The Fed's new monetary policy framework

The Fed's much-anticipated new monetary policy framework is now public. Fed Chairman Jerome Powell outlined the policy framework last week in Jackson Hole; you can view his speech here. Overall, I thought Powell's delivery was very good. While there's room for improvement, I think the new framework is a step in the right direction (George Selgin provides a good critique here). There were three things in Powell's speech that stuck out for me. I discuss these below. 

Shortfalls vs. Deviations

At the 22:30 mark, Powell reports what may very well be the most substantive change to the monetary policy statement. Here, he states that the FOMC will now interpret important macroeconomic time-series like GDP and unemployment as exhibiting "shortfalls" instead of "deviations" from some ideal or "maximum" level (a frustratingly vague concept). 

The practical effect of this shift is to remove (or make less prominent) in the minds of FOMC members the idea that the economy is, or will soon be, "overheating" (i.e., embarked on an unsustainable path that can only end in misery for those most vulnerable to economic recession). 

The idea of "deviation from (some) trend" seems like a plausible description of the postwar U.S. up to the mid-1980s. Severe contractions were usually followed by equally robust recoveries. However, this representation seems to break down since the "great moderation" that began in the mid-1980s. Since then, economic recessions have not been followed by above-average growth. Instead, each recession seems better described as a "growth shortfall." We're not entirely sure what accounts from this cyclical asymmetry, but it seems consistent with Milton Friedman's "plucking model." I think we can expect a stream of research resurrecting this old idea (see here, for example).

 

In any case, the upshot here is that, to the extent that "overheating" is no longer considered a serious threat, the FOMC will be less likely to implement "preemptive" policy rate hikes. This constitutes a tacit acknowledgement that the period leading up to "lift off" and what followed might have been handled better. As I wrote at the time (see my discussion here), standard Phillips Curve logic did not seem to support tightening (unemployment was above the estimated natural rate, inflation was below target, and inflation expectations were declining). But the Committee somehow talked itself into the need to "normalize," to act preemptively and not get caught "behind the curve." In fairness, monetary policy is always about balancing risks (in this case, the perceived risk of overheating). In the near future, less weight will be assigned to the risk of overheating. 

 

The Maximum Level of Employment

 

At the 22:30 mark, Powell states "Of course, when employment is below its maximum level, as is so clearly the case now, we will actively seek to minimize that shortfall..."

 

I have a hard time not interpreting "maximum" here as "socially desirable." I think most people would agree that the 2008 financial crisis caused employment to decline below its maximum level. The workers rendered idle in that episode constituted a social waste, and the Fed was right to loosen monetary policy to stimulate economic activity in the face of recessionary headwinds. 

 

But the recession induced by the C-19 is very different from standard recessions. This was laid out very clearly by St. Louis Fed President Jim Bullard on March 23, 2020: Expected U.S. Macroeconomic Performance during the Pandemic Adjustment Period. According to Bullard, the temporary removal of some workers from their jobs is not, in this case, a waste of resources. The decline in employment in this case should be viewed as an investment in public health. That is, the maximum level of employment declined and its recovery is driven mostly by the contagion dynamic (as well as improvements in social distancing protocols, masking, testing, treatments, etc.). The role of monetary policy here is to calm financial markets (which the Fed successfully accomplished in March) and to aid the fiscal authority with its income maintenance programs. In short, the primary monetary/fiscal policy objective here is to deliver insurance, not stimulus. 

 

Monetary stimulus is appropriate, however, to the extent that demand factors (e.g., individually rational, but a collectively irrational restraint on spending) are inhibiting the recovery dynamic. The evidence for this is usually assumed to be found in falling inflation and inflation expectations, and declining bond yields. And usually, this makes sense, because we usually assume that recessions are caused by collapses in aggregate demand (as in 2008-09). But what if the increase in the demand for money (safe assets in general) is driven by a collectively rational fear? We'd expect to see the exact same inflation and interest rate dynamic, but the role for stimulative monetary policy would be more difficult to justify (though the desirability for insurance remains). 

 

So, maybe it is not so clearly the case now that employment is below or, at least, far below its "maximum" level. Note that a significant part of the decline in aggregate employment is coming from the leisure and hospitality sector: 

Arguably, we do not want, at this stage of the pandemic, to promote the indoor dining experiences people enjoyed earlier this year. This activity will return slowly as economic fundamentals improve. The "full employment" level of employment in this sector is clearly below what it was in Jan 2020. But, to be fair, it is entirely possible, and perhaps even likely, that the level of employment even here is lower than the "full employment" level. It's very hard to tell by how much though. 

 

Average Inflation Targeting

 

At the 24:00 mark, Powell explains how AIT will help anchor inflation expectations. Missing the inflation for a prolonged period of time will cause expectations to drift away from target and line up with the historical experience. This view of expectation formation is firmly rooted in the "adaptive expectations" tradition. That is, expectations are assumed to be formed by looking backward instead of forward. 

 

People sometimes claim that adaptive expectations are inconsistent with "rational" expectations. But this is not necessarily the case. In fact, it makes sense to use the historical record of inflation realizations to make inferences about the long-run inflation target if people are not sure of the monetary authority's true inflation target; see, for example, here: Monetary Policy Regimes and Beliefs. 

 

It's still not entirely clear to me whether FOMC members view AIT as a policy to pursue passively (i.e., let inflation creep up to and beyond target on its own) or actively (i.e., take explicit actions to promote an overshoot of inflation). If it's the former, then I'm on board with the idea. But if it's the latter, I am not. In particular, with the liquidity-trap-like conditions we're presently in, the Fed does not have the tools (or political will) to boost inflation persistently. It is likely to fail, just as the Bank of Japan failed. (I explain here why it's more difficult for a central bank to raise the inflation target than to lower it.) So, as I've advocated many times in the past, why not just declare 2% as a soft-ceiling and let fiscal policy do the rest? 

 

My view rests on the belief that missing the inflation target from below by 50bp over the past eight years is not a significant macroeconomic problem (especially given how crudely inflation is measured). The FOMC did view it as a problem, but mainly, it seems, because of the embarrassment associated with missing its target. "We are a central bank. We have an inflation target. Central banks are supposed to hit their inflation targets. We need to hit our inflation target to remain credible." This is why earlier FOMC statements emphasized the Fed's "symmetric" inflation target. That did not work and so now we have AIT which, I'm afraid, might not work either. Happily (for those who want to see higher inflation), Congress seems comfortable with the idea of producing large budget deficits into the foreseeable future.  

 

So, if we get higher inflation, it will largely be a fiscal phenomenon. The purpose of AIT is to accommodate any rise in inflation for the purpose of increasing inflation expectations and avoiding the specter of deflation (people often point to Japan as a case to avoid, by Japan seems to be doing fine as far as I can tell). There is the question of how the Fed would react should inflation rise sharply and persistently above 2%. Even if the event is unlikely, it would be good to state a contingency plan. In the past, the Fed could be expected to raise its policy rate sharply. But this event, should it transpire, will almost surely take place during an employment shortfall (since this is now the acknowledged new normal). The only prediction I'll make here is that the FOMC will have a lot of explaining to do in this event. 

 

Some thoughts on yield curve control

There's been a lot of talk about "yield curve control" (YCC) as of late. I found the recent exchange between Joe Weisenthal and David Beckworth (with many others chiming in) very interesting:

I normally enjoy Joe's hot takes, but this one... yikes. It is a good example, in my view, of why relying too heavily on the "money view" (i.e.liquidity preference view) of interest rates can cause one to miss the forest for the trees. Let me explain...1/n https://t.co/EjUltLb4dF

— David Beckworth (@DavidBeckworth) August 9, 2020

A number of us gathered on Zoom to discuss the subject. What follows is my own take on YCC and some of the issues involved. If you're interested in joining in on a future Zoom discussion, let me know.  

One thing I learned from the people I talked to is that my notion of YCC seemed to differ from the way they were thinking about it. Most people seem to have in mind the idea of YCC as a form of interest-rate peg, only with the fixed peg applying to interest rates at all maturities, for example, in the manner of Fed policy over the period 1942-47 and 1948-51.

In contrast, I view YCC as a state-contingent policy that pegs (or sets a narrow corridor for) rates at all maturities. The slope of the curve may be held fixed, with policy determining the level shifts in the yield curve (so, basically an extension of the Taylor rule applied to interest rates at all maturities). Or policy could also change the shape of the curve, making it steeper or flatter (so, basically replicating "Operation Twist" type interventions). Let me distinguish this notion of YCC by labeling it RBYCC (rule-based YCC).

What is the rationale for RBYCC? The RB part has the standard rationale. But what's the point of YCC then? The way I look at things is as follows. For some odd reason, the Treasury finances the deficit by issuing U.S. Treasury Securities (USTs) with different maturities. These securities are nominally risk-free. But if they all constitute risk-free claims to cash (reserves), then why do/should these objects sell at different prices? And even if there is some "preferred habit" force at work, why doesn't the treasury exploit the apparent arbitrage opportunity, selling securities that trade at a premium (typically bills) and repurchasing securities that trade at a discount (typically bonds). Indeed, why even issue securities that the market discounts (a polite way of saying hates) in the first place? (I offer one rationale here: Maturity Structure and Liquidity Risk).

If by "liquidity" we mean the ability to convert a security in reserves (or bills), then it is clear that the liquidity of USTs is a policy choice. It would be a simple matter for the Fed and/or Treasury to set up a standing facility prepared to buy/sell USTs of any maturity on par with reserves, for example. This policy would have the effect of eliminating discounts across all securities. If you don't like this policy, then you'll have to explain why it's a good idea for government securities to trade at discounts relative to each other. To me, this is like saying a $10 bill should be discounted relative to two $5 bills. (Note: I am not saying such an argument does not exist--indeed, my paper above makes one such argument.)

The effect of RBYCC would be to render all USTs equivalent to reserves (which itself leads to the question of why deficits can't be financed entirely with interest-bearing reserves). The same would be true of YCC with a fixed pattern of discounts, as in the U.S. from 1942-47. This much was recognized by Friedman and Schwartz in their Monetary History when they wrote "The support program converted all securities into the equivalent of money" (pg. 563). In theory, this type of policy should work as well or better than simply targeting the short rate. It eliminates the liquidity premia on government debt (i.e., it satiates liquidity demand) and it permits the usual sort of Taylor rule to stabilize the economy.

As mentioned above, however, most people probably think of YCC as a peg-like policy. One argument against interest rate pegs is that they induce instability. The U.S. experience over 1942-47 and 1948-51 is widely interpreted as having promoted excess inflationary pressure. Let me briefly review those episodes here. 

At the time, the Fed set the short rate at 3/8% and capped a long rate at 2.5%. Measured inflation remained low, thanks to wartime wage and price controls. Interestingly, the 2.5% cap seemed non-binding. It is likely that long yields remained low because investors expected the Fed to keep the short rate low for the indefinite future. We know that at the time, investors were selling bills to the Fed and acquiring higher-yielding bonds to exploit the apparent arbitrage opportunity (see Chaurushiya and Kuttner, 2004). Inflation only took off once wage and price controls were lifted in 1946. While this burst of inflation was likely only temporary, a concern over inflation led the Fed to raise the short rate to 1% in late 1947 when inflation had already declined from 20% to 10%. Inflation then stabilized for about a year at 8%, before declining sharply to 2.75% in 1948. At this time, the economy went into a recession, lasting until the last quarter of 1949. The inflation rate fell below zero in May 1949 and stayed below zero until July 1950 (so, well over a year of deflation). 

Let me summarize this episode. Under this YCC policy, inflation fell from a peak of 20% in March of 1947 to about 10% in November of 1947 with the bill rate still pegged at 3/8%. Then, with the rate hike pegged at 1%, inflation continued to fall rapidly, hitting a low of negative 3% in August of 1949 (near the end of the recession). It took until June 1950 for inflation to rise to 0%. 

Inflation then began to rise rapidly after June 1950 – the month the United States entered the Korean War. The Treasury wanted to keep interest rates low to facilitate war finance. The Fed favored high interest rates to combat inflationary pressures created by the war. Inflation peaked in early 1951 at 9.5%. Chaurushiya and Kuttner, 2004 write:

“It became abundantly clear during this period that the interest rate caps were hampering the Fed’s ability to achieve its monetary policy objectives and, in particular, its efforts to contain rapidly rising inflationary pressures.”

This experiment in YCC ended with the Treasury Accord in March 1951. And the narrative that YCC is is inconsistent with inflation control was born.

My own interpretation of these events and of the efficacy of YCC is as follows. First, it seems a bit of stretch to "blame" inflation over this episode as the consequence of YCC. For most of the 1942-51 period, the U.S. was at or recovering from war. Wars are known to place great fiscal strain on governments and financing a war effort with higher-than-normal inflation is likely desirable from the perspective of optimal public finance policy. That is, the U.S. would have likely experienced higher-than-normal inflation under any reasonable interest rate policy. 

I interpret the interest rate hike in 1947 as an example of how RBYCC can work to control inflation. In this example, the short rate was increased to 1% and the long-rate remained capped at 2.5%, in effect flattening the yield curve. The disinflationary impact of this rate hike seems evident in the data above. So, it seems clear that RBYCC can be used to control inflation, even if it seems to have been employed rather clumsily in 1947. 

As usual, looking forward to your comments/criticisms, which can be left below. 

Why the Fed Should Create a Standing Repo Facility

     

I was invited recently to take part on a panel discussion
on Modernizing Liquidity Provision as part of a conference hosted jointly by CATO and the Mercatus Center entitled A Fed for Next Time: Ideas for a Crisis-Ready Central Bank.  My post today is basically a transcript of the presentation I gave in my session. I'd like to thank George Selgin and David Beckworth for inviting me to speak on why the Fed should create a standing repo facility, an idea that Jane Ihrig and I promoted early last year in a pair of St. Louis Fed blog posts here and here.

In those posts, Jane I argued that the Fed should create a standing repo facility that would be prepared to lend against U.S. Treasury securities and possibly other high quality liquid assets (HQLAs). We distinguished the facility we had in mind from the discount window in two key respects. First, unlike the window, it would restrict collateral to consist only of HQLA; and second, it would grant access to non-depository institutions, in particular, to dealers and possibly even to all the counterparties that are presently permitted to access the Fed's ON RRP facility.

At the time, we motivated the facility as a way for the Fed to conduct monetary policy in a manner consistent with the FOMC's preferred operating framework of ample reserves together with its 2014 Policy Normalization Principles and Plans which stated, among other things, the desire to hold "no more securities than necessary to implement monetary policy efficiently and effectively."

Jane and I speculated that a significant source of the demand for reserves over other HQLAs came from the Global Systemically Important Bank's (G-SIB's) perceived need for resolution liquidity. We reasoned that these G-SIBs might be more inclined to hold higher-yielding HQLAs over reserves if it was known beforehand that the former could be readily converted into reserves on demand at a standing facility at pre-specified terms. At the same time, the facility would provide a ceiling on repo rates and eliminate the need to estimate the so-called "minimally ample" level of reserves. That is, the facility would automatically flush the system with the reserves it needed as reserve supply and demand conditions varied because of adjustments in the Treasury General Account or other economic factors. Finally, we doubted whether the facility would lead to any significant amount of disintermediation as some people feared. In our view, it would serve mainly to cap the terms of trade in a number of over-the-counter (OTC) repo transactions involving Treasury debt.

The title of this session is "Modernizing Liquidity Provision." We're here today, of course, because of the massive Fed-Treasury interventions in response to the COVID-19 pandemic. Jane and I didn't tout the standing repo facility as a crisis tool because we figured that in a crisis, investors were unlikely to have much difficulty in finding buyers of U.S. Treasury securities. Since the 2008-09 financial crisis, we've grown accustomed to the idea of USTs serving as a flight-to-safety vehicle. And, indeed, this seems to have been the case as the present crisis initially unfolded. Bond yields began to drop sharply in Februrary and then again following the Fed's rate cut on March 5, with the 10-year hitting a low of 54bp on March 9.

But then something happened that I don't think anyone was expecting (certainly, I was not). In particular, after March 9, there's clear evidence of selling pressure stemming from what looked like a repo run on treasury securities. That is, for a variety of reasons there was an enhanced demand for cash which, in this instance, led to sales of U.S. Treasuries, depressing their value as collateral--effectively evaporating a significant portion of the supply of safe assets--which led to margin calls, which led to further selling pressure, and so on.

When the Fed cut its policy rate to 10bp on March 16, bond yields continued to rise, with the 10-year hitting almost 120bp on March 18. Bond yields came down only after the Fed intervened first with its discretionary repo operations and then with $1.5T of outright purchases of securities. This episode reminds us again that cash is king in a crisis and that U.S. Treasury securities are not always considered cash-equivalent in a crisis.

A natural question to ask here is whether disruptions like this constitute a policy problem. After all, it's not like bond traders are unfamiliar with the notion of interest rate volatility. When I glance at the data, the absolute size of this volatility seems more or less stable since the mid 1980s. However, because interest rate levels are so much lower today, a 50bp move is quantitatively more significant in relative terms. This wouldn't be much of a problem, in my view, if treasury securities served merely as pure saving instruments. But for better or worse, the UST has evolved over time to become an important form of wholesale money. In particular, it is used widely as collateral in the repo market (the so-called shadow bank sector). Its value as collateral stems in large part from its perceived safety and liquidity. And most of the time, the U.S. Treasury market is liquid. Except for when it isn't, of course. And so the question is, when it isn't liquid, does it matter and, if so, should something be done about it?

My views on this questions are informed by both by theory and from what I know of the history of the U.S. Treasury market (e.g., Garbade 2016). Theory tells us that in a fiat money system, there's no fundamental difference between account entries at the Federal Reserve and (say) at Treasury Direct. They are both electronic ledgers containing interest-bearing accounts. There are legal differences, of course. Only depository institutions have access to Fed accounts, whereas treasury securities can be held much more widely. Treasury securities are more complicated objects because they differ from each other in terms of coupon, time left to maturity, and possibly other characteristics. For this reason, treasury securities, as with most bonds, trade in decentralized over-the-counter markets instead of centralized exchanges.

While OTC markets may have their advantages (they evidently displaced the centralized exchange of bonds in the 1920s), their decentralized structure can be problematic. When investors become fearful, bond dealers and other traders may become unwilling or unable to execute trades, so that meaningful price information is lost. Safe assets may trade at significant discounts or premia, not for any fundamental reason, but simply because liquidity (market participation/communications) has vanished. Such events have implications that extend beyond the treasury market because, as is well-known, the yield on Treasury debt serves as a benchmark for many other financial assets. Unnecessary and avoidable problems in the treasury market can spillover into other financial markets, bringing grief to the broader economy.

From this perspective then, I am led to ask the question: in what world does it make sense to permit risk-free claims to fiat money like treasury securities to suddenly become illiquid? (This question is distinct from the one that asks whether risk-free claims to fiat money should be made illiquid--as in, the issuance of non-marketable debt; see here, for example) There is really no good reason, as far as I know.

I therefore continue to believe that a standing repo facility makes a lot of sense for the U.S. economy. And I again want to stress that this is not an hypothetical proposal. Many of the world's leading central banks operate such facilities. The Fed has had its ON RRP facility in place since 2013. Indeed, the Fed even implemented a repo facility (called the FIMA repo facility) in March of this year where foreign central banks can borrow funds at 25bp above IOER by presenting U.S. Treasury securities as collateral. The same type of facility set up for domestic purposes (ideally with Treasury support) could simultaneously help the FOMC achieve interest rate control, shrink the size of its balance sheet, and prevent unnecessary violent disruptions in the treasury market by setting a corridor around treasury yields at different maturities. The size of the corridor could ultimately be adjusted to help achieve yield curve control if desired. But this is a separate issue, so let me end here. Please feel free to comment below.