There are a number of weaknesses to using “central planning” in the implementation of monetary policy. Today we see another example—slow decision-making. Whereas markets move at lightening speed, committees of bureaucrats tend to move quite slowly. In 2008, that slowness created a deep recession. I don’t expect that outcome this year, but policy has recently been falling “behind the curve”.
Take a look at this FT graph showing the probability of various policy paths in 2019:
Notice that as recently as March 2019, interest rate futures markets were expecting no change in rates during 2019. At the same time, other financial markets were showing a great deal of optimism. Stocks were near record highs and the labor market was extremely strong. Inflation had been running only slightly below the Fed 2% target. Monetary policy was roughly on course.
Just 2 months later, there are signs that the equilibrium interest rate has fallen sharply. The markets now expect a number of rate cuts this year, and yet despite that anticipated policy “easing”, expectations for growth and inflation seem to be falling.
A naive observer (i.e. 99.9% of pundits) would interpret this picture as follows. In March, there was no expectation that the Fed would ease policy in 2019. Two months later, markets believe that President Trump will succeed in getting the Fed to enact the monetary stimulus that he has been calling for.
In fact, almost the opposite is true. One should never interpret the stance of monetary policy from the level of interest rates. A cut in interest rates may indeed reflect easier money, if associated with faster expected NGDP growth. But it can also reflect tighter money if the central bank is responding too slowly to a fall in the equilibrium interest rate. In this case, the latter interpretation is more likely (albeit not certain), given the recent performance of the markets.
This is why we should never have given central banks the responsibility of targeting interest rates. It’s the original sin of modern central banking. Instead, central banks should target (market) NGDP growth expectations and let markets determine interest rates. Only markets can keep up with fast moving changes in the equilibrium interest rate—bureaucrats are simply not up to the job.
President Trump will likely get the lower rates he’s been asking for, which only goes to prove the old maxim “be careful what you wish for.”
PS. Let me explain the “albeit not certain”, which appeared in this post. All judgments about monetary policy are provisional. Markets provide the best guess based on current information, but will be wrong on many occasions. As new facts come in, market participants continually revise their forecasts. It’s entirely possible that the Fed will be right and the markets will be wrong regarding the path of rates in late 2019. But if I were a betting man . . .
READER COMMENTS
John Hall
May 31 2019 at 4:35pm
Kashkari was out and about saying he didn’t think it was time to cut rates yet. When even the most dovish person at the Fed is against cutting rates…
rodrigo
May 31 2019 at 5:29pm
Its scary to think how bad the market needs to get in order for the fed to react. I just wish the futures market would price the rate cut for July instead of September so we don’t have to suffer through the carnage.
I don’t expect that outcome this** year
Matthew Waters
May 31 2019 at 6:05pm
CBs should not make their ultimate target interest rates. They should ultimately target inflation, NGDP or gross income.
However, I am unsure what a system will look day-to-day without the Fed using some sort of overnight interest rate. The mechanical tools of IOR, discount window, repos and overdrafts all require the Fed to make some judgment as to the short-term interest rate.
Scott Sumner
May 31 2019 at 8:37pm
Matthew, That’s false. They can use ordinary open market operations to target NGDP futures prices. There is no need to pay IOR or make discount loans.
Matthew Waters
May 31 2019 at 9:26pm
In practice, pre-2008 monetary policy had very slight and gradual changes in the outright holdings. The outright SOMA account increased by around $25bil a year and the change was spread over the year.
Since almost no reserves existed, daylight overdrafts peaked at $200bil a day. The overnight fluctuations due to Treasury auctions or tax payments were handled almost entirely through repos. With both repos and overdrafts, the counterparty handled the trading and took the price risk first. Price risk only mattered in case the dealer or bank defaulted.
So it’s not absolutely, theoretically impossible to have minute-by-minute OMOs to handle reserve fluctuations. But IMO, it would have more frictional costs and be less efficient than either the pre-2008 or IOR systems.
Matthias Görgens
May 31 2019 at 11:45pm
Why would there be a need to have minute by minute OMO?
Even once a week would be plenty.
With stable ngdp expectations, the markets would stabilise themselves.
Travis Allison
May 31 2019 at 6:21pm
Scott,
You have talked about the Wicksellian natural rate of interest before, but do you find it a useful concept? It can’t be observed and even theoretically it would bounce around every day according to shocks, creating positive or negative feedback effects depending on where the Fed had pegged the Fed Funds rate. For me, NGDP expectations seem a more stable and useful paradigm.
The one exception (as an intuition pump) is that when the Fed cuts rates too slowly, the Wicksellian rate will fall even more. So the Fed is behind the curve. I have an intuition for that situation. Could you explain it to me in terms of NGDP expectations? Is it the case that NGDP expectations are based on the Fed cutting the Fed Funds by an average of x% and if the Fed cuts by less than x%, NGDP expectations are revised downward because NGDP expectations are based on a probability expectation of Fed easing by different amounts?
Scott Sumner
May 31 2019 at 8:40pm
Yes, the Wicksellian equilibrium rate is not a useful concept. I am forced to use the concept because the Fed targets interest rates and other pundits insist on describing monetary policy in terms of rates. Using the equilibrium rate I am able to explain why they are wrong.
When the Fed cuts rates more slowly than the equilibrium rate falls, then NGDP growth expectations decline.
Thomas Hutcheson
Jun 2 2019 at 6:33am
Why not just say NGDP is still below the trend level (mainly because of past failures to achieve symmetric inflation, in letting inflation fall below 2% for more quarters than it exceeded 2%) and so the Fed should cut rates, stop paying IOR, renew QE, buy foreign exchange assets, or use whatever instrument it finds handiest to achieve its dual mandate?
Scott Sumner
Jun 2 2019 at 1:08pm
Thomas, That’s fine too.
Don Geddis
May 31 2019 at 6:56pm
@Matthew Waters: “I am unsure what a system will look day-to-day without the Fed using some sort of overnight interest rate.”
It’s almost trivially simple: use only the money supply. If (expectations of) your variable (e.g. NGDP, or maybe inflation) are below target, then increase the (growth rate of) the money supply. If the variable is above target, then decrease the money supply.
Interest rates don’t matter. Ignore them.
“The mechanical tools of IOR, discount window, repos and overdrafts all require the Fed to make some judgment as to the short-term interest rate.”
The Fed should abandon IOR, and the discount window is not significant.
It is foolish to set a short-term interest rate target, at best, every six weeks. Interest rates ought to float day by day, hour by hour, perhaps minute by minute, according to current market conditions.
Currently the Fed takes action to maintain interest rate targets, and those intermediate targets only change slowly. The Fed ought to instead target the nominal variables of ultimate interest (such as NGDP) directly. Perform minute-by-minute OMOs in order to maintain stability in the actually important macroeconomic conditions.
This current framework of an every-six-week (at best!) update to an intermediate target (interest rates) is counterproductive. Instead, daily OMOs should be driven by real-time updates to the ultimate target.
Matthew Waters
Jun 1 2019 at 5:30pm
Don,
For context, see the detailed report from the NY Fed on OMOs in 2007. The overnight changes in reserve demand (such as Treasury auctions or April 15th) were entirely accommodated by repos, since these reserve demands were transitory. For daylight overdrafts, reserve demands varied by $100 to $200 billion every day.
By contrast, the permanent SOMA portfolio changed VERY gradually. There were perhaps $5 billion in outright purchases each month. Like any other broker relationship, the Fed had to pay primary dealers commissions or bid-ask spreads for executing the trades. The later QE accumulation was still relatively small on a daily basis compared to pre-IOR fluctuation in reserves.
https://www.newyorkfed.org/medialibrary/media/markets/omo/omo2007.pdf
I don’t want to defend the status quo too strongly. I would change a lot personally. But also, the “have NGDP futures and immediate OMOs” seems very out of step with the way policy has always actually worked at the Fed. The Fed having some overnight interest rate makes even drastic policy far easier.
For example, in a Sep 2008 zero bound scenario, the Fed can do an open tender for Treasuries and STRIPS at zero yield. That obviates the need to work through primary dealers. But the open tenders or repos of longer-term Treasuries require some short-term interest rate anchor.
Scott Sumner
Jun 2 2019 at 1:11pm
Matthew, There’s a difference between needed an interest rate to conduct a transaction and needing to target interest rates.
Benjamin Cole
May 31 2019 at 9:16pm
I am one in a thousand!
Scott Sumner says 99.9% of pundits are wrong in their analysis of the Fed and its actions and the direction of interest rates.
I am more of a wag than a pundit, but I agree with Scott Sumner in this case.
I am not sure I would call the labor market “extremely strong.” Wages haven’t really budged much, and real weekly median wages are not much different than they were in 1978 (that is as far back as the FRED chart goes. Real weekly median wages may be lower than they were in 1972).
Presently unit labor costs are a drag on the Fed’s 2% inflation target.
The best tonic for America would be a couple of generations of extremely tight labor markets— interestingly, tight labor markets no longer seem associated with inflation. Seize the day!
Matthias Görgens
May 31 2019 at 11:49pm
Haven’t both Scotts debunked the whole stagnant wages myth recently?
At most there’s a disconnect between median wages and average labour income. But there’s not much monetary policy can (or should) do about that.
Benjamin Cole
Jun 1 2019 at 8:19pm
https://fred.stlouisfed.org/series/LES1252881600Q
Matthias:
Nothing in macroeconomics is ever debunked, because no one is ever wrong.
The above FRED chart shows median weekly wages have not really budged much since 1978. The chart does not extend back before 1978 , and mercifully so, as I suspect real median wages were higher around 1972 or the late 1960s.
There is the additional problem of measuring real wages when it comes to soaring housing costs which are the norm along the West Coast and in the New York Boston regions.
Interesting question: supposed two generations ago a single person would move to Los Angeles and rent a single or efficiency apartment. Suppose today that same single person moves to Los Angeles but shares the apartment with a roommate. As a percent of income, the same fraction of household income is allocated to housing costs—- but obviously living standards are much reduced.
The most important thing in macroeconomics is to choose those data series that confirm your biases!
bill
Jun 1 2019 at 10:03am
In a certain sense, the market is saying it expects the Fed to do a poor job. If the Fed weren’t behind the curve, the most common market expectation would be No Change. And tied for (a distant?) second would a cut or increase.
Thaomas
Jun 1 2019 at 5:23pm
I don’t think “slow” is the right concept. The Fed’s failure to reduce interest rates (which are instruments, not targets) results form the failure to have a symmetric inflation target (i.e. and price level trend target) instead of of an inflation ceiling for the “prices” half of it’s dual mandate.
Paul
Jun 1 2019 at 11:34pm
“Instead, central banks should target (market) NGDP growth expectations and let markets determine interest rates.”
I’d love to see how a central bank targets something like NGDP without working through the banking sector and affecting financial conditions (interest rates). Even your NGDP futures idea simply substitutes a fed funds targeting regime for a base-targeting regime (the banking system is still the medium of transmission).
Your conception of a central bank is one that somehow exists (and can credibly drive expectations) but does almost nothing to implement policy. “You guys better spend or we’re going to create more money (which we can only do by changing the interest rate target!)!” is not monetary policy.
Comments are closed.