Periodically, policymakers and market participants have speculated as to whether or not there is a bubble in Treasury securities. Here’s Sheila Blair yesterday:

The Fed has maintained interest rates at or near zero for four years running, even though the financial system has been relatively stable since 2009. The Fed’s actions have kept Treasury bond prices high (while keeping the government’s interest costs low), but the fundamentals do not support the high valuations, given the fiscal mess we are in. Sooner or later, the bond bubble will burst. History has shown that a structurally weak economy combined with a fiscally irresponsible government propped up by accommodative central-bank lending always ends badly. Absent a change in policies, a toxic brew of volatile interest rates and uncontrollable inflation could define our future.

This is typically derided by left-wing critics of the Bernanke Fed as hard-money fetishism. Unemployment is still high. Inflation is still low. Clearly, what we need is not for the Treasury bubble to burst, but rather an even more accommodative central bank, a burst of catch-up nominal growth fueled by inflation.
I would like to argue that, in fact, both the hard-money and the soft-money perspectives on our current monetary situation have a part of the truth. I am increasingly convinced that, to the extent that monetary policy is implicated in our current mess, what we’re seeing is precisely the limitation of unconventional monetary policy.

Do make my argument, I’m going to wade into last week’s debate between Ryan Avent and Matt Yglesias about whether or not eliminating cash would improve monetary policy at the zero bound. But first I need to recap the debate.

Right now, when the Fed thinks we are entering a recession, it lowers rates to increase the money supply in response to elevated demand for money (equivalent to lower demand for goods and services), and when inflation begins to pick up the Fed raises rates to reduce the money supply in response to lower demand for money (equivalent to higher demand for goods and services). If they do their job well, the economy should hum along a more stable trajectory than would be the case if the money supply were fixed. That’s the theory, anyway, and in the 1950s-1960s and 1980s-1990s the theory seemed to work well in practice.

But if inflation is stable and low (which is what we want), then interest rates will also generally be rather low, which means that if we tip into recession there is very little room to cut rates before they reach zero. This is the “zero bound” problem. Once rates hit zero, the central bank can only increase the money supply through unconventional means – buying longer-term bonds, for example.

Yglesias argued several months ago that eliminating cash would eliminate the zero-bound problem:

Stop for a moment and ask yourself whythe interest rate can’t be reduced much below 1 percent. The trouble is cash. At any given time, relatively little paper currency circulates in the United States. Instead, most of the American money supply consists of bank accounts and other electronic stores of value. People prefer to keep money in bank accounts because it’s convenient and because you get interest on it. If the rates were driven below zero—in effect a tax on holding cash in the bank—people would just withdraw money and store it in shoeboxes instead. But what if you couldn’twithdraw cash? What if all transactions were electronic, so the only way to avoid keeping money in a negative-rate account was to go out and buy something with the money? Well, then, we would have solved our depression problem. Too much unemployment? Lower interest rates below zero, Americans will start spending and investing again, the economic will grow, and unemployment will go back down to its “natural rate.”

To which Avent responded last week:

[I]nflation and negative interest rates are basically the same. Both take an amount of money in the possession of an individual and erode its purchasing power over time. Arithmetically, higher inflation reduces the real cost of borrowing, generating, at the zero lower bound, that all important negative interest rate. You can target a rate of inflation in order to generate precisely the desired interest rate, thereby encouraging households to spend, borrow, and invest in a manner consistent with full employment.

Now, Mr Yglesias might have argued here that the central bank is incapable of creating a higher level of inflation—that the problem remains a technical one—but he doesn’t. Rightly so, in my view; a central bank that can create an unlimited amount of money and engage in open-market operations should have no trouble creating inflation. Rather, he argues that higher inflation is a “bizarre and unpalatable proposal…for the economic and political elite”. But in what way is higher inflation different in its impact on the elite than negative interest rates? One could argue that political backlash is more likely with negative rates; the government has at least a veneer of protection from anger over higher inflation in the form of money illusion. With negative rates, the dynamic would be explicit: the value of the money you have in your saving account is getting smaller, and that is a direct result of government policies. Maybe the citizenry will pull its money out and gleefully go a-spending. Or voters will arm themselves and install Ron Paul as supreme leader.

 To which Yglesias responded:

It is true that there are lots of different ways the Fed can do. But during the Great Moderation the thing the Fed did do was stabilize the macroeconomy by cutting interest rates. Everyone anticipated the Fed’s behavior to follow a Taylor-type rule in which inflation and unemployment data determined interest rates. You would read paradoxical-sounding stories about the stock market jumping on disappointing jobs data, precisely because everyone felt they understood how everything worked. The problem with the zero lower bound then becomes that as rates got closer and closer to zero nobody knew what was going to happen. . .

But if there were no logistical barrier to going negative, people could just assume that the Fed would keep using its tool of choice and its rule of choice and do what the rule said. No muss, no fuss. It would feel a bit funny the first time it happened, but since the economy would have recovered faster the Fed would have gone back to raising rates faster and ultimately there’d be less discontent than there is with the status quo.

Yglesias and Avent have wandered into a cul-de-sac, I think, in speculating about what would be more or less politically difficult – targeting inflation or simply cutting rates – while skating breezily over what is technically more difficult. The fact is, the political difficulty in pursuing unconventional monetary policy is closely related to its technical difficulty – or, rather, to the fact that we don’t know whether unconventional monetary policy works predictably well.

The Fed normally operates by manipulating the overnight rate not simply because of tradition, but because it is a relatively simple instrument. It is trying to manipulate the most money-like instruments in the market – because it is responding to changes in the demand for money (over which the government has a monopoly). When it starts behaving unconventionally, buying longer-term government bonds or even bonds not issued by government (or, in the extreme, even stocks or real assets), it can be less and less certain of the precise consequences of its effects – what the magnitude of the impact will be on inflation expectations, but also what other follow-on consequences there will be for the markets and the economy.

Take a look at the bond market today. Longer-term bonds are relatively low-yielding. Because short rates can’t go below zero, you’ve got a very flat yield curve. A flat yield curve is, generally, an expression by the market of low expectations for long-term nominal growth. That’s not what the Fed wants, ultimately. It wants the market to expect higher nominal growth, because those expectations will drive investment. Normally, it would engineer an upward-sloping yield curve by cutting short rates. But because it can’t cut short rates any further, it’s intervening at the long end of the curve. It’s trying to push people out of “safe” long-term bonds and into riskier assets (or into spending), but in the process it’s engineering the flattening of the yield curve. Until they get the change in nominal growth expectations they are hoping for (whether because inflation expectations pick up or because real growth expectations pick up), they are, in fact, engineering precisely the bubble Blair is warning about in her column. That’s an unintended side effect of unconventional monetary policy.

Yglesias and Avent act as if everybody knows that the Fed can produce precisely the desired level of inflation by a simple act of will: target the level, and the market will expect that the target will be hit. Leaving aside questions of whether the Fed can measure inflation well enough to target it, the assertion would still only be true if the Fed credibly declared that it didn’t care about anything else except the inflation target. Which it cannot credibly declare. It’s just not conceivable that the Fed would ever be believed in stating that no other political or economic factor matters to it at all. Not even the ECB says that. If unconventional monetary policy has less-known side effects – or unknown side-effects – then policymakers are rationally going to be more cautious in applying it.

This caution, in turn, feeds the bubble phenomenon. A bubble takes time to build up. Expectations that certain conditions are going to be semi-permanent need to set in. That’s precisely what’s happened with respect to interest rates. The Fed has said that it is going to keep rates low indefinitely because expectations for growth remain low. So the market has every reason to believe that the curve will remain flat. Which means it can discount the risk of a sudden spike in rates from very low levels. The aggressive discounting of a downside risk is pretty much the definition of what a bubble looks like.

Now, imagine what would have happened if short rates could go below zero. First of all, the Fed could have responded much faster to the recession, and could have intensified its response as the depth of the recession became clear. And precisely because it could have responded faster, the recession would never have gotten as bad as it did. Negative nominal short rates would have constituted a bailout of the banking system that favored the strongest firms rather than the weakest ones – since banks borrow short and lend long, all banks would have seen profitability increase sharply – and the toxic assets on their balance sheets might therefore have more quickly found a clearing price. The structural problems of the American economy would remain. I still believe the natural rate of unemployment would have increased significantly, because rational expectations for real growth drop when you realize that much of the growth you just experienced was illusory and bubble-fueled. But we would not have overshot so much to the downside, and therefore we would not have had to play catch up. And, because we would not have had to play catch up, we would not now be dealing with worries about a new bubble at the same time that we are worried about terrifyingly high lingering unemployment.

Yglesias ends his most recent piece thusly:

I say all this, I note, not to argue that we need to scrap paper money. The point is that it’s very bad for the Fed to have a policy rule that breaks down in moments of severe crisis. It’s like having an umbrella that dissolves in water. We either need to run a background level of inflation that’s high enough to avoid zero bound episodes, or else shift the policy lever to something that’s not effected by these issues.

But a rule is only as good as the tools for applying it. A higher level of background inflation means a higher level of uncertainty about inflation – the primary reason why low inflation is favored over high inflation is precisely that the variance on low inflation is lower, and stable inflation expectations are conducive to higher real growth, which, in turn, is the driver of real increases in wealth. And shifting the policy lever given current tools means institutionalizing unconventional monetary policy – in spite of the fact that we have a higher degree of uncertainty about the side effects of using these unconventional tools. This, again, would increase uncertainty about the effectiveness of monetary policy.

It might still be a good idea, on balance, to have higher background inflation. It might still be a good idea, on balance, to institutionalize unconventional monetary policy. But there would be real costs associated with these choices. It behooves the soft-money crowd to reckon with those costs, just as it behooves the hard-money crowd to reckon with the costs of long-term unemployment (which it frequently doesn’t do).

By contrast, the costs of eliminating cash strike me as quite limited – and the upside quite significant. Pushing short rates below zero is not equivalent to pushing down long rates by means of unconventional policy. It’s not equivalent – and it’s superior. Better tools don’t come along that often. Yglesias shouldn’t back away from promoting one when one comes along.