Paul Krugman presents a simple formal model of a liquidity trap. He shows that monetary policy won't work, but fiscal policy can work, to bring the economy to full-employment.
Actually, that's not right. What his model shows is that current monetary policy won't work (because the nominal interest rate is at zero); but future monetary policy will be very effective in his model. A (permanent) 10% (say) increase in the future money supply in his model will cause a (permanent) 10% increase in the future price level, and this will cause current real interest rates to be minus 10%, which should be enough to get the economy back to full employment.
Krugman knows this, of course. It's not a new point. The question is: how can a central bank change future monetary policy today? How can a central bank change people's current expectations about future monetary policy and price levels? The biggest failure of current monetary policy is not what central banks are doing today; it is that they are failing to guide expectations about what they will do tomorrow.
I have read enough comments at various blogs to know that people's expectations of the future US price level are all over the map. Some (deflationists) expect future prices to be lower than today's; others (inflationists) expect future prices to be much higher than today's. There is no anchor. Anybody could be right.
The US Fed offers no guidance on expectations about future price levels. At least, I haven't heard any, and it should be offering guidance loud enough and clear enough that everyone should hear it. The Bank of Canada does offer some guidance: it remains committed to its 2% inflation target.
For a central bank in a liquidity trap, formal commitment to a 2% inflation target would give some help. But formal commitment to a price-level path rising at 2% per year would be better.
Readers who already understand the difference can skip this paragraph. If a central bank controls inflation perfectly, there's no difference between targeting 2% inflation and targeting a 2% price-level path. In both cases the price levels grows: 100, 102, 104, 106, etc. (I can't be bothered to do the compounding). But central banks can't control inflation perfectly. If an inflation targeting central bank makes a mistake in one year (only), the price level will go: 100, 101, 103, 105. If a price-level path targeting central bank makes the same mistake, the price level will go: 100, 101, 104, 106.
Both inflation targeting and price-level path targeting central banks will be driven off course during a liquidity trap. We don't know how long the liquidity trap will last, and we don't know how much deflation there will be while we are in the liquidity trap. With inflation targeting, all we know is that inflation will eventually return to 2%, but that the price-level will be permanently lower, by an unknown amount. With price level path targeting we know that there will be no permanent effect on the price level, and that the longer the liquidity trap lasts, and the worse the deflation, the higher the rate of inflation will eventually be to get us back on target.
It's that last bit, in bold, which shows the benefits of price-level path targeting. Expected inflation is the medicine for a liquidity trap, because it can make real interest rates negative even when nominal interest rates cannot be. Price-level path targeting is a regime in which the medicinal dose gets bigger as the disease gets worse.
Canada is not yet in a liquidity trap. The decision on whether to continue with inflation targeting or switch to price level path targeting will be made in 2011. Perhaps that's soon enough; perhaps not. In any case, the Bank of Canada's actual policy looks very close to price level path targeting.
The US is in a liquidity trap. The Fed does not even have formal inflation targeting to provide some sort of anchor for expectations of future price levels, and it shows. There is a strong argument for Ben Bernanke to announce a target time path for the future US price-level, even if he needs to admit his uncertainty about how long it will take to get onto that path. As long as people believe he will eventually get onto the path, the announcement will work.
Current fiscal policy is getting all the attention; future monetary policy is being ignored.
That is a very interesting post. Can you imagine the political uproar though if the Bernanke Fed announced that it wanted to see the US price level 5-10% higher in the next 2-3 years? (side questions - which price level? would housing be included in the price index chosen? does this qualify as asset price targeting?)
Posted by: brendon | January 05, 2009 at 06:05 PM
Thanks Brendon! I'm not really a Fed-watcher, but I think the Fed has made a number of statements about trying to keep inflation around 2%, though without a formal commitment, so a path for the price level growing at 2% should not be too revolutionary. The hardest part would be trying to explain to the math-challenged the difference between a 2% inflation target and a 2% price level path! And given the very revolutionary changes to the Fed's balance sheet, I don't think it would raise eyebrows in comparison.
Which price index? Dunno. I think it should probably be a broader price index, including some asset prices. But that might make it harder to communicate. Because asset prices are more flexible, and the data on many asset prices (not houses) is available more quickly, and because it is the level of real asset prices (as well as the rate of change of nominal asset prices) which affects AD, I think price-level path targeting is even more appropriate when asset prices are included in the index.
But my head's not clear on the interrelationship between the two questions: narrow vs. wide price index; price level vs. inflation targeting.
Posted by: Nick Rowe | January 05, 2009 at 07:09 PM
Nick: wouldn't this also serve to reduce a lot of the perceived inflation risk from long term nominal interest rate bonds? It might offer some comfort to people retiring that there is a target price level for 10, 15, 20 years out.
Posted by: Andrew F | January 05, 2009 at 10:05 PM
I guess my thinking was specific to the current US housing situation - if the Fed announced a broad price level target that included a significant weight on housing it might stem the decline in the market -though somewhat artificially since inventories are still too high.
Posted by: brendon | January 05, 2009 at 10:14 PM
Andrew: yes, the main argument for price level vs inflation targeting is that the price level becomes much more predictable at long horizons. Under inflation targeting, the price level follows a random walk (with drift), so the forecast variance goes to infinity in the limit.
Brendon: I'm not sure if we would want to influence the real price of houses, provided we don't need to do so to get out of the liquidity trap.
I expect my main motive for writing the post was frustration with all the focus on fiscal policy, plus quantitative and qualitative easing with current monetary policy, with no focus on what might be the most effective way to stimulate demand -- directly anchoring expectations of future price levels.
Posted by: Nick Rowe | January 05, 2009 at 11:06 PM
Nick: There is a lot of concern that recent global monetary measures may have been too lax (a Greenspanesque policy response on a global scale)and some fear inflation down the road. Personally, I can't help but wonder if this concern could be used to the central banker's advantage. Especially in situations where inflationary expectations are sticky around where the central bank says inflation will be.
This is what I have in mind. Purely hypothetically, if (say) there were problems lowering the nominal rate below 0.5% (below that some capital markets may not function well if there are transactions costs), and (say) the real rate of interest required to close the output gap (at this juncture) was -3%, then inflationary expectations need to be at least 2.5% to do the job. I wonder whether a credible inflation targeting central bank should consider announcing that it would tolerate inflation at the upper end of its target band (say 3%) while interest rates are at 0.5% as long as it takes to close the output gap and for real rates to normalise. Such a conditional announcement would essentially provide the same outcome as a switch to price level targeting but without a formal switch in the policy rule. The problem is that such a policy may be seen to erode the credibility of the central bank. What do you think?
Posted by: Hume Torrens | January 15, 2009 at 02:18 PM
Hume: a conditional inflation targeting policy like you describe, if it were made symmetric (so it works in reverse if inflation gets too high) would be very similar to price level targeting, but might be harder to communicate. In the US, where they don't have a formal target (but I'm hearing more talk in the last few days of them adopting one) they might as well go to a price level target right away. In Canada, your proposal would make more sense, since it's sort of finessing the existing inflation target, and still keeping within the 1% to 3% target range. Or they could just wait until the inflation target comes up for renewal, since the problem doesn't (yet) seem quite as urgent here as in the US. Dunno.
Posted by: Nick Rowe | January 15, 2009 at 02:53 PM