Thursday, October 5, 2017

U.S. Monetary Policy: What's Up?

To frame the issues, let's look at some objective measures of the Fed's performance. Just to be fair, we'll evaluate performance in terms of the objectives laid out in the FOMC's January 2017 goals statement.

1. "...inflation at the rate of 2 percent, as measured by the annual change in the price index for personal consumption expenditures, is most consistent over the longer run with the Federal Reserve’s statutory mandate." Note that the price index the Fed has chosen as most appropriate is the raw, headline pce deflator, not the pce deflator stripping out food and energy, the cpi, the core cpi, or any other measure. So we should hold them to that choice. Here's year-over-year inflation rates, since the last recession:
Obviously the Fed hasn't been hitting 2% inflation spot on, but what's actually feasible, or even desirable? The Bank of Canada, to take an example that's close at hand for me, actually sets a target band of 1-3% for inflation, so by that criterion the Fed has done pretty well - within the 1-3% band for most of the last seven years, except during 2015 after the fall in energy prices, which perhaps is understandable. The current inflation rate is 1.4%, which is tolerably close to the Fed's ideal. But the the Fed also has a "symmetric inflation goal." Missing the target sometimes is OK, but the FOMC would like to miss on the high side about as much as it misses on the low side - roughly, average inflation should be close to 2%. Over the last year, inflation was above 2% for two months, and below for 10 months, with an average of 1.6%. Not quite symmetric, but not so bad.

2. "... it would not be appropriate to specify a fixed goal for employment...Information about Committee participants’ estimates of the longer-run normal rates of output growth and unemployment is published...For example, in the most recent projections, the median of FOMC participants’ estimates of the longer-run normal rate of unemployment was 4.8 percent." For good reasons, the FOMC doesn't want to be specific about numerical goals related to the second part of its mandate - sometimes called "maximum employment," whatever that is. There's a hint though, about what the FOMC members might care about, which is the "long run normal" or natural rate of unemployment. This is rather ill-defined and hard to measure. To my mind, the economics profession would be better off if we refrained from mention of "natural" anything. One danger associated with the natural rate, as for any other ill-defined and hard-to-measure variable, is that a policymaker can start making stuff up, so as to manipulate the policy discussion. In the FOMC's most recent projections, the range of estimates for a long-run unemployment rate fall in a range of 4.4%-5.0%, with a median of 4.6%, so notions of what is normal have fallen since January 2017, when the FOMC put together its long-run plans revision. Here's what the actual unemployment rate looks like:
The current unemployment rate, at 4.4%, is as low as any FOMC participant thinks is normal over the long run, conditional on things going really well, apparently. So, by that measure the Fed is doing great.

Just to check, we can look at another measure, which is a measure of labor market tightness (or it's inverse, as conventionally measured in the labor literature). This is the ratio of the number of unemployed to total job openings:
Shortly after the last recession ended, this measure peaked at about 6.6 unemployed people per job opening, and it's now down to close to 1 unemployed person per job opening. Indeed, that's close to the lowest reading since the BLS started collecting this particular job vacancy data. So, by conventional measures the job market is very tight, which should be viewed as extremely good performance on the Fed's part.

What about growth in real GDP?
If we think that part of the Fed's job is to smooth growth in real GDP, then that chart looks pretty good. I've put in a 2% growth path, and the deviations from that growth path are small. Of course, people might complain that 2% growth is lower than the 3% (roughly) post-WWII average, but that shouldn't be the Fed's concern. Received wisdom in the economics profession is that monetary policy can't do much for long-run growth, other than keeping inflation low and predictable.

So, what's to fix here? Inflation is tolerably close to 2%, the unemployment rate is very low, the labor market is extremely tight, and real GDP is growing smoothly. The only improvement to be made is some fine tuning so that inflation fluctuates symmetrically around the 2% target. How should that be done? I'll assume, consistent with my last post, that QE doesn't matter. So the only issue is what should be done to the fed funds target range (or, more accurately, the interest rate on reserves and the interest rate on overnight reverse repurchase agreements), so that the average inflation rate is 2%? Well, you don't have to be a neo-Fisherite to understand that, if inflation is persistently lower than what you want, on average, then the nominal interest rate needs to be, on average, higher in the future. What's needed here is some tweaking of the Fed's policy interest rate target. How much? As mentioned above, the average inflation rate over the last year is 1.6%, so one or two more interest rate hikes will do the trick. Twenty five to fifty basis points' increase in overnight interest rates is small potatoes for real economic activity - note that the labor market continued to improve in the face of the last four interest rate increases.

So, that's what I'd do. What does the FOMC have on its mind? In the last FOMC projections, the median long-run prediction of Committee members for the fed funds rate is 2.8%, with a range of 2.3%-3.5%. That's come down considerably, with recognition by the committee that the low real rates of return on government debt we are observing are likely to persist. A persistently low real rate of return on short-term safe assets implies of course that the nominal short term interest rate consistent with 2% inflation is low. I'm saying that what Janet Yellen would call the "neutral interest rate" (the interest rate target at which the Fed achieves its goals in the long run) is more like 1.5%, and not 2.3%-3.5%.

To get more information on what the FOMC is likely to do over the near future, we'll look at Janet Yellen's last speech on "Inflation, Uncertainty, and Monetary Policy." First, Yellen tells us how inflation has been low, and then says why she thinks low inflation is bad:
Sustained low inflation such as this is undesirable because, among other things, it generally leads to low settings of the federal funds rate in normal times, thereby providing less scope to ease monetary policy to fight recessions. In addition, a persistent undershoot of our stated 2 percent goal could undermine the FOMC's credibility, causing inflation expectations to drift and actual inflation and economic activity to become more volatile.
The second sentence is important. The Fed committed to a 2% inflation target because the assurance of predictable inflation minimizes uncertainty, and makes credit markets, and (by some accounts) the markets for goods and services work more efficiently. If the Fed consistently undershoots its inflation target, people will either think the Fed is incompetent, or that it is willfully abandoning its promises, neither of which is good - for the institution or the economy. But in this instance, the Fed isn't missing by much, so what's the big worry? As I mentioned above, this requires some fine-tuning, but don't get bent out of shape about it.

The first sentence in the quote was really interesting. She's got the causality backward. Pretty much all of us now accept that it's the central bank that controls inflation. That is, central bank actions or, more accurately, the central bank's policy rule, causes inflation to be what it is, combined of course with other factors outside the Fed's control - including the factors determining the long-run real rate of interest on government debt. So, low inflation does not lead to "low settings of the federal funds rate." It's the low settings for the fed funds rate that lead to the low inflation. In a world in which the central bank targets the nominal interest rate to control inflation, that's how it works, and central bankers would be wise to absorb that idea. Stop the neo-Fisherian denial, and get with the program!

The speech uses a two-equation model (written down in the appendix) to frame the issues. It's a Phillips curve model. Arrgghh. Even Larry Summers recognizes that Phillips curves are unreliable. As he says:
The Phillips curve is at most barely present in data for the past 25 years.
For example, in the recent post-recession period, here's what we get when we plot inflation against the measure of labor market tightness I used above (ratio of unemployed to job seekers):
The line connects the observations in temporal sequence from right to left. Over this period of time, I think the Fed would claim that inflation expectations are more or less "anchored." So, what we should see in the chart, if the Phillips curve is to be at all useful, is a set of observations tracing out a downward-sloping relationship. But, more often than not, inflation and my "slackness" measure are moving in the same direction. There's nothing new about macroeconomists raising issues with the Phillips curve as a cornerstone for policy. It's been in disrepute for much of the last 45 years or so, both on theoretical and empirical grounds. Unfortunately, the Phillips curve was dragged out of the gutter, dressed up, and rehabilitated by New Keynesians, which is another story altogether.

But, like a lot of people, Janet Yellen is a true believer, and her staff will aid her in that belief by going on a fishing expedition and finding a Phillips curve, and a sample period, for which all the signs in the regressions (if not the magnitudes) come out "right." Here, "right" is whatever conforms to the beliefs of the boss. Sure enough, in the appendix to Yellen's speech, there is a two-equation Phillips curve model. Core inflation is determined by past core inflation, inflation expectations, resource slack, and the relative price of imported goods, and core inflation, energy price inflation, and food inflation determine headline inflation. Yellen's worries about future inflation outcomes are essentially those of the true believer. Do we have the right slackness measure, or the right inflation expectations measure, and how much should we worry if expected inflation falls?

Though Yellen lays out an explicit model of inflation, she doesn't exactly tell us how policy is supposed to work within that framework. Even true believers will sometimes tell you that "the Phillips curve is now very flat," meaning that they think a tighter labor market will put little or no upward pressure on inflation. If we want to stick with the Phillips curve framework, what's left then? The Fed has to focus on anticipated inflation. But how do they move that around? Modern macroeconomics tells us that our views about future outcomes are shaped by what we know about policy rules, in a manner consistent with what we know about how the world works. I got no sense from Yellen's talk of how the Fed thinks its policy rule affects inflation expectations.

But here's the essence of the FOMC's current policy view:
...without further modest increases in the federal funds rate over time, there is a risk that the labor market could eventually become overheated, potentially creating an inflationary problem down the road that might be difficult to overcome without triggering a recession.
So, in spite of the fact that the Phillips curve doesn't fit the data, the most recent manifestation being the failure of the very tight labor market to make inflation go up, policy going forward will be driven by the fear that the Phillips curve will somehow wake up and re-assert itself. Summers thinks that's wrong, and rightly so.

But, I think Yellen and her colleagues are actually following the right policy. Modest increases in the Fed's interest rate target in this context is the correct prescription. But doing the right thing for the wrong reason won't help you in the long run. We're fortunate, though, that not much is likely to go wrong here. If inflation stays low, and the FOMC loses its appetite for interest rate increases, so what? Low inflation is fine, and it's close enough to 2% as not to be embarrassing. No big deal.

Monetary policy is the least of our now-staggering problems. Unfortunately, the wingnut in the White House is busy creating more difficulty for us, and making the problems we have worse. Fortunately, he has as yet not screwed up the Fed, and the slate of would-be Fed Chairs doesn't include anyone outlandish. More on that later.

Wednesday, October 4, 2017

Whatever Happened to Normalization?

What's become of the Fed's normalization plans? To get this straight, recall what's been abnormal about Fed policy for the last nine years or so. Here's a chart of the effective fed funds rate, and securities held outright by the Fed:
Abnormal policy began at the height of the financial crisis in late 2008, when the FOMC agreed on a plan to target the fed funds rate in a range of 0-0.25% - a policy that continued until "liftoff" in December 2015. As well, beginning in early 2009, the Fed embarked on a sequence of quantitative easing (QE) exercises, which increased the quantity of securities held outright by a factor of more than five. Further, the Fed got rid of essentially all of its Treasury bill holdings, and increased the average maturity of Treasury bonds and notes held. The Fed also purchased a large quantity of mortgage backed securities (MBS) - close to $1.8 trillion. So, the Fed increased the size of its balance sheet substantially, lengthened the average maturity of securities held, and departed in a big way from a policy of "Treasuries only."

As outlined in this FOMC document, the FOMC began thinking seriously about how Fed policy might return to normal, and what "normal" might be, as early as June 2011. A formal normalization plan was posted by the FOMC in September 2014, and this is essentially what has been implemented since, more or less. The plan was:

(i) Begin increases in the fed funds rate target.
(ii) Reduce the size of the balance sheet by stopping the reinvestment policy, after increases in the target policy rate are well underway.

Increases in the fed funds rate target began in December 2015, and we have since had three more, with the target range increasing from 0-0.25% to 1-1.25% currently. Balance sheet reduction did not commence until October of this year, when the FOMC issued an addendum to the 2014 plan. The addendum contains explicit details about how the balance sheet reduction will occur. Reinvestment - a policy by which assets in the Fed's portfolio are replaced as they mature, holding the nominal size of the balance sheet constant - did not stop abruptly, but its cessation will be phased in. It appears that the New York Fed did not purchase assets with a view to smoothing quantities that mature over time, and the FOMC seems concerned that the balance sheet not decline in a lumpy fashion, as it would without the caps on portfolio reduction outlined in the addendum.

Some questions that might come to mind (or should) on normalization, along with my answers:

1. Why did normalization start with interest rate increases first, then reductions in balance sheet size? It might seem logical, since QE followed the reduction in the nominal interest rate target to zero (effectively), that the Fed would normalize by first reducing the balance sheet to a normal size, and then increase interest rates. Indeed, there are some good reasons why this is what should have happened. As short-term nominal interest rates increase, the profit that the Fed makes on the spread between the return on its assets and what it is paying out on its liabilities declines. As a result the Fed makes a smaller transfer to the Treasury each year. QE took place in the context of relatively low yields on Treasury bonds and MBS, and with a larger balance sheet, the asset portfolio is being financed by a larger fraction of interest-bearing reserves and a lower fraction of zero-interest currency. If short-term rates go high enough, transfers to the Treasury will stop. Economically, this is unimportant, as this amounts to the difference between interest paid on reserves by the Fed vs. interest paid on government debt by the Treasury, but politically this could be very dangerous territory. The Fed should not give ammunition to its enemies in Congress. Added to this is the argument that QE was an experiment, with poorly understood consequences. Thus, the sooner the Fed ended the program, the better. So why not reduce balance sheet size before engaging in liftoff? Likely, because the FOMC was spooked by its experience in 2013. At that time, after the FOMC meeting that ended on June 19, Ben Bernanke announced that a winding-down, or "tapering" of the Fed's QE program was likely to being later that year, and that the program would probably end in mid-2014. The financial market response to that announcement, and earlier public statements by Bernanke, is sometimes called the "taper tantrum:"
In the chart, you can see an increase of more than 100 basis points in the 10-year Treasury bond yield, observable in both the nominal yield and the inflation-indexed yield. Somehow, this wasn't the response the Fed expected, but if the market viewed Bernanke's statement as news about forthcoming interest-rate target increases, the reaction doesn't seem outlandish in retrospect. In any case, this experience appears to have colored FOMC views on the importance of QE, and made them skittish about unwinding the program. Thus the idea that interest rate increases should be well under way before the Committee would even think about balance sheet reduction.

2. What's different about raising interest rates when the Fed has a large balance sheet? In theory, when there are reserves in excess of reserve requirements in the financial system overnight, the interest rate on excess reserves (IOER) should determine the overnight rate. This is called a floor system, under which interest rate control is easy, as the overnight interest rate can be essentially administered by the central bank. But in the United States, things aren't so simple. For the details see this article, and this one. Basically, there are regulatory features of the US financial system that restrict arbitrage in the overnight market, so that the "effective" federal funds rate is typically lower than IOER. And, as was also the case before the Fed began paying interest on reserves in late 2008, all fed funds trades don't happen at one interest rate on a given day - indeed, much of the fed funds market is conducted over-the-counter. The Fed was concerned, before liftoff happened, about its ability to achieve a given target range for the fed funds rate - would the fed funds rate even go up with increases in IOER? To assure that this would happen, the Fed expanded the market for its liabilities by making use of an overnight reverse repurchase agreement (ON-RRP) facility. ON-RRPs are loans to the Fed, usually overnight, secured by securities in the Fed's portfolio. These Fed liabilities are just reserves by another name - they can be held, for example, by money market mutual funds, which are prohibited from holding reserve accounts with the Fed. Currently, IOER is set at 1.25%, the ON-RRP rate is set at 1.00%, and on most recent days the effective fed funds rate is 1.16%. Here's a chart showing what has happened with Fed interest rate control since liftoff:
The chart shows takeup on the ON-RRP facility (quantity of ON-RRPs outstanding) and the effective fed funds rate. Initially, the Fed was willing to commit up to $2 trillion in collateral to ON-RRPs, but takeup is typically in the range of $50 billion to $250 billion, running up to $400 billion to $500 billion only at quarter-end (this for regulatory reasons). As well, the effective fed funds rate has recently been coming in consistently at about 9 basis points below IOER (except at month-end, again for regulatory reasons), and more detailed data shows that most trades happen around the average. In one sense interest rate control since liftoff appears to be a success. But it's not clear that the ON-RRP facility is necessary for its stated purpose. The fed funds rate might be about where it is now even without an active ON-RRP facility. We could go further though, and question this whole operating strategy. There is no good reason for the Fed to focus on a fed funds rate target. Fed funds are unsecured, and currently most of the trade in this market is just a means for some GSEs to earn overnight interest on reserve balances. Even in pre-crisis times, it would have made much more economic sense if the Fed had announced its overnight interest rate target as a target for an overnight repo rate. In the current context, why isn't the ON-RRP rate set equal to IOER? Maybe that would kill the fed funds market, but so what?

3. What happens to reserves when Fed assets mature and there is no reinvestment? The balance sheet of the Fed balances, just as any balance sheet does, so for any transaction that occurs affecting either assets or liabilities, implying a debit or credit, there must be an offsetting debit or credit. Suppose first that the maturing asset is a MBS. The issuer of the MBS - which would be a GSE if the MBS is held by the Fed - pays the face value of the debt to the Fed, and the payment will be made by reducing the balance in the GSE's reserve account by that amount. But the MBS that the GSE issued is composed of bits and pieces of underlying mortgage debt. Suppose that the reason the MBS matured was that the underlying mortgage debt was paid off. Then, mortage payments are made to the GSE, and ultimately those payments will involve an increase in the GSE's reserve balances, and a reduction in reserve balances held by private financial institutions. So reserves held by private sector institutions (a Fed liability) and MBS holdings (a Fed asset) decrease by the same amount. Suppose, alternatively, that a Treasury security held by the Fed matures. The Treasury holds a reserve account with the Fed, and a maturing Treasury security implies that the Treasury's reserve account balance (the account is called the "General Account") falls by the face value of the debt. But that has no implications for the private sector - it's just an accounting transaction between the Fed and the Treasury, like internal budgeting transactions between the English department and the Economics department at the University (if such a thing ever happens). There would be implications for the private sector if the Treasury, now finding itself short of reserve balances to pay for stuff, issues more debt to replenish those reserve balances. Then, the new Treasury debt is purchased by the private sector (the Fed won't be buying it, as it's not reinvesting) with reserve balances, so reserves held in the private sector fall. If you think about this a bit, you'll see that, if we think the level of of reserve balances held by the private sector is part of monetary policy, then the Treasury can engage in monetary policy, by varying the quantity of reserves in its reserve account. Look at this:
Note that both the level and variability of balances in the Treasury's general account have increased by a huge amount since the financial crisis. Maybe the Treasury thinks this doesn't matter now, as it won't mess with monetary policy, but that view is at odds with what the Fed says - which is that QE matters in a big way. If QE matters so much, then the big increase in average balances in the General Account in 2016 should have been a significant "tightening" (because it implies a reduction in privately-held reserves) that the Fed would be concerned with. What's going on?

4. When will balance sheet reduction stop? What the FOMC's normalization addendum says is that they don't know, so let me fill you in on what the issues are. The Fed needs to decide on a long-run operating strategy for monetary policy. They could adopt a channel system for monetary policy, like what Canada has, for example. This would involve operating with a small balance sheet. For example, in Canada, where there are no reserve requirements, overnight reserves are essentially zero. The target overnight rate in such a system is bounded by the interest rate at which the central bank lends (the discount rate, for the Fed) on the high side, and IOER, on the low side. Alternatively, the Fed could stick with the floor system under which it operates now, according to which there are excess reserves in the system overnight. The question then is how much reserves you need to make the floor system work. Basically, financial institutions have to be more or less indifferent between lending to the Fed overnight, and lending to to private entities overnight, so that the interest rates on Fed liabilities determine all overnight rates. Evidence from the Canadian experience from Spring 2009 to Spring 2010 suggests that number is smaller than some people seem to think. Probably less than $100 billion. In addition there are issues concerning what overnight rate the Fed should be targeting. In many countries the central bank targets a repo rate, which makes sense, as the central bank should be interested in a secured overnight rate, that is not contaminated by risk. Why persist in speaking to a fed funds rate target, particularly in a financial crisis?

5. Did QE actually work? Don't expect to get good information from the Fed about this. Central bankers want to at least keep up appearances. Who wants a central banker who's not knowledgeable and trustworthy? As I mentioned above, the Fed has many enemies - in Congress and elsewhere - and it's typically optimal for the institution if Fed officials don't admit to not knowing stuff. Truth is that I've never seen any solid evidence that the people who implemented QE in the Fed system actually have a grip on how it might work - either in theory or in practice. Bernanke once said that"QE works in practice but not in theory." I've heard that repeated many times, usually by a person with a smug look on his or her face. Basically, the statement's B.S. The evidence that QE works is weak or nonexistent. I've written about this in more detail in this St. Louis Fed article. Most of the pro-QE evidence comes from questionable event studies, and the evidence we have seems consistent with QE having no effects for the Fed's ultimate objectives. For example, the Bank of Japan has for more than four years engaged in a massive QE experiment that has had no discernible effect on inflation. And QE does in fact work in theory - at least in the 1950s and 1960s vintage theories that Berananke trotted out to justify the policy in the first place. More careful thought might make one think that QE could actually be harmful, by withdrawing useful collateral from financial markets and replacing it with inferior reserves (talk to people who understand "financial plumbing," for example Peter Stella or Manmohan Singh) It's possible that QE could do some good, if the Fed had the proper liabilities at its disposal. QE is basically an attempt by the central bank to engage in debt management, which is the job assignment of the Treasury in the United States. Maybe the Fed can do a better job of debt management than the Treasury, but if so there should be a public discussion, and an explicit assignment of tasks. And if the Fed is doing debt management it needs to be able to issue tradeable debt instruments of all maturities.

So, where are we? With the unemployment rate at 4.4% and the inflation rate at 1.3%, the Fed is achieving its goals, within reasonable tolerance. We're no longer in emergency territory, yet the Fed has an emergency-sized balance sheet. The plan they have issued to reduce the balance sheet is overdue, and it's quite timid. For example, note that during the QE3 program (the final stage of the Fed's asset purchase program), the Fed bought $85 billion per month in long-maturity Treasuries and MBS and that, even after a phase-in period, under the disinvestment program the reduction will be $50 billion per month, at most. Chances are that, when a recession comes along, the balance sheet will still be large, the interest rate target will quickly go to zero, and asset purchases will resume. If the Fed's balance sheet achieves any semblance of "normal" in my lifetime, I'll be amazed.