Wednesday, September 29, 2010

Yeah, the Fed Did It.

(More precisely, the Fed aggravated it; the Fed could have prevented most of it; the extent to which we depend on the Fed's good judgment should shrink, not grow.)

I believe we've been having a three-component economic crisis, each component building on and worse than the one before it, with overambitious or overconfident regulators/legislators making each component far worse than it needed to be. The three components I see are

  • the trend-following housing (and financial services) bubble, which the Fed worsened slightly by false reassurances;
  • the security-seeking, trend-breaking cash crunch which the Fed worsened greatly by too-tight monetary policy;
  • the longer-run tech-based employment recalculation. (All right, the Fed is not guilty here but it mostly hasn't happened yet, and our sensitivity to the mistakes the Fed made this time is growing with time. Cheer up, the worst is yet to come.)

bubble: I've commented on the housing bubble before, and how I think it was worsened by regulators and legislators (and raters) who denied the problem. Investor irrationality was real, but part of that irrationality was the willingness of investors to trust pronouncements by Greenspan and Bernanke, by Barney Frank and others on both sides of the aisle, and of course their willingness to believe that AAA meant "safe". I'm not arguing that interest rates were or weren't too low. (I do not believe that was the problem.) I'm not saying that the regulators were (or are, or will be) stupid or malevolent. I am saying that they were, quite obviously, wrong, and that those who relied on their assurances did very badly. (Those who simply said "prices are rising, I'll bet everything I can borrow that the trend will continue" did exactly as badly; there are always some of those.) I'd fix that (following Arnold Kling) mainly by going back to a world of high down payments. You could still give 100% financing if you wanted, but any federal support (including FDIC guarantees for a bank that offers mortgages) should depend on at least 20% down payment. Leverage would shrink generally, and underwater mortgages would be extremely rare. This would reduce homeownership rates, of course, and that may be regrettable, but it's not obvious that people are helped by encouraging them to make commitments they are likely to break, or be broken by. Of course this version of Kling's reform won't happen; what we're getting instead is expanded trust in that which failed before, to which we add taxpayers having to guarantee more than 95% of mortages, still being pushed on those who can't afford them.

cash crunch: As I've said before, I've become a semi-Sumnerite:

the real problem right now is not a “real” problem. The real problem is a nominal problem. When the growth rate of nominal GDP falls sharply there is always a severe recession. We have a severe nominal shock, a problem which has been understood by economists at least as far back as Hume. At the time, it always looks like the “real problem” was some symptom of the monetary shock, such as financial panic. Thus in the 1930s people thought the collapsing financial system caused the Great Depression, only later did we discover it was too little money.
Investors' efforts to minimize individual risk ended up adding to systemic risk. Actually it seems to me that we knew by February 2008, when Roubini said,
"Cash is king in 2008,"... the U.S. went into recession in December and will stay there for at least a year.
The Federal Reserve under Bernanke ignored what Bernanke had written academically; it brought down interest rates and then declared a "liquidity trap". It did expand the monetary base, but not nearly enough to satisfy demand -- and they neutralized part of their monetary expansion in fall 2008, by paying interest on excess reserves, encouraging hoarding by banks. Cash remained king, mostly because people were worried about too much risk in their portfolios. Here I would agree with Sumner that we should target NGDP (nominal GDP, aggregate cash flow) but I worry that buying Treasury bonds with cash, exchanging one low-risk item for another, might not succeed; we need to cope with people trying to shed risk. The Fed's purchases of mortgage-backed securities seems like a really bad idea: this is not absorbing risk in the sense of variability, it's buying a bet that already failed and attempting to prop up a market that should go downwards because there are too many houses out there for a while. So,
  • I'd make NGDP measures tradable in the form of Shiller's trills, creating a permanent market growing to perhaps a billion trills, paying one-tenth of one percent of our GDP, owned by citizens or foreigners but not by our own government.
  • Like Sumner, I would announce that we're targeting a 5% growth trend in trill yield (i.e., in NGDP), based on the pre-2008 trend so that if it rises too fast or too slow in one year we compensate the next; this is "level targeting".
  • I would give the Fed a stock-bonds-cash portfolio to be rebalanced daily, where the cash can be effectively imaginary (set it at last year's NGDP, most will never be printed) and all stocks are treated equally via a Wilshire Index fund; this rebalancing portfolio is the key difference between me and everybody else, hence probably totally wrong, but it makes sense to me. If investors starts selling stocks, the Fed will automatically buy, or sell if everyone else is buying, so this couple-of-trillion portfolio would automatically tend to stabilize the market. It would probably make money for taxpayers, too.
  • How would it stabilize the NGDP trend? When trills (next year's trills; buy them now!) start to fall, the Fed would change the portfolio proportions, giving cash for stocks and perhaps bonds, absorbing risk and satisfying the demand for cash. When trills start to rise above the price level target, the Fed portfolio proportions would change back.
  • Actually, I might make this last item more indirect: I might start a prediction market on the proportions required to achieve the actual NGDP target. In effect, I'd be giving knowledgeable parties something to bet on, so that they'd make money by getting it right. I don't want them able to make money by betting on the actual cash value of a trill's annual yield: that's (2008 yield)*(1.05^N), so the "right answer" is known in advance. Bet on the unknown path to that, instead. The Fed would use this prediction market to guide the proportions.
Instead of this, of course, we're giving the Fed a more complex mission as if its people had enhanced credibility. Since their credibility with me has gone way down, I don't find this reassuring.

employment recalculation: Kling talks about recalculation, reallocation of resources including labor in the constant search for "sustainable patterns of specialization and trade", and the unemployment this causes. Sumner acknowledges that some recalculation was required at the beginning, but mostly he just means the structural issues of too big a housing sector (and finance.) Delong and Krugman point to aggregate-demand-based unemployment and say that structural unemployment is on the way, but not yet a big deal. (Of course current unemployment is made worse by underwater mortgages which keep people from moving where the jobs are, and therefore by low-down-payment policies. And it's made directly worse by the cash crunch which motivates companies to sit on their cash, and it's made worse by regulatory uncertainty (and especially health care) and inflation uncertainty. But this is talking about aggregate demand v. structural, with recalculation as part of a slightly different story.)

I'd agree with them all, mostly, but add that recalculation is growing as an issue in a way they haven't (to my knowledge) discussed. My feeling is that overall technological productivity will gradually become the biggest factor in continuing unemployment, in the sensitivity of unemployment rates to (failures in) NGDP trends. I think that our increasing wealth and productivity means that a sharply decreasing fraction of the population is generating stuff we actually need, and a less-sharply decreasing fraction of the population is generating stuff we think we need. When money-trends continue, this doesn't matter because people buy whatever they were planning to buy. When money-trends fail and people want to hide their money, only the essentials keep going and that's a shrinking part of the economy. In the long run, (almost?) all production of goods and services is optional. In the short-to-medium run it would be enough to have the Fed do its job, making sure money-trends continue so people are comfortable buying stuff they want, not just what they think they need. In the long run, we will also need a negative income tax.

My approach to this stuff would be even more drastic, and therefore more unlikely, than my approaches to the preceding problems. So I won't finish this part of this post.

Footnote, since this is stuff that wasn't part of the way I thought through 2008: Aggregate cash flow is NGDP, Nominal Gross Domestic Product, the sum of all the money we pay (or get paid) for all the goods and services we use (and produce). You can divide that by your best guess at an inflation multiplier to get "Real GDP", the theoretical "constant-dollar" value of all those goods and services, but your paycheck and mortgage payment and grocery bill are paid in actual nominal cash flowing around and around, keeping our individual financial plans going by fulfilling the promises that we need to make economics ("sustainable patterns of specialization and trade", as per Arnold Kling) work. If expected NGDP drops, then you're already in a recession. I didn't really follow this argument when Tyler Cowen first recommended Sumner's blog. In the end, it's not that complicated. Think of a zillion spreadsheets carrying business plans and personal plans forward a few years, each projecting current trends. Aggregate cash flow -- that's "Nominal GDP". NGDP. Money. Some of those spreadsheets, some of those plans, will fail and others do better than expected, but generally the aggregate cash flow rises each year as population goes up, as productivity goes up, and as inflation goes on. If it falls or rises a little away from the expected trend implicit in all those individual plans, we adapt. If it falls sharply below trend, then cash isn't going around as expected and plans start failing simply because cash isn't going around: businesses fail and it's not their fault. Things are broken. We have a recession, a bad one.

In fact, people act by plans and promises, betting on their projections, so we get a recession as soon as the expected NGDP growth fails so that people stop buying and employers stop hiring.

Q: That sort of sounds almost convincing. Very odd. But isn't the future causing the present here?

A: Gee, thank you. It's actually close to tautological: expected NGDP is the aggregate of expected cash flow, and your belief that you're no longer going to be able to buy the goods and services you expected to buy will immediately change your behavior, the recession hits as soon as you expect it. So it's your beliefs about the future causing your behavior in the present.

In the current case we had a small recession because a whole lot of investors had believed our regulators and legislators who downplayed the risks of the bubble. They -- the investors -- had believed in the AAA ratings. When they hit reality they bounced, and needed more cash.

Q: But is this the Fed's fault? I mean, apart from Greenspan and then Bernanke denying the bubble?

A: The Fed has a dual mandate: they are supposed to manage inflation and unemployment, by managing the money supply. I'm saying that I mostly believe Sumner: the Fed did expand money somewhat, but they could have avoided most of the pain we've felt if they'd done more. So yeah, it's their fault.

Q: Done more? Done what? Lowered interest rates below zero?

A: Well, first by not paying interest on (excess) reserves, which was and is contractionary. Second, by announcing an inflation target or better an NGDP level-targeting sequence. Third, by expanding their open market purchases; preferably by starting the kind of automatically daily-rebalancing portfolio I described above.

Q: I understand why paying interest on reserves is contractionary; why are they doing it?

A: I don't really understand, but I think it's simply a way to give the banks money so they don't fail, while pretending that it's not Main Street bailing out Wall Street. I'm getting very cynical in my old age.

Ryan Avent of the Economist said

It's getting ever more difficult to avoid concluding that the Fed's inflation target is not the 2% we'd all come to expect, but something much closer to zero. This obviously impacts economic behaviour. The Fed could potentially have a significant effect on conditions simply by letting markets know that it's not actually happy with the current inflation trajectory.
Recently (Sept 2010) Bernanke has said that, with good effect; let's hope he goes further. There's some evidence that it will happen, e.g. Calculated Risk's Fed's Lockhart: The Approaching Monetary Policy Decision Dilemma
I think a consensus is building for QE2 in early November.
But I don't trust Bernanke to follow through, or at least I don't trust the Fed he leads...and that's what it depends on.

Or then again (I hope), maybe not.

Update: I see Avent saying in The perils of prediction: Forget forecasts, trust markets | The Economist that

I like to point out that in June of 2008 the Federal Reserve forecast real GDP growth in 2009 of 2.0% to 2.8%, when in fact the economy shrank in 2009 by over 2%. Of course, this doesn't mean that central banks have no basis on which to make policy. All they need do is look at the evidence in front of them. Markets...
I trust markets a lot more than I trust the Fed.

Perhaps I should note that Sumner does not blame the Fed for failure to predict, as he said in TheMoneyIllusion » The Fed doesn’t have a crystal ball

All the major investment banks with their million dollar Ivy League employees missed this crisis (and its eventual impact), and yet the Fed was supposed to have predicted it? The Fed pays much lower salaries than Wall Street.
Indeed, I wouldn't blame the Fed for the housing bubble recession-trigger at all if Greenspan (and then Bernanke) had simply said "Bubble-detection is not part of my job, I can't help you with that." But this is not what I understood them to be saying.

update: Ah-ha! An actual reputable economist, Nick Rowe, says at least that

If I had my druthers, the Fed would buy stocks. Something like the S&P500 index.
This is not equivalent to saying that the Fed should do a large part of its monetary policy via a rebalancing portfolio somewhat similar to what investment people prescribe for individuals, but it's a start. Yay!

(Or then again, maybe not.)

upd: The same Nick Rowe is quoted approvingly by Brad Delong in Against Money-Financed Fiscal Expansion, For Open Market Operations in Equity Indexes

OK. Start with the Fed buying bridges. That will work. Now, wouldn't it be nice if the Fed could also sell those bridges again later, if it needs to, as it probably will. Bridges aren't very liquid. And, the Fed is good at clipping coupons on bonds, but perhaps not very experienced at collecting tolls on bridges. Hmmm. Maybe if the Fed just bought shares in bridges instead, that would be as good as bridges, but even better from the practical point of view. Hmmm. Why stop at bridges? Why not buy shares in everything? Why not just buy the Wilshire 5000, or some such index?
Excellent. The right index identified, along with the need for later sale; we are close to portfolio rebalancing.

Labels: , , , , , ,