Sunday, October 12, 2014

The market monetarist smell test

I gave myself a quick whiff this week to determine if I pass the market monetarist smell test. This is by no means definitive, nor is this an officially administered MM® test.

To be clear, my preferred policy end point is market choice in centralized banking. In other words, you, me, and my grandma should be able to start up a central bank. But that's a post for another day. First-best option aside, here's my reading of a few market monetarist ideas.

Target the forecast

**** 5 stars

Big fan. Targeting the forecast would take away the ad hoccery and mystique that surrounds central banks. We want central bankers to be passive managers of yawn-inducing utilities, not all-stars who make front covers of magazines.

First, have the central bank set a clear target x. This is the number that the central bank is mandated to hit in the course of manipulating its various levers, buttons, and pulleys. Modern central banks sorta set targets—they reserve the right to be flexible. Bu this isn't good enough. To target the forecast, you need a really clear signal, not something vague.

Next, have the central bank create a market that bets on x. Either that, or have it ride coattails on a market that already trades in x. If the market's forecast for x deviates from the central bank's target, the central bank needs to pull whatever levers and pulleys are necessary to drive the market forecast back to target.

The advantage of targeting the market forecast is that the tasks of information processing and decision making are outsourced to those better suited for the task: market participants. Gone would be whatever department at the central bank whose task is to fret over incoming data to determine if the bank is on an appropriate trajectory to hit x. Gone too would be the functionaries whose job it is to carefully wordsmith policy statements. The job of Fed-watching—the agonizing process of divining the truth of those policy statements—would disappear, just like lift operators and bowling alley pinsetters have all gone on to greener pastures. Things would be much simpler. If the market bets that the central bank is doing too little, its forecast will undershoot the target and the central bank will have to loosen. Vice versa if the market thinks the central bank is doing too much.

Targeting the forecast is the "market" in market monetarism. It's elegant, workable, and efficient—let's do it.

NGDP targeting

*** 3 stars

Meh, why not?

If we're going to target the forecast, we need a number for the market to bet on. Using the same target that central banks currently use is tricky. Most central banks are dirty inflation targeters. They try to keep the rate of change in consumer prices on target, but reserve the right to be flexible. Central banks have been willing to tolerate a little more inflation than their official target, especially if in doing so they believe that they can add some juice to a slowing in the real economy. Alternatively, they may choose to undershoot their inflation target for a while if they want to put a break on excessively strong output growth.

An NGDP target may be a good enough approximation of a flexible inflation target. NGDP is real GDP multiplied by the price level. If a target of, say, 4% NGDP growth is chosen, and the real economy is growing at 3%, then the central bank will only need to create 1% inflation. But if output is stagnating at 0.5%, then it will create 3.5% inflation.

So NGDP targeting affords the same sort of flexible tradeoff between the price level and real output that dirty inflation targeting affords, while serving as a precise number for markets to bet on.

The quantity of base money

* 1 star

Market monetarists have a fixation on the quantity of base money. This is where the monetarism in market monetarism comes from. Specifically, market monetarists seem to think that a central bank's policy instrument is, or should be, the quantity of base money. The policy instrument is the lever that the Carneys and Draghis and Yellens of the world manipulate to get the market to adjust the economy's price level.

But modern central banks almost all pay interest on central bank deposits. The quantity of money has effectively ceased to be a key policy instrument. (The Fed was late, making the switch in 2008). Shifting the interest rate channel (the gap between the interest rate that the central bank pays on deposits versus the rate that it extracts on loans) either higher or lower has become the main way to get prices to adjust.

This doesn't mean that the base isn't important. Rather, the return on the base is the central bank's policy instrument—it always has been. This is a big umbrella way of thinking about the policy instrument, since the return incorporates both the interest rate paid on deposits and the quantity of money as subcomponents. Reducing the return creates inflation, increasing it creates deflation.

Market monetarists seem to think that the interest rate channel ceases to be a good lever once interest rates are at 0%. But this isn't the case. It's very easy for central banks to reduce the return on deposits by imposing deposit rates to -0.5% or -1.0%. Going lower, say to -3%, poses some problems since everyone will try to immediately convert negative yielding central bank deposits into 0% cash. But if a central bank imposes a deposit fee on cash, a plan Miles Kimball describes more explicitly here, or withdraws high face value notes so that only ungainly low value notes remains, which I discuss here, there's no reason it can't drop rates much further than that.

If anything, it's the contribution of quantities to the base's total return that eventually goes mute. In manipulating the quantity of central bank deposits, central banks force investors to adjust the marginal value of the non-pecuniary component of the next deposit. Think of this non-pecuniary component as package of liquidity benefits that imbue a deposit with a narrow premium in and above its fundamental value. Increasing the quantity of central bank deposits results in a shrinking of this premium, thereby pushing their value lower and prices higher, while decreasing the quantity of deposits achieves the opposite. At the extreme, the quantity of deposits can be increased to the point at which the marginal liquidity value hits zero and the premium disappears, at which point further issuance of central bank deposits has no effect on prices. Deposits have hit rock bottom fundamental value.

So in sum: yes to targeting the forecast, and I suppose that an NGDP target seems like a good enough way to achieve the latter, and to hit it let's just keep using rates, not quantities. Does this make me a market monetarist?

Of course there's more to market monetarism than that, not all of which I claim to understand, but this post is already too long. Nor am I wedded to my views—feel free to convince me that I'm deranged in the comments.

Incidentally, if you haven't heard, Scott Sumner is trying to launch an NGDP prediction market.


  1. Part 1
    Comment in multiple parts due to character limit)

    I think the argument for NGDP (actually Scott's favorite is NGDP per capita) is more principled than you're suggesting. I'll try to elaborate, but I'm not an economist by training so please excuse me if my language is imprecise.

    Obviously it's a little icky to have central planning of monetary policy rather than some sort of market-oriented solution, but we're sort of stuck with central banking until we can figure out how to make free banking work. But there does seem to be wide agreement on the objective, which is that monetary policy should seek to strike a balance in the supply and demand for money which is neutral, and we'll loosely define neutral as meaning that policy should not provide shocks to that balance which have real effects, or at least, not effects that disturb pricing or contractual equilibria elsewhere in the economy.

    But what policy would attain that goal? People tend to bring up a couple areas where the value of money is non-neutral and has real effects:

    1. The price level prices are sticky, so monetary shocks cause inefficiencies in pricing.

    2. The wage level wages are sticky downward, so monetary shocks cause inefficiencies in compensation. This can be thought of as part of (1), but in practice they are often broken apart.

    3. Debt burden contracts are usually in the unit of account, so changes in the supply and demand for the unit of account benefit either the creditor or the debtor, and can sometimes result in default.

    I'm sure there are more, but these are important and often cited.

    1. Part 2
      Comment in multiple parts due to character limit)

      Let's start with (1), which is the focus of inflation targeting. In theory prices should adjust to any change in the purchasing power of money, but there is friction that prevents this (often referred to as price stickiness), so it seems like changes in purchasing power are bad. But aside from philosophical difficulties (purchasing power is not well-defined; are computers now 10x cheaper than 10 years ago because they've gotten more powerful?), it is arguable that there are cases where the purchasing power of money *should* change. For example, in the case of a general supply shock, we actually *do* want to reduce the purchasing power of the medium of account. So the central bank should only target inflation induced by demand shocks. But now we end up with a circularity: we only care about price level changes that are caused by demand shocks for the medium of account, and not supply shocks, but what does it mean to have a demand shock?

      Scott argues that for this reason, the price level is a secondary indicator of the demand for the medium of account, not a primary one. Demand for the medium of account is literally just the amount exchanged, or NGDP. So NGDP is actually the thing we want to target to deal with (1).

      Moving on to (2), NGDP literally is the sum of all income, so targeting NGDP will trivially keep (2) stable.

      Finally for, issue (3), NGDP is also a manifestly good solution. Under NGDP targeting, we ensure that contracts can be fulfilled, because all contracts become a percent of the total value of the goods and services provided that year, rather than a fixed amount of goods and services. If I can count on there being $X circulating in the economy in 2100, I can reasonably approximate the difficulty of providing some fraction of X in that year. Even if asteroids obliterate our civilization in intervening years, if the central bank keeps its promise to provide a certain level of NGDP at that point in the future, it will still be possible to provide that sum, whereas under CPI targeting, I may be promising more than the total value of NGDP in existence that year. A much less extreme version of this problem occurred in 2008, when a shortfall in NGDP made many contracts untenable, and defaults flourished.

      Just food for thought. Again, intuitively this seems like a supply and demand problem that should be solved by markets, but under a central banking regime we're stuck centrally planning it.

      BTW, when Scott started blogging I think his #1 ask was to stop the IOR policy, or even institute negative IOR, as you suggest. He switched to supporting QE later when that idea seemed to be more tractable among policymakers. I don't think [sophisticated] market monetarists are wedded to a particular monetary policy mechanism, except to insist that there is no ZLB.

    2. Plenty of good points.

      Don't the current flexible inflation targeting regimes (which include an output gap in their target) do pretty much the same thing as NGDP targeting in terms of taking care of 1, 2, and 3?

    3. They did until 2008 :-).

  2. There are 1,001 different possible monetary policy "instruments", that could be used to hit an NGDP (or whatever) target. The quantity of base money is one, the price of gold is a second, the stock market index is a third, the price of land is a fourth, etc. A nominal interest rate is just one of many.

    These "instruments" are not truly instruments. They are communications languages.

    There is absolutely nothing special about using an interest rate instrument. Plus, it has the wrong units (1/years); it does not have $ in the units. That means that if you set the interest rate wrong, and leave it there, the economy will eventually implode or explode. It is an unstable choice of instrument. Plus, it has the ZLB problem, so it fails you as a language when you hit the ZLB, and it fails you just when you need it most.

    "At the extreme, the quantity of deposits can be increased to the point at which the marginal liquidity value hits zero and the premium disappears, at which point further issuance of central bank deposits has no effect on prices. Deposits have hit rock bottom fundamental value."

    Sure, when the central bank has bought everything.

    1. Don't you think Miles Kimball's plan pretty much takes care of the ZLB problem?

      "Sure, when the central bank has bought everything."

      The premium disappears a lot earlier than that. Once the overnight rate has fallen to the deposit rate, the marginal liquidity value has hit zero.

      I don't understand why wrong interest rates cause the economy to implode. Too low of a interest rate on central bank deposits causes the price level to rise to a point at which people expect the purchasing power of deposits to increase at a rate fast enough to bring the return on money back in line with other assets. It stabilizes itself.

    2. JP: "The premium disappears a lot earlier than that. Once the overnight rate has fallen to the deposit rate, the marginal liquidity value has hit zero. "

      Line up all the non-money assets, from the most liquid on the left to the least liquid on the right. The central bank starts buying at the most liquid end, and keeps buying till it gets to the other end. The marginal liquidity value never hits zero until it has bought all the assets.

      "It stabilizes itself."

      Suppose, for example, the central bank suspends gold convertibility, but people expect gold convertibility to be restored, in 2525, at the *original* price of gold. Then what you say would work. Deflation causes expected inflation; and inflation causes expected deflation. In both cases the real interest rate can adjust to equilibrium, even if the nominal rate is fixed. But drop that assumption, and it won't work.

      Miles' plan is equivalent to raising the inflation target when you hit the ZLB.

    3. Nick, I see central bank monetary policy as the manipulation of the marginal liquidity value of the bank's own liabilities, not the marginal liquidity values of other assets. Once that margin has been reduced to zero the central bank has depleted all its gunpowder.

      I'm still not sure on why my interest rate example doesn't result in a stable outcome. Say the risk free rate of return is 4% and the market expects flat prices, but then Gideon Gono reduces the return on central bank liabilities (the interest rate + liquidity return) from 4% to 2%. The price level will rise to a high enough point that the market expects deflation of 2%. At this point central bank liabilities provide the same original 2% return plus a 2% expected capital gain, so that their total return is 4%, once again in line with the economy wide rate of return.

  3. JP: central bank monetary policy can be seen as the manipulation of the marginal liquidity value of the central bank's own liabilities **relative to the marginal liquidity values of other assets that are owned by the public**.

    Suppose that expected inflation adjusts adaptively (and positively) to past actual inflation. Then it is unstable. You need to assume that if actual inflation increases this causes expected inflation to decrease, to get stability. We know from the 1970's that expected inflation increased when actual inflation increased. It took some serious effort to bring expected inflation back down.

    1. Sure, but there's a point at which the relative marginal liquidity values can't change anymore and we're just swapping assets with equal liquidity values. We can assume the marginal liquidity of private assets is close to zero for private assets, given competition. Once a central bank has issued enough deposits to drive their marginal liquidity value to 0, then there's nothing left to wring out. And that happens pretty quick, once the rental value of central bank deposits is 0.

      Gotta go cook the bird. Will give some more thought.

    2. "You need to assume that if actual inflation increases this causes expected inflation to decrease, to get stability. We know from the 1970's that expected inflation increased when actual inflation increased. It took some serious effort to bring expected inflation back down."

      Fair enough, but why is this specific to interest rates? If you set the quantity of central bank deposits above the demand for deposits, the same process should ensue. The way I see it, setting the quantity of deposits above their demand or rates below the market rate really amount to the same thing -- in either case you've reduced the total return on deposits below the market rate.

  4. market monetarism is a conspiracy theory akin to the "the drug companies have a cure for x but the disease is too profitable" or "the oil companies are preventing the deployment of new energy technologies".
    with the addition of the idea that children grow because we buy them bigger clothes.
    with a sprinkle of buying everything on earth.

  5. Market Monetarism depends on a machine that tells of the future like in the movie Paycheck, starring Ben Affleck.