"Monetary policy can cause bad things to happen in financial markets, which can cause bad things to happen in the rest of the economy. Therefore NGDP targeting is wrong."
I made up that quote. But if I had to summarise M.C.K.'s long article in two short sentences, that is how I would do it.
Here's a longer passage from M.C.K.:
"Banks and other financial intermediaries usually create credit whenever
they can earn what they believe is a risk-adjusted spread between their
funding costs and the rates they charge their borrowers, both of
which are affected, if not determined, directly by the monetary
authority. Tobias Adrian and Hyun Song Shin have shown
that the balance sheets of financial firms that mark their assets to
market grow and shrink based on changes to the level of short-term
interest rates. Meanwhile, Markus Brunnermeier and Yuliy Sannikov have shown that monetary policymakers can alter the willingness of banks to create credit by adjusting the shape of the yield curve."
Yes, if the central bank raises or lowers interest rates, this will affect financial markets. But I thought we had gone beyond thinking of monetary policy in terms of raising or lowering interest rates. Or buying or selling bonds in an open market operation. Or raising or lowering the money supply. Or raising or lowering the exchange rate. Those aren't monetary policies.
Targeting 2% inflation is a monetary policy. Keeping the money supply growing at 4% per year is a monetary policy. Keeping the exchange rate fixed at $0.95US is a monetary policy. Targeting "full employment" (at least, trying and failing) is a monetary policy. Following the Taylor Rule is a monetary policy. Targeting a 5% level-path for NGDP is a monetary policy.
We can debate the merits of those different monetary policies. Which one is best?
A monetary policy is not just the central bank doing something right now. A monetary policy is some sort of rule that tells the central bank the different things it should be doing under all sorts of different circumstances in the past, present, and future.
We have to think of monetary policy that way. First, because the effects of the central bank doing something right now, and whether those effects are good or bad, will depend on the circumstances. Second, and more importantly, what the central bank does right now isn't the only thing that matters. What it did in the past matters too. And what it is expected to do in the future matters even more.
Suppose I argued that inflation targeting was a bad policy, because targeting inflation might require the Bank of Canada to raise interest rates, which would cause the exchange rate to appreciate, which would cause Aggregate Demand to fall, which would cause a recession. A defender of inflation targeting would say "Hang on; wouldn't the Bank of canada only raise interest rates when AD was growing too strongly anyway? And if rasing interest rates would cause an exchange rate appreciation which would cause a fall in AD which would cause a recession, wouldn't it also cause inflation to fall below target? So why would the Bank of Canada do it, if it were targeting inflation?"
Inflation targeting doesn't mean ignoring the exchange rate. Of course the Bank of Canada will look at what is happening to the exchange rate when it chooses the overnight rate to keep inflation on target. The exchange rate is one of many indicators that give the Bank of Canada the information it needs to keep inflation on target.
If I were seriously worried about the possible bad consequences of exchange rate fluctuations, and if I thought there were a genuine conflict between keeping inflation on target and avoiding those bad consequences, I ought to propose a monetary policy like fixing the exchange rate. Fixed exchange rate vs inflation targeting; which one is best? That's a proper debate between two proper monetary policies.
Or maybe a hybrid policy would be better than both. Perhaps the central bank should target the sum of CPI inflation plus exchange rate depreciation?
It would be the same if I argued that targeting NGDP was a bad policy, because targeting NGDP might require the central bank to lower interest rates, which might cause asset prices to rise, which might have bad consequences.
If I were seriously worried about the possible bad consequences of financial market fluctuations, and if I thought there were a genuine
conflict between keeping NGDP on target and avoiding those bad
consequences, I ought to propose a monetary policy like (say) targeting asset prices. Targeting asset prices vs targeting NGDP; which one is
best? That's a proper debate between two proper monetary policies.
What would happen to the real economy if the central bank did whatever it takes to keep asset prices, or total credit, or whatever, growing at a steady (say) 5%? Can you think of any circumstances under which keeping financial markets stable might destabilise the rest of the economy? Which one matters most?
Unless you propose an alternative monetary policy target, we can't even begin the debate.
Worthwhile Canadian Initiative 
Nick, you have set up a straw man here. The relevant comparison is strict NGDPLT vs NGDPLT corridor, where the point in the corridor is chosen according to the financial stability considerations.
Nick:
We monetary theorists.
Why do we think that our concerns are central to central banking?
Suppose that what central banks care about is stablizing financial markets with these nominal anchor concerns being a constraint.
Why should be Fed be expected to focus on the macroeconomy? Do we expect the Department of Agriculture to worry about much more than farm prices and output?
123: Your NGDPLT “corridor” seems to me like hybrid between NGDPL targeting and asset price targeting. So again, why is it good to have that asset price component to monetary policy and what are the risks associated with it? That is what the debate should be about.
123: “straw man”? What do you mean “straw man”? That “straw man” is me, 20 years ago, when I used to argue that the Bank of Canada should target the TSE300 total return index!
But J.V. has the right answer. The corridor is a lexicographic hybrid, where the central bank targets asset prices as long as NGDP stays within the corridor, and NGDP otherwise. But it’s a strange sort of hybrid. Does NGDP not matter at all inside the corridor? Do asset prices not matter at all outside the corridor? Don’t we economists normally assume smooth trade-offs?
Bill: yep. These finance guys!
Monetary policy is all hogwash. The basic purpose of an economy is to produce what the people want, both as expressed by what they do with their cash and what types of public spending they vote for. Ergo if stimulus is needed, it’s household incomes that should be boosted (e.g. via Warren Mosler’s payroll tax reduction) and public spending that should be increased.
Why boost just investment or asset prices, which is what monetary policy does? You might as well boost an economy just via people with black hair, while blondes, red heads, etc wait for a trickle down effect.
Ralph:
1. OK, so they make you Prime Minister. What instructions do you give the Bank of England?
2. If monetary policy cannot affect the rest of the economy, how do you explain the fact that Canadian inflation has been almost exactly 2% for the last 20 years, ever since the BoC started targeting 2% inflation? Total coincidence?
3. Suppose some nutters took over the Bank of England, and decided they wanted to reduce the money supply by 90%/raise interest rates to 90%/cut the price level by 90%/destroy the UK economy/whatever. Would you shrug your shoulders?
Nick,
For me the main issue is what Stein said four days ago, thus the “straw man”.
Corridor is a practical solution when you have one instrument and two targets, the main target and a secondary one. NGDP is the main target, and we allow only small ex-ante deviations from our target. Financial stability is a secondary target that is mainly adressed with the other tools.
If you want a smooth trade-off approach – no problem. Some time ago I had a post where I argued for two instrument two target approach. Second target is the expected volatility of NGDP that is closely related to financial stability, second instrument – NGDP options. When there is a buildup of financial imbalances that the central bank fears, the central bank should buy insurance against the volatility of NGDP, thus raising the market estimate of NGDP volatility and slowing down the creation of new debt.
There are many ways a central bank can adjust the money supply to hit whatever target it chooses and each of these ways affect to some degree not only the money supply but also the distribution of income or wealth (that is: all monetary policy has some fiscal side effects that may or may not be intended by the central bank).
Adjusting the money supply via the CB buying/selling low-risk bonds in theory seems one of the most neutral (low fiscal side effects) ways of affecting the money supply. However combined with other govt policies (insurance schemes on bank accounts, promoting and subsidizing high-risk lending etc) and perhaps with human psychology, it can be shown to have toxic side effects if these polices encourage the money from assets purchases to go into further higher-risk purchases that can become bubbles.
For this reason I would favor discussion of other forms of adjusting the money supply (a tax/subsidy on all final sales for example using newly created money if necessary) that would have a less direct effect on asset prices than QE has.
Hi Nick,
I didn’t mean to suggest that monetary policy has no effect. My point was that it impinges initially on a narrow range of economic activities, and that is undesirable all else equal (in particular if the lags involved in monetary and fiscal policy are equal).
Re Canada’s good record on 2% inflation, the Bank of England has also aimed for 2% and not succeeded very well.
Re your third point, I don’t see why what I’m suggesting leads to violent changes in the money supply, prices, etc.
I think 123 is getting at what I wanted to say, but let me put it my way.
The impact of monetary policy on a macroeconomic target (let’s say NGDP) is not deterministic, and presumably a rational central bank would have preferences over the entire distribution of outcomes, not just over the mean (or the median, or whatever summary statistic it uses). For simplicity, let’s just say they care about the first and second moments. In the light of what happened in 2008 and 2009, it seems pretty clear that they should care about the second moment, particularly if they’re doing growth rate targeting, because in that case a big miss has a big persistent impact.
A properly stated argument for bubble-popping is mostly an argument about the second moment of the primary target: bubbles pop at uncertain times and the pop, if it comes late and of its own accord rather than being early and deliberately precipitated, can cause the target variable to be realized at a value far from the mean. For a central bank that doesn’t deliberately pop bubbles, this has two implications: (1) the second moment of the target will be too big, and (2) until the bubble pops, if the central bank is accurately targeting the first moment, the realizations will be too high because the as-yet-unobserved crash brings down the distribution. (The central bank’s “peso problem” if you will.)
Now the second moment is less of an issue if the central bank is targeting a level rather than a growth rate, but it still matters. You won’t get the permanent impact of a one-time miss, but you will get excessive volatility. A unintentional miss in one direction commits the central bank to an intentional compensating move in the other direction, and obviously we would all prefer that the path be smoother. So there is still reason to be concerned about the second moment.
There’s something to be said for the simplicity of targeting the first moment and ignoring the second, and I tend to favor such a policy, but it is a bit worrisome. Ideally I’d like to see the bubble problem addressed via a different instrument: for example, create an authority that can raise the natural interest rate (thus forcing a first-moment-targeting central bank to raise the actual interest rate) by declaring impromptu tax cuts and/or transfer payments to households that are likely to be liquidity constrained.
@Andy Harless
I fully agree ( including the part about the fiscal stimulus ).
On the other hand, when the financial bubble pops, it should be the responsibility of the central bank to prevent the second moment from exploding, if the central bank anticipates it can be done with expected profit.
I also think that level targeting will certainly help with the next crisis, but over the long term the Minsky cycle will reach higher levels where second moment will do lots of damage if unchecked despite level targeting.
Did you read my NGDP volatility blog post? If not, google “NGDP options” (with quotation marks).
The problem is we don’t know which target(s) would be best, though we have some idea which might be better than others under certain conditions, so while we must choose a target, we still need to monitor those other variables to see if they represent accumulating problems and potentially modify our target(s) and actions. (I do not consider asset prices one of these, it is not equities but debt that can be dangerous.)
Andy: I’ve been mulling over your comment.
An analogy: an econometrician might choose an estimator he knows is biased, if it has a lower variance than an alternative unbiased estimator?
Nick,
It’s not a problem in estimation; it’s a problem in feedback control. Should the control function be of the form F(s) or F(s, ds/dt)? This depends on the transfer function to be controlled and the desired output behavior. In electronics you might be faced with a tradeoff between rapid response and ringing or overshoot. Or perhaps a proposed solution has too little margin to be reliably stable.
But what I think is being missed here is that their are ordinary bubbles caused by irrational exuberance where popping is relatively painless, and large scale debt bubbles, where private debt is large compared to NGDP. Of course there’s no particular magic level of debt that will cause disaster, but the presence of the debt makes any policy of austerity extremely dangerous since it can cause the economy to head for a bad equilibrium.