Scott Winship has a technical study finding all sorts of flaws in the Census data.  Here’s the punchline:

In short, while the middle class””and especially the poor””saw declines in market income after 2007, the safety net appears to have performed just as we would hope, mitigating the losses experienced by households. By 2011, the safety net had returned middle-class and poor households’ incomes to the highest levels ever seen. Since then, the situation has likely improved. Disposable income among the poor and middle class is probably at an all-time high.

And Matt Yglesias tells us why this should have been obvious, if you simply stop and think about living standards:

Just in time for the Fed taper.

I only met Bernanke once, as I recall it was in the 1990s.  He presented a paper at Bentley on the role of bank failures in the Great Depression.  I only asked one question: “How much would output have declined if the Fed had successfully stabilized NGDP, but the banking panics had still occurred.”  I don’t recall the exact answer, my sense is that he said output would have declined, but obviously much less than it actually did.  Bernanke certainly thought the monetary channel was very important, and saw his credit channel paper as supplementing the Friedman and Schwartz explanation.

I’ve always been skeptical about the credit channel.  Yes, it certainly matters to some extent, but given the decline in NGDP, and the bad supply-side policies, we got roughly the Depression I would have expected. So I don’t think the bank failures add much, except that they obviously help explain why money was so tight. After all, under the gold standard the hoarding of currency and reserves is even more harmful than under fiat money, as the Fed cannot offset as easily.

In the current recession a policy of NGDP targeting would have prevented a steep rise in unemployment, even if lots of banks had failed.  Of course, if we had been doing NGDP targeting, the number of bank failures would have been only a small fraction of what actually occurred.  Most of the failures were linked to loans to developers than went bad when the economy tanked, not the subprime mortgage mess.

Mark Sadowski sent me a study by Jeffrey Miron and Natalia Rigol that shows that much of Bernanke’s results hinge on a single month’s data, March 1933.  And that’s a very odd month to draw any implications from, as not only was there an extraordinary number of bank failures, but there was also a national bank holiday. So the entire banking system was shut down for much of the month.

We reconsider this issue by reporting regressions that drop March, 1933 from the sample entirely (along with appropriate lags). Columns (5)-(8) of Table 2 show the results. In this specification, the bank and business failure variables still enter negatively, consistent with Bernanke’s hypothesis, but the 7 coefficients are no longer statistically significant. Thus, exclusion of the Bank Holiday does not reverse Bernanke’s results but it weakens them substantially.

.  .  .

To the extent U.S. experience during the Great Depression – and especially the view that bank failures play a significant, independent role during that period – formed the intellectual foundation for Treasury and Fed actions, however, our results suggest at least a hint of caution. If the Great Depression does not constitute evidence for Too-Big-to-Fail, then what historical episodes do provide that evidence? We leave that question for another day.

In a world of no Too-Big-to-Fail, no FDIC, and NGDPLT, the optimal bank regulatory regime is no regulation at all.

Of course we don’t live in that world.  What’s the optimal regulatory regime in a world with TBTF, with FDIC, and without NGDPLT?  How the hell would I know?

The good news is that if we ever get to NGDPLT, we won’t need TBTF and FDIC.

Several commenters asked me why the liquidity trap hypothesis implies that countries are unable to devalue their currencies in the forex markets.  There are several ways of answering this.  A liquidity trap implies the central bank cannot inflate, but currency devaluation tends to raise the price level.

But that doesn’t really answer the question.  It seems obvious that countries can devalue, why do the Keynesians think this becomes impossible at the zero bound?

Here’s why people are confused.  We’ve been taught that “liquidity traps” are all about the zero lower bound on nominal interest rates.  That does play a role in the hypothesis, but on closer inspection it’s actually another zero bound that is crucial, the zero lower bound on eligible assets not purchased by the central bank.

Think about the common reply to liquidity trap worries; “The central bank could buy all the assets on planet Earth–surely that would depreciate the currency!”  Yes, but then Keynesians raise practical objections:

1. The central bank can only legally buy certain assets.

2.  The central bank may be fearful of having a large balance sheet.

Those objections then become the real reason for monetary policy ineffectiveness, not the zero bound. Some argue that monetary policy crosses over into being fiscal policy in this area.  I disagree; the EMH implies the expected gains and losses on purchase of foreign government bonds is roughly zero.  Even unconventional OMOs simply do not have the fiscal implications (futures tax liabilities) of government spending.

Now that we understand that the real problem is not enough eligible assets, or unwillingness to expand the balance sheet, it becomes clear why some countries (in their view) cannot devalue.  In order to peg the exchange rate at a lower level they’d have to sell so much domestic currency that their balance sheet would swell to unacceptable levels.  So there is a theory there.

Evan Soltas once showed that as soon as the Swiss National Bank finally bit the bullet, devalued, and then set a firm peg of 1.2/euro, they actually needed to buy fewer assets than before.  So although the worry about balance sheets might provide some sort of theoretical justification for liquidity traps, as a practical matter it’s simply not an issue.  The real problems lie elsewhere–refusal to set a robust enough NGDP target, and do level targeting.

PS.  The exchange rate does help in one respect.  It forces people out of the horrible Keynesian/Woodfordian “rental cost of money” approach to monetary policy, and into a much more enlightening Fisher/Warren “price of money” approach.  By doing so it allows us to think about the problems much more clearly, and suddenly some of those Keynesian interest rate-oriented concerns seem to fade away.

I can see from comments that people are very confused about the role of the medium of account.  I am going to address two misconception in the post.  First I will explain exactly the sense in which the MOA is more central to monetary economics than the medium of exchange, and which sense it is not.  Then I will explain why nominal price levels differ between countries.

Obviously it’s going to be impossible to explain the difference between a MOA and a MOE if the same asset serves both purposes.  We’d end up with an “angels on the head of a pin” debate.  Who could say why money is important if it serves both roles?  At a minimum we need a MOE than is not a MOA.  So let’s assume that a gram of gold is the MOA and a gram of silver is the MOE.  Prices (and wages) are denominated in terms of gold grams, but bills at the cash register are paid in silver coins. They have those modern cash registers with built in computers. The exchange rate is set each morning based on conditions in the gold and silver markets.  An easy way to envision this process is to assume some countries use gold and some use silver. Then the store just looks up the foreign exchange rate each morning and programs the cash register.

Now for my claim that the MOA is more “fundamental” than the MOE.  What do I mean by this?  I mean one thing and one thing only.  Here goes:

A discovery of a new process for easily turning lead into gold would have massive inflationary implications for the economy.  NGDP would soars and debtors would gain while creditors lost.

A discovery of a new process for easily turning lead into silver would reduce the nominal price of silver, but otherwise have no important implications for the economy.

That’s the sense (and the only sense) in which I think the market for the MOA is “interesting” and the market for the MOE is “uninteresting.”  Now I am quite aware of that fact that in the gold alchemy case, the value of the existing stock of (silver) MOE measured in terms of the MOA does soar, so one can also argue that the MOE is important for that reason.  I understand that argument, but simply don’t find it persuasive.  It still seems to me that gold is the dog and silver is the tail.

Now on to international price level differences.  Imagine that Austria uses grams of gold as their MOA, and Argentina uses grams of silver as their MOA.  A visitor from Austria notices that prices in Buenos Aires seem 40 times higher than in Vienna, or at least the price tags on goods in stores that would say “2 grams” in Vienna say “80 grams” in Buenos Aires.  So the “nominal price” (nominal means number) is 40 times higher in Argentina.  Real prices are fairly similar, perhaps slightly lower in Argentina.  The real ratio of price levels is also called the “real exchange rate.”

The Austrian visitor wants to know why nominal prices are 40 times higher in Buenos Aires.  Do you explain the difference with reference to:

1.  The path of nominal interest rates in each country over time.

2.  The path of real interest rates over time.

3.  The path of market interest rates minus the Wicksellian interest rate over time.

4.  Fiscal policy in each country.

5.  Minimum wage laws.

6.  The difference in MOA.

I say answer 6 is not just the right answer, it’s the only non-insane answer.  The market for the MOA determines each price level.  Period, end of story.

And logically, if the market for the MOA determines what sort of numbers you see on price tags in 2013, it will also explain what sort of numbers you see on price tags in the year 2023.  And that means (purely as a matter of logic), that it explain the inflation rate between 2013 and 2023.

Questions?

PS.  In 1900 Japanese prices were only modestly higher than US prices (nominally), whereas now they are roughly 100 times higher.  How did this happen?  The Keynesian model can’t really tell you.  If you instead focus on the MOA in each country (let’s say currency printed by each government), it’s easy to understand why Japanese prices are now 100 times higher than US prices.  These price differences don’t just happen, there are reasons.  And don’t say the reason is “the exchange rate.”  The exchange rate is simply another price, it begs the questions.  It’s the market for the MOA that determines nominal values.