’s had a lot of influence at central banks around the world,” said Michael Woodford, a professor at Columbia University and Mr. Svensson’s former colleague at Princeton University. Mr. Svensson’s view also contrasts with admonitions from European Central Bank President Jean-Claude Trichet, who last weekend urged euro-zone governments to cut back on public spending.

As I thought, Svensson was the man behind the negative interest rate policy in Sweden:

Earlier this month, he argued for the Riksbank to cut its main interest rate to zero, objecting to the eventual decision of the other board members to cut it to 0.25% from 0.50%.
But he did score a victory in that the Swedish central bank set its deposit rate in negative territory. That means that banks pay to store their excess reserves with the Riksbank, which gives them an extra inducement to turn those reserves into loans, rather than hoarding them.
“There is nothing strange about negative interest rates,” he told the other five Riksbank board members, according to minutes from the July 1 meeting.

I also said that we shouldn’t put too much weight on the Swedish policy, because currency depreciation is a foolproof way out of liquidity traps for small countries like Sweden.  So in terms of nominal aggregates like inflation and NGDP, it is simply a matter of determination.  Git er done!  Small countries can easily get as much inflation as they want.  I also argued that bigger countries might get some political flack from devaluing their currencies.  Svensson seems to agree:

Mr. Svensson has also advocated currency depreciation, or at least the prevention of rises in the currency’s value, as a tool against deflation for smaller countries like Sweden.
A weaker krona “is part of expansionary policy and means that activity is higher and unemployment lower in this country than they otherwise would be,” he said. He emphasized that a weaker krona shouldn’t be taken as a threat to trading partners. Even if a weak krona would make imports more expensive, he believes the boost to Sweden’s demand would ultimately result in more imports. “We can consume, invest and import more,” he said.
However, that logic doesn’t apply to major economies, such as the U.S., where a cheaper currency doesn’t have much impact on domestic demand but has a big impact on the rest of the world as its large amount of exports suddenly become cheaper, he said.

If you look closely, you will also see that he contested the same beggar-thy-neighbor argument that I tried to refute in my “What’s Good for China is Good for America” post; the idea that currency depreciation in a recession hurts other countries.  As Svensson points out, currency depreciation allows countries to grow faster and import more.

The following quotation discusses the difference between inflation rate targeting and price level targeting.  As far as I know there are no central banks that target price levels.  But he and I both believe level targeting would have been a big advantage in this crisis:

Within Sweden, he said keeping interest rates low will help moderate job losses and some inflation shouldn’t be feared. Prices are well below the point where they should be over the long run, he says, and a rise in prices would represent a return to those long-term trends.
Mr. Svensson describes this concept — price-level targeting — as a benchmark with a “memory.” Most central banks that have an inflation target aim to get to the same level of inflation, say 2%, whether or not inflation was far slower than that the year before. By focusing on prices, the central bank allows for inflation to overshoot targets temporarily to offset those occasions when it had undershot inflation targets the year before.

Price level targeting helps stabilize expectations, AD would not have fallen nearly so far last fall, as even a slight fall tends to create expectations of faster future growth.  Under our current backward-looking, memory-less, inflation rate targeting regime, falling AD just leads to expectations of more of the same.

I also discussed how Sweden was the only central bank in the 1930s to have an explicit price level target (FDR had a more vague goal of returning prices to pre-Depression levels.)  So they were very progressive even back then:

His idea fits with Swedish tradition. The Riksbank is the only central bank — back in the 1930s — to target price levels rather than inflation rates.

I have also argued that the Fed should target not just prices, but rather NGDP, which includes both prices and output.  This is an example of a “flexible price target,” as it anchors the price level in the long run, but allows some give during supply shocks.  Svensson doesn’t specifically favor NGDP, but basically likes the idea of a flexible price target:

“Flexible inflation targeting is a very good monetary policy regime,” Mr. Svensson said. “Flexible price-level targeting may be even better in the long run.”

I should mention that another of the more progressive countries in the 1930s was Britain, which left the gold standard in 1931, before any other of the major western economies.  Leigh Caldwell has a nice post today where he argues that Britain has also adopted one of my ideas.  I’m not so sure, but I do like the idea of targeting “nominal demand,” which is basically NGDP.  (Actually I believe it’s the final sales version recommended by Bill Woolsey, if I’m not mistaken.)

It’s interesting to consider the small group of economists who have suggested that monetary policy can and should have done more, and/or the payment of positive interest on reserves was a mistake:  Thompson, Hall, myself, Woolsey, Glasner, Svensson, Jackson, etc.  Every one of them was in the even smaller group of economists that focused on forward-looking monetary regimes of one form or another.  Once you start looking at monetary policy in a forward-looking fashion, everything seems different.



15 Responses to “My Doppelganger”

  1. Gravatar of Alan Alan
    22. July 2009 at 15:54

    I agree with most of the points you make, but I have one objection with the following phrase (which is more Svensson’s phrase):

    “Most central banks that have an inflation target aim to get to the same level of inflation, say 2%, whether or not inflation was far slower than that the year before. By focusing on prices, the central bank allows for inflation to overshoot targets temporarily to offset those occasions when it had undershot inflation targets the year before.”

    Most central banks I know of – like Canada, Australia, NZ and Britain – have an inflation target (more precisely, a target range) OVER THE CYCLE, not on an annual basis. Thus such CBs do allow for over- and under-shooting of inflation over the cycle.

    I do agree with your point that inflation targeting has more of a “bygones-be-bygones” approach – and hence allows for less over- and under-shooting than price level targeting. But I think it’s not totally correct to claim that inflation targeting (especially when done over the cycle) is “memory-less”.


  2. Gravatar of Alan Alan
    22. July 2009 at 16:25

    Actually Scott, out of the group of CBs I cited, I think its just Australia and NZ that target inflation over the cycle. I think Canada and Britain specify their inflation targets on a per annum basis (I think Brazil is another example)


  3. Gravatar of Lorenzo (from downunder) Lorenzo (from downunder)
    23. July 2009 at 00:04

    The first two comments are already about Oz, as is the question that struck me.

    One of the things I like about this blog, Scott, is that you very interested in experiences outside the US: something a lot of American commentators, even economic commentators, do not venture into much. Including when discussing the 1930s.

    It is very noticeable that the Oz economy is doing much better than the US, as the latest ABS (Australian Bureau of Statistics) figures indicate

    (Side point: the ABS presents its key national indicators rather better than the US equivalent )

    The latest external shock has reinforced our experience of the 1997 Asian crisis — a floating exchange rate is a very good thing for us. The old, unreformed Oz economy used to do particularly badly from external economic shocks: we now do much better.

    Anyway, you might be interested in Tony Makin on monetary policy being much more useful than fiscal policy,25197,25777608-7583,00.html

    But it also seems to me that our experience can be useful in shedding light on the entire crisis. After all, our housing prices are even more inflated than American ones,
    and in all our metropolitan markets, since we are all “Zoned Zone”, we have no “Flatland” (to use Krugman’s very useful distinction). Yet, we have had no financial system meltdown, no liquidity trap, etc. This I put down to
    (1) much better prudential financial regulation
    (2) we did not do anything anywhere near as stupid in housing markets apart from land-use regulation creating supply-response-suppressing scarcity (i.e. zoning) which, admittedly, is fairly stupid, but apparently not too stupid. (At least so far: some feel we can keep this up

    Just a thought that we might provide a useful comparison economy.

  4. Gravatar of Alan Alan
    23. July 2009 at 04:00

    Although I know I run the risk of over-commenting on one particular blog piece, as an Aussie, I can’t help but chime in when someone mentions something about the sunbaked country – especially when its related to economics and our relative stability during the financial crisis.

    I agree with Lorenzo’s points; all I would add is that Australia has no real subprime market, which meant that we didn’t have any of the problems that plagued the US in 2007.

    In addition, our inflation-targeting regime is much more forward looking than the US (or Britain’s); this has helped to maintain the stability of inflation expectations in contrast to that of the US. Although the RBA uses inflation – not price level – targeting, the key difference to that of most other IT-countries is that the RBA’s target is based on the outcome over the cycle, and so, IMO, is somewhat akin to Scott’s/Svensson’s idea of level targeting and monetary policy with “memory”.

    Oh, and it probably helps that we’re major trading partners with China…..

  5. Gravatar of ssumner ssumner
    23. July 2009 at 04:01

    Alan, I have to plead ignorance here, but I was under the impression that central banks did not try to offset past mistakes. Thus if inflation was currently too low, I thought that they would try to get back on target over an extended period, which might be one year, or might be over the cycle. But getting back on target is not the same as making up for past mistakes. For a central bank to do what Svenssen and I want, they would actually need to spell out a target path many years out. I really don’t think there is any central bank that says “if the price level today is 100, we will try to make the price level in 2024 equal to 130, and we will continue to shoot for 130, no matter how far off we get. So if the actual price level in 2023 is 143, we will shoot for 10% deflation between 2023 and 2024. Does any central bank work that way? I’m not sure, but I doubt it.

    Thanks Lorenzo, I have been a big fan of Australia in at two levels. One is the country and people, which I enjoyed when I lived there in 1991. The other is the economic reforms undertaken since the 1980s. Many people on the left idolize the Western European model. But Australia is a much more plausible model for the US. There share of government in GDP is even slightly lower than the US, but you have universal health insurance. Unfortunately I don’t see any signs we are learning any lessons from Australia.

    In the macro area there are more lessons to be learned. Australia does have big current account deficits like the US, but not big budget deficits. Thus I don’t think the two are linked (as many in the US argue.) In addition, by aggressively easing monetary policy you reduced interest rates enought to prevent a sharp fall in GDP. I also notice the 1.5% CPI number there, which is in contrast to our negative 1.4% number. I think Makin is right in the article you attached.

    Finally, I think Australia presents massive problems for the Austrian school of economics. Austrians keep saying that using monetary policy to try to prevent a recession will just store up more problems for the future. Australia’s last recession was when I was there in 1991. They avoided a recession in 2001 (I believe also with currency depreciation), and they have since had a housing “bubble.” And yet again they might avoid recession, or at worst have a very mild one. So when is the “lucky country’s” day of reckoning that the Austrians keep predicting?

    When people ask me what I have in mind regarding macro policy, Australia is as good a model as any. BTW, they were hit bad by the fall in commodity prices, so no one can simply say they didn’t suffer from any shocks capable of producing a recession.

    I don’t know much about Australian housing regulation, but I do think we can learn a lot from both Australia and Canada. I would like to see tighter rules on these sub-prime mortgages (zero money down.) I presume the Aussies didn’t get heavily involved in that type of mortgage. That may seem to violate my free market ideology, but since the government is on the hook anyway with deposit insurance and “too big to fail” it makes sense to encourage banks to be more careful.

  6. Gravatar of ssumner ssumner
    23. July 2009 at 04:05

    Thanks Alan, That fills in some of the gaps that I was wondering about when I typed my previous response. I’m glad to hear about both the lack of subprime lending, and the forward-looking monetary policy–that is music to my ears.

  7. Gravatar of Alan Alan
    23. July 2009 at 04:35

    You are 100% correct – the RBA and RBNZ (and, for that matter all other CBs) do not explicitly try to offset past mistakes in the sense that if inflation is 1% in one year, then the CB aims for 3% inflation in the next year (assuming a 2% inflation target per annum). There is certainly no explicit mandate to target the price level.

    I guess I was trying to distinguish 2 types of inflation targeting: where the target is set for a specific horizon (e.g. 1 year, as is the case with Canada and Britain); and where the target is based on a more forward-looking over-the-cycle horizon. The latter regime may allow for some over- and under-shooting (since the target is defined over the cycle), although I agree with you that this possibility can be made more explicit under a price-level targeting regime. FYI, I think there’s an old RBA paper (c.1995) that discusses some of these issues:

    Sorry I should have been a bit clearer.

  8. Gravatar of Bill Woosley Bill Woosley
    23. July 2009 at 05:36


    I think targetting a growth path for the CPI is a bad idea.

    The current policy of taking it where it is, and having be expected to change from where it is the proper amount seems sensible to me. Of course.. zero inflation sounds sensible too. So, the price level is shocked up. The central bank sets poilcy to keep it at that new level. The price level is shocked down, the central bank tries to keep it at that level.

    The Svensson argument sounds pretty good right now. The demand for money rose, aggregate expenditure fell, and the price level… well.. Let us pretend it clearly fell. Getting aggregate demand and the price level back up sounds sensible.

    But, suppose we had an adverse aggregate supply shock to the price level. What do we do then?

    Along with the arguments you made in the last post, about the problems with measuring the price level, this is another reason why nominal income targetting seems better. (Yes, final sales is the better measure, but the difference is just inventory investment.)

    Anyway, I adjovate getting nominal income back to an appropriate growth path. Frankly, all this discussion of getting expected inflation up to 2% is irritating to me. From the current price level?

    Still, I think a careful analysis of what would happen with various sorts of shocks that cause overshooting would be appropriate.

  9. Gravatar of Bill Woosley Bill Woosley
    23. July 2009 at 07:00


    Mises was all focused on how it is impossible to measure the price level. Just because it can’t be measured doesn’t mean that the price level (or purchasing power of money) doesn’t impact things.

    Mises, at least sometimes, defined inflation in terms of monetary disequilibrium. An increase in the quantity of money beyond the demand to hold it. (Which is an increase in the quantity of money, ceteris paribus.) He argued that the purchasing power of money must fall to equilibrate this. But measuring the purchasing power of money using a price index was hopeless. He didn’t even like the term “price level.”

    It has been a long time since I read Mises, and there are some who comment on the blog who appear to have him at their finger tips always, but I think it has to do with the demand for money by the market being the sum of the individual demands, each of which depends on the subjective expecation of the purchasing power of money. It is different for each person. Not just their expectatiosn, but what particular set of goods whose prices are relevant. Using the term “price level” suggests it is somehow objective and public for all.

    I tend to understand inflation using ceteris paribus assumptions as well, but not so closely tied to monetary disequilibrium. It is a change in the level of prices at unchanged relative prices. The exchange ratios are the same, but all the prices change in proportion.

    That is the concept of the price level.

    In reality, changes in the price level are combined with changes in relative prices. Changes in the nominal prices of the goods and the exchange ratios are mixed up. So the _measures_ of inflaton are all imperfect.

    Generally, price indices are used. But they have inherent problems. And that is where you are focused now.

    If you take a specific bundle of goods, you can measure its total price. But the relative price of the bundle can change, so that isn’t the price level. And actual measures of the price level change the compostion of the measured bundle as well as making quality adjustments because they are measusing the “cost of living” rather than “the purchasing power of money.”

    But to me, economic theory is interesting. And we can do that without any measures.

    Now, if you want to target some macroeconomic statistic, then it has to be measured. And the difference between the concept (say, total spending on final goods and services) and the measure (GDP or final sales) should always be kept in mind.

    Mises, in fact, complained about price level targetting because of these problems with price level measurment. (He favored a gold standard.)

    I am more and more irritated with talking about an 2% inflation goal. Targetting a growth path of nominal income is what I think should be done. But I think we can talk about the implications of this for inflation in response to various sorts of shocks, even if we cannot measure inflation.

    I think you should just stick with the notion that measuring problems in price indices is a reason not to target them or their growth rates and that total spending doesn’t have this problem.

    But the notion that we can’t measure it, so the concept is unimportant.. is a fundamental error.

    Is this have something to do with the positivism taught at Chicago?

    By the way, I want to second the comment by BB that an index of wages is frought with difficulty too. I presume you would favor increasing it 3% or something? If you really had it fixed, wouldn’t changes in relative wages involve all the stickiness problems you complain about?

    It would seem to me that as globalization reduced union monopoly power, their wages must fall. That automation might require drops in the wages of unskilled workers.

    Of course, with a target for nominal income growth, some nominal wages would need to sometime decrease.

  10. Gravatar of JTapp JTapp
    23. July 2009 at 17:40

    You often leave Ken Rogoff of your list of advocates for more responsive monetary policy during this crisis and I often wonder why. He was calling for an explicit inflation target of at least 6% last spring in a couple of newspapers. I’ve always thought he was a monetarist…not so?

  11. Gravatar of Lawton Lawton
    23. July 2009 at 19:55

    Here’s a down-under blog that covers the RBA and “economics and financial markets from a classical liberal perspective”. by Dr. Stephen Kirchner

    Among other interesting work is skepticism about bubbles, and of course a strong preference for monetary over fiscal stimulus.

  12. Gravatar of ssumner ssumner
    24. July 2009 at 04:25

    Alan, No problem. I think there are two issues being mixed up here, and it is causing confusion:

    1. Rate vs level targeting
    2. Inflation vs. NGDP targeting.

    I just read the abstract, but I think what the Aussie central bank might be thinking is that if the price level spikes up because of a supply shock, you don’t want to immediately bring it back down. But this is actually just another way of saying that the inflation rate is the wrong variable, we should be targeting NGDP. The solution to problems with targeting inflation is not to target inflation in a more flexible long run fashion, but to target NGDP.

    Under NGDP targeting I do strongly support level targeting not rate of change targeting. I arbitrarily picked February 2009 as my starting point, which was actually a very conservative choice, because the economy was already severely depressed by that time. Thus I favor a 5% growth path from that date. Since we have already fallen behind, from this point forward we should aim for 6% NGDP over the next year, and then 5% thereafter. If the Fed ever adopts futures targeting, then I might favor gradually moving to a lower number than 5%, perhaps 3% or 4%.

    Bill, See my answer to Alan. I also find it very irritating to have to talk about inflation targets, but since everyone else looks at things that way, I feel that I need to provide at least some kind of number. Right now TIPS spreads are less than 1% over 2 years, and less than 1.5% over 5 years, so I am very comfortable asking for stimulus. Nick pointed out that another futures indicator (CPI futures) is a bit higher on the 5 year, but it also shows excessively low inflation for the next two years.

    I see your point about the problems with CPI level targeting with a memory. But I think my response to Alan shows what I think the real problem is. I don’t think the issue you raise regarding level targeting would be a problem for NGDP level targeting. After a strong boom a few quarters of subpar NGDP growth, (like late 2006), does not cause any macro problems, and after a big fall a few quarters of above 5% growth is welcome.

    Bill#2, I partly agree with you about wages, which is why I prefer NGDP targeting. But not totally. Yes, it is true that relative wage changes also may be less than desired, and that that may cause economic losses, but aggregate changes cause far worse losses, due to mass unemployment. Relative wage frictions do not cause the economy to move away from the natural rate, rather they are just a factor behind the frictional rate of unemployment. So I continue to insist there there is a real powerful argument behind Thompson’s proposal. But I also do partly agree with the earlier criticism you referred to. If we can only measure a subset of wages (hourly workers) and if that subset sees their real wages fall from globalization, then yes the target is thrown off a bit. So I prefer NGDP.

    I agree that a variable can be important despite the fact that we can’t measure it. But I am also saying that we can’t even define it, or if you accept the textbook definition them we ought to call “inflation” wage inflation, because it is well known that people only feel like their utility is constant if they are “keeping up with the Jones.” I hate this type of psychological concept. But if you try to pin things down in a more concrete way, rather than just talking about psychology, it becomes very difficult in a world where products are changing fast.

    My criterion is what is useful?

    My answers:

    For living standards, the CPI.
    For PPP, only traded goods.
    For business cycles and Fisher effect, income or wage inflation.

    This is my most important point, not the problem that the CPI is difficult to measure.

    JTapp, Maybe Rogoff should be included. But I try to focus on those who I know for a fact thought that the Fed made errors of omission late last summer or early fall. I plan a new post on this soon where I will try to clear things up a bit. BTW, Mankiw called for a 3% inflation target in the spring, level targeting, so he was also critical of the Fed. I think Mishkin also advocated an explicit inflation target in the spring. But here is the thing, Mishkin is also full of praise for the Fed in his May AER piece, so that is not as radical a critique as I am making, and as at least some others are making. But I should say something about Rogoff.

    Thanks Lawton, I will take a look and comment later,

  13. Gravatar of Current Current
    24. July 2009 at 08:52

    Despite Bill’s having read it a long time ago he gets Mises exactly right. Mises describes these things on page 189-194 of “The Theory of Money and Credit”.

    Defining it is troublesome and there is no one right way, as Mises and Scott mentions.

    Although Bill and Scott seemingly disagree I’m not sure they really do. Scott, what Bill is saying is that we can theoretically disconnect “influences on prices that come from the money side” from “influences on prices that come from the goods side”. Although we can’t invent an index that can correctly navigate that divide it remains a useful distinction.

    Regarding targeting the inflation rate over the cycle. If this is a done then surely employees will realize it. Surely they will negotiate pay deals so that their pay increases in that manner. So, surely in the very long run it will become similar to long run targeting of stable prices?

  14. Gravatar of ssumner ssumner
    25. July 2009 at 05:06

    I am not sure that targeting inflation over the cycle is as effective as targeting it over a 12 month period, but its is better than nothing.

    Thanks for the Mises passage, I agree with what he wrote, but again, my main point isn’t so much the difficulty of measuring price inflation, but rather that even if we could measure it perfectly, it may be wage inflation that is more meaningful for both business cycle theory and the Fisher effect.

  15. Gravatar of Current Current
    29. July 2009 at 09:03

    Scott: “my main point isn’t so much the difficulty of measuring price inflation, but rather that even if we could measure it perfectly, it may be wage inflation that is more meaningful for both business cycle theory and the Fisher effect.”

    Ah, I see what you mean. I’m not sure I agree with you I’ll have to think more about it.

Leave a Reply