Blog by David Beckworth, an associate Professor of Economics at Western Kentucky University, has recently launched an awesome podcast that features well-known economists, many of whom are featured on this list. David’s blog highlights is a great place to start to learn about specific economic events.
Some economists advocate nominal GDP targeting as an alternative to the Taylor Rule. These arguments are largely based on the idea that nominal GDP targeting would require less knowledge on the part of policymakers than a traditional Taylor Rule. In particular, a nominal GDP targeting rule would not require real‐time knowledge of the output gap. We examine the importance of this claim by amending a standard New Keynesian model to assume that the central bank has imperfect information about the output gap and therefore must forecast the output gap based on previous information. Forecast errors by the central bank can then potentially induce unanticipated changes in the short‐term nominal interest rate, distinct from a standard monetary policy shock. We show that forecast errors of the output gap by the Federal Reserve can account for up to 13% of the fluctuations in the output gap. In addition, our simulations imply that a nominal GDP targeting rule would produce lower volatility in both inflation and the output gap in comparison with the Taylor Rule under imperfect information.
Many of you may have seen this article before since it has been a working paper for many years now. I am glad to finally get it published.
Ten years after the financial crisis there is a new appreciation for the role household debt and financial fragility play in the business cycle. As a result, policymakers are looking for tools to promote financial stability. A number of recent studies claim that nominal GDP (NGDP) targeting is just such a tool. For it can theoretically reproduce the distribution of risk that would exist if there were widespread use of state-contingent debt securities. This paper empirically test this view by exploiting an implication of the theory: those countries whose NGDP stayed closest to its expected pre-crisis growth path during the crisis should have experienced less financial instability. This paper constructs an NGDP gap measure for 21 advanced economies to test this implication and finds there is a meaningful role for NGDP in promoting financial stability.
There are a lot of other interesting papers in this Cato Journal that were presented at the conference. So take a look. For those interested, here is the video of my panel at the conference:
P.S. In my last post I asked if the Fed's floor system was about to fold. Well, the answer is no, for now. My concerns about overnight interest rates rising above the IOER have faded as they have for the most part converged back to the Fed's target interest rate range. I also said I would outline in the next post how the Fed could transition to a symmetric corridor symmetric corridor if this collapse was imminent. That promise is still good, but on hold for now. I am doing some more reading and thinking on this topic and will return to it.
Last December, I participated in an AEI event where I made the case that the Fed's current floor operating system could collapse into a corridor operating system fairly soon. My argument was that even without a significant reduction in the supply of reserves, a large shift in the demand for reserves could be sufficient to move the Fed off the perfectly elastic or 'flat' portion of the bank reserve demand curve. The Fed, in other words, could have a relatively large balance sheet and still end up in a corridor operating system.
Graphically, such a development is depicted in the figures below. The figure on the left shows a floor operating system with a large supply of reserves on the flat portion of demand curve. In this system, the IOER is both the target and overnight interest rate. The second figure on the right shows what I imagined could be happening. The demand for reserves was shifting outward because of new regulatory requirements and the supply of reserves was shifting inward as the Fed began shrinking its balance sheet. As depicted, these actions together would push the Fed off the flat portion of the reserve demand curve. In turn, this would cause overnight rates to rise above the IOER and end the Fed's floor system.
When I brought this up late last year it was pure speculation on my part, but it was informed by the Senior Financial Officer Survey and the Fed's balance sheet reduction plans. In my AEI talk, I told George Selgin, who dislikes the floor system, that if this comes to fruition Christmas will come early for him.
Well, Christmas did not come early for poor George. He may, however, get a late gift from Santa Claus as there is some evidence my prediction may be coming true. Jeff Cox reports that interbank interest rates are rising above the IOER rate. The figure below shows the overnight bank financing rate (OBFR), the new and improved interbank interest rate measure, has started rising above the IOER. The old federal funds rate (FFR) has been above the IOER for almost a month. (The closely-related overnight Libor replacement, the Secured Overnight Funding Rate (SOFR) tells a similar story.)
To be clear, this move is small from a broader perspective as seen below. Nonetheless, this rise in both series is part of a longer-term change in their trend, where previously they were consistently below the IOER but now are bouncing above it. If they continue to rise above the IOER, the floor system's days are numbered.
Now the DTCC repo rate has been above the IOER for awhile, but the BNY repo rate has not. This is relatively new. And like the interbank rates, the repo rates collectively have been gone from trending below the IOER to bouncing above it as seen below. Again, if this upward movement is sustained the floor system will fold.
Now, with all that said, there is something of a puzzle here: there has been no revival in interbank lending. One would expect, all else equal, that a rise in interbank interest rates above the IOER to spark some interbank lending. Instead, it appears interbank lending has been flat to declining:
One possible resolution to this puzzle is that banks are lending to the overnight treasury repo market rather than to each other. The overnight yield is slightly higher in this market and according to the Fed's H8 database there has been an explosion of reverse repo activity as seen below.
Maybe part of the new normal is that the treasury repo market has permanently displaced interbank lending. In any event, these developments all point to some big changes taking place that could force the Fed back to a corridor system.
If that is the case, I would recommend the Fed get ahead of this transition and intentionally guide itself to a symmetric floor system like the one in Canada. In my next post, I will offer some practical suggestions for making this journey.
P.S. The technical definition for the "federal funds sold and reverse repos" in the H8 database is as follows: "Includes total federal funds sold to, and reverse RPs with, commercial banks, brokers and dealers, and others, including the Federal Home Loan Banks (FHLB)." So maybe part of the explanation for the lack of interbank lending is that bank are lending indirectly to each other via the repo market.
Over the past decade, inflation has persistently undershot the Fed's inflation target. The Fed's preferred measure of inflation, the core PCE deflator, has average 1.56 percent over this time compared to a target of 2 percent. The Fed officially begin inflation targeting in 2012, but was implicitly targeting 2 percent long before that time. So below-target inflation has been happening for close to a decade and for many observers it is a mystery.
There have been a spate of articles as to why the Fed has not been able to hit its inflation target. Some have wondered if the Fed really understands or even controls the inflation rate. Even Fed officials have been perplexed by the low inflation since it cannot be explained by their Phillips curve models. As a result, they sometimes attribute the persistently low inflation to developments such as falling oil prices, demographics, global competition, changes in labor’s share of income, safe asset shortage, and even the rise of Amazon.
These explanations, however, are not satisfactory since the Fed should be able to determine the inflation rate over the medium to long-run. That is, the Fed should be able to respond over time to developments that might cause inflation to drift off target. The Fed should be, in theory, the final arbiter of the trend inflation rate.
So why has inflation been so low? In my view, the answer is simple: the Fed is getting the inflation it wants. There is no mystery. One does not get a decade of trend inflation that is below target by accident. Instead, revealed preferences tell us inflation is where it is because the FOMC allowed it to be there. Put differently, the Fed has chosen not to fully offset the shocks and secular forces listed above that have pushed inflation down. This is a policy choice.
Fed officials and others may disagree, but the revealed preference argument is hard to ignore. Moreover, there are other reason to believe that the low inflation is, in fact, the desired outcome of the FOMC. They are presented below.
SEP Core Inflation Forecasts
The first reason to believe the low inflation is a desired outcome comes from the FOMC itself. The FOMC's Summary of Economic Projections (SEP) provides a central tendency forecasts for core PCE inflation. The FOMC's definition of the SEP is as follows (my emphasis):
Each participant’s projections are based on his or her assessment of appropriate monetary policy.
The SEP, in other words, reveals FOMC members forecasts of economic variables conditional on the Fed doing monetary policy right. And up until recently, doing monetary policy right was not overshooting 2 percent inflation in the following year, as seen in the figure below. Even now, 2 is still seen largely as a ceiling. There is nothing symmetric about 2 percent in these SEP forecasts.
Most FOMC members, therefore, have treated 2 percent as a ceiling over the past decade. This is "appropriate" monetary policy for them. Keep in mind, that at this forecast horizon most of them also believe they have meaningful influence on inflation. Both of these observations point to the low inflation as a choice.
We directly estimate the Federal Open Market Committee’s (FOMC) loss function, including the implicit inflation target, from the tone of the language used in FOMC transcripts, minutes, and members’ speeches. Direct estimation is advantageous because it requires no knowledge of the underlying macroeconomic structure nor observation of central bank actions. We find that the FOMC had an implicit inflation target of approximately 1.5 percent on average over our baseline 2000 - 2013 sample period.
Fed officials, via their words, actually want 1.5 inflation on average. And shocker of all shockers, they are very close to getting that just that rate of inflation since 2009.
The Neel Kashkari Counterfactual
The third reason to believe low inflation is a desired outcome comes from imagining a counterfactual FOMC. Imagine a FOMC that has twelve members that are all clones of Neel Kashkari, as seen below. In this FOMC, where interest rates were not raised over the past few years--and maybe even lowered--do we really think inflation would be the same? I find that hard to believe.
To be clear, I do think there are important secular forces pushing down trend inflation, like the demand for safe assets. But again, the Fed should be able to offset such pressures if it chose to do so. The real question, then, is why the Fed has settled for trend inflation near 1.5 percent. That is a question for a different post. This post is simply a retort to all those who think the low inflation is a mystery. Folks, it is not a mystery. It is a choice.
It is worth nothing that this choice is actually more than a choice for trend inflation. It is implicitly a choice for lower trend aggregate demand (AD) growth. As seen below, aggregate demand growth was averaging 5.6 percent in the decades before the crisis. Since the recovery started, it has averaged about 3.6 percent. That is a 2 percentage point decline in the trend. The red line in the figure shows what a naive autoregressive forecast would have predicted over the past decade conditional on past nominal expenditure history. There has been a sizable AD shortfall.
In my view, it is this dearth of robust aggregate demand growth rather than the low inflation that is a problem. The slowdown in AD growth has arguably contributed to problems like hysteresis and populism. If so, this policy choice has been costly.
The change is being attributed to growth concerns and a weakening of financial markets. I do not want to go through all the indicators supporting the Fed's worries, but I do want to see whether Nominal GDP (NGDP) lends support to this decision. As many readers of this blog know, I believe that properly evaluated NGDP growth is probably the best indicator of the stance of monetary policy. Okay, so what does it say?
To answer that question, I like to look at a measure called the 'sticky forecast' growth path for NGDP and compare it to the actual level of NGDP. In this note, I show the spread between these two measures--the NGDP gap--provides a good measure of the stance of short-run macroeconomic policy. Here is the intuition for the metric:
The idea behind the sticky forecast path for NGDP is twofold. First, the public makes many economic decisions based on a forecast of their nominal incomes. For example, households may take out a 30-year mortgage based on an implicit forecast of their nominal income over this horizon. The actual realization of nominal income may turn out to be very different than expected, but the households may not be able to quickly adjust their plans given sticky debt contracts and other commitments that constrain them. Therefore, the consequences of previous forecasts are often binding on them and slow to change even if their nominal income forecasts have been updated. Second, in addition to these old forecasts and decisions whose influence lingers, new forecasts and new decisions are being made each quarter for subsequent periods that will also have lingering effects. Together, this means future periods have many overlapping and different forecast applied to them that only gradually adjust.
Given the public's expectations of nominal income, the sticky-forecast path of NGDP can be viewed as the neutral level of NGDP. Unlike the unobservable u*, r*, and y*, this neutral measure is simply a weighted quarterly forecast of nominal income found in the Survey of Professional Forecasters. There is no need for guesswork. The details of its construction are in the note, but it is worth mentioning that the NGDP gap created by this measure is remarkably similar to many measures of slack. I should also note that I constructed this measure using the IMF NGDP forecasts for 21 countries in a forthcoming paper (working paper version) and find it works well cross country too.
Here are what the sticky forecast and actual NGDP paths looks like:
And here is the NGDP gap--the percent difference between the two series:
This NGDP gap shows a standard story: aggregate demand growth overheated some in the late 1990s and to a lesser extent in the early-to-mid 2000s followed by a sharp collapse in 2008. A slow recovery followed that stalled around 2015-2016 and then started rising again. Currently, the NGDP gap is slightly below the neutral level of zero percent. This graph suggest it was appropriate for the Fed to pause on rate hikes this week. It also indicates, however, that the Fed arguably should not have started raising rates in 2015.
While the NGDP gap provides a nice cross check on the stance of monetary policy, it can also be used in an explicit monetary policy reaction function. Here is one I created:
Here it is a market interest rate, the first term measures the gap between the forecasted and targeted NGDP growth rates over the next year, and the second term is the NGDP gap as noted above. The 1-year treasury yield is used for it and the 1-year NGDP growth forecast comes from the Survey of Professional Forecasters. The NGDP target is set to 5.5 percent for 1985-2008 and 4 percent for 2009-2018 to reflect the actual trend NGDP growth rates experience during those times. The figure below plots the rule for different values of the coefficients:
Again, we see that if anything, the prescribed target interest rate is a little below the actual one suggesting the Fed's pause is appropriate. So overall, a smart move by the Fed and arguably one that is overdue.
Finally, I want to share one more application of the NGDP Gap that I have made before. The NGDP Gap does a decent job explaining nominal wage growth as seen in the figure below. The red dots show the portion of the scatterplot attributed to past few years. While the R2 is 65 percent for the overall sample, the sample since 2016 has an R2 of 85 percent. Here is hoping the Fed starts paying more attention to NGDP as a cross check for assessing the stance of monetary policy.
The December 2018 FOMC minutes are out and reveal members continue to discuss the potential long-run frameworks for monetary policy implementation. Their discussion as to whether they should keep their current floor operating system or move to a corridor operating system can be illustrated using the figure below:
The FOMC likes the floor system since it separates the size of the Fed's balance sheet from the setting of its target interest rate. This added flexibility is possible because the reserve supply schedule is on the horizontal part of the reserve demand curve as seen above. Here, banks will take all the reserves sent their way--killing off interbank lending--as their demand for reserves is perfectly elastic. The floor system puts the reserve schedule back on the downward slopping part of the reserve demand curve. That creates an opportunity costs for reserves and resurrects interbank lending.
Recall that the floor system is a byproduct of the crisis. It was part of the unconventional monetary policy actions taken during that time. Consequently, the Fed is now discussing how to normalize its operating system. As I have argued in a recent paper and in various blog posts, I prefer the Fed leave the floor system and move to a symmetric corridor system. In my view, the political an economic costs exceed any benefits of a floor system.
I do not want to rehash these arguments, but I do want to respond to a claim made by the FOMC members as reported in the December 2018 minutes. Specifically, the FOMC claims there will be much greater interest rate volatility under a corridor system. Here is the relevant part (my stress):
Reducing reserves close to the lowest level that still corresponded to the flat portion of the reserve demand curve would be one approach consistent with the Committee's previously stated intention, in the Policy Normalization Principles and Plans that it issued in 2014, to "hold no more securities than necessary to implement monetary policy efficiently and effectively." However, reducing reserves to a point very close to the level at which the reserve demand curve begins to slope upward could lead to a significant increase in the volatility in short-term interest rates and require frequent sizable open market operations or new ceiling facilities to maintain effective interest rate control. These considerations suggested that it might be appropriate to instead provide a buffer of reserves sufficient to ensure that the Federal Reserve operates consistently on the flat portion of the reserve demand curve.
Well, if there were any doubts as to where the FOMC is leaning in this debate over operating systems this paragraph should put the doubts to rest. FOMC members apparently love their flat reserve demand curves. So much so, they cannot handle the imagined horrors of interest rate volatility under a corridor system.
Yes, the horrors of interest rate volatility in a corridor system. I mean, how can central banks like the Bank of Canada (BoC) impose such a cruel system on their financial system? How dare the BoC leave the peaceful sanctuary of a floor system and move to the interest rate jungle of a corridor system! Just look at the all the interest rate volatility they are imposing on the Canadian financial system.
Oh wait, the BoC corridor system actually looks okay. Yes, there is some interest rate volatility for the overnight repo rate relative to the BoC's interest rate target, but the repo rate stays well within the corridor bounds.
Maybe the FOMC means interest rate volatility in a corridor system compared to a floor system, like the one it runs. After all, the FOMC is a true believer in its own operating system. The FOMC did say in the minutes that the "efficient and effective implementation of monetary policy" most likely requires providing "a buffer of reserves sufficient to ensure that the Federal Reserve operates consistently on the flat portion of the reserve demand curve."
With such confidence in their own floor system, it must be that the FOMC members are indeed thinking of the interest rate stability in their system. Right?
Oops, maybe not. Overnight U.S. repo rates do not look so stable compared to Canada. Maybe the scales of the above figure overstate the volatility of repo rates in the United States? How about comparing the actual spread between the overnight repo rate and target rate for the two countries and their different operating systems?
Okay, maybe the corridor system is not so bad after all. Maybe the FOMC is thinking of a return to an asymmetric corridor system like the one that existed pre-2008. There might be more interest rate volatility in returning to that system, but most advocates of a move to corridor system are not advocating such a move. Instead, we want a move to symmetric corridor system where the IOER pins down the lower bound and the discount rate anchors the upper bound.
I am glad the FOMC is debating the future of its operating system. My hope is that Fed does not get blinded by its own experience with an asymmetric corridor system and instead looks elsewhere in the world for understanding how a symmetric corridor system can work.
I have a new working paper titled "Better Risk Sharing Through Monetary Policy? The Financial Stability Case for a Nominal GDP Target". I presented this paper at the recent Cato Monetary Policy Conference.
Here is the abstract:
A series of papers have shown that a monetary regime targeting nominal GDP (NGDP)
can reproduce the distribution of risk that would exist if there were widespread use of state contingentdebt securities (Koenig, 2013; Sheedy, 2014; Azariadis et al., 2016, Bullard and DiCecia, 2018). This paper empirically evaluates this view by exploiting an implication of the theory: those countries whose NGDP stayed closest to its expected pre-crisis growth path during the crisis should have experienced the least financial instability. This paper constructs an NGDP gap measure for 21 advanced economies that is used to test this implication. The results strongly suggest that there is a meaningful role for NGDP in promoting financial and economic stability.
And an excerpt:
The key insight of Koenig (2013), Sheedy (2014), Azariadis et al. (2016), and Bullard and DiCecia (2018) is that in a world of incomplete markets where there is non-state contingent nominal contracting, an NGDP target can reproduce the risk distribution that would occur if there were complete markets and state contingent nominal debt contracting. An NGDP target, in other words, can make up for the lack of insurance against future risks that could affect debtors’ ability to repay their debt. Conversely, an NGDP target can also make up for the lack of insurance against potential returns a creditor might miss out on because their funds are locked up in a fixed-price nominal loan. Bullard and Dicecia (2018) show that this result holds even when the heterogeneity among debtors and creditors modeled approximates that of the actual income, financial wealth, and consumption inequality in the United States. They note this makes NGDP targeting “monetary policy for the masses.”
Readers of this blog will know I share many of George's concerns about the floor system that are outlined in his book and I would like to see the Fed move to a symmetric corridor system. The FOMC spent a good portion of its November meeting discussing this issue. My comments at the AEI event, however, were not on the tradeoffs between a corridor and floor system but rather on how close the Fed currently is to a corridor system. There are some indicators that the Fed may not be too far away.
To illustrate my point, consider the figures below. The question I considered is how far the Fed is from transitioning from the figure on the left below to the one on the right? Note, that there are two ways to make this move. First, the supply of reserves (red line) can shift back until it hits the slopping part of the reserve demand curve. Second, the demand for reserves (blue line) can shift out until the slopping part of the demand hits the reserve supply schedule. Or, there can be some combination of both developments. In either case, reserves become relatively scarce, unsecured interbank lending revives, and reserves once again will bear an opportunity costs.
In my remarks I noted that there have been both supply and demand shifts that have moved the Fed closer to a corridor system. (For more details on this simple supply-demand model see my recent paper.)
On the supply side, there has been a reduction in reserves going on since August 2014. Initially, the decline was largely due to the the growth of the Treasury General Account (TGA) and overnight reverse repos displacing reserves. Since October 2017 it also been the result of the Fed decreasing reinvestment of principal payments on its asset holdings. Collectively, these actions have led to a decline of about $1 trillion dollars in reserves over the past four years. Reserves are now just over $1.7 trillion. This can be seen in the figure below:
On the demand side, there have been new regulations and apparently a general rise in risk aversion that has led to a higher demand for reserves than existed before 2008. The main regulatory development is the liquidity coverage ratio (LCR) which requires banks to hold enough liquid assets to cover 30 days of withdrawals. The LCR treats bank reserves and treasuries as the top "high quality liquid assets" (HQLA) in meeting this requirement.
For most of the past decade the IOER rate has been higher than that yield on overnight treasury repo rates creating an elevated demand for reserves to meet the HQLA. As seen below, however, this relationship has changed with measures of treasury repo rates bouncing around--and increasingly above--the IOER rate. As I have explained elsewhere, this is largely due to President Trump's budget deficits sharply increasing the supply of treasury bills.
This reversal in relationship should mean, all else equal, that that banks would be indifferent between holding bank reserves and treasury bills. However, the most recent Senior Financial Officer Survey indicates that even in this "constellation" of interest rates banks still want to hold a lot of reserves relative to the pre-crisis levels. Specifically, the banks indicate they would want to hold $617 billion in reserve balances. Now the survey only covers banks that currently hold two-thirds of bank reserves. Consequently, extrapolating this number out another third puts the total desired reserve balances at $927 billion. This implies a large rightward shift of the reserve demand curve compared to pre-crisis demand for reserves.
If the Fed continues to reduce its balance sheet by $50 billion, then it would reach a point of reserve scarcity by early-to-mid 2020. This can be seen in the figure below.
Put differently, the Fed could be back in a corridor system as early as 2020 if it were to put its balance sheet reduction on autopilot. Note that balance sheet would permanently larger--even after accounting for currency or NGDP growth--at a size near $3 trillion.
Interestingly, the Senior Financial Officer Survey also indicates that if the Fed really wanted to shrink the supply of reserves to levels comparable to the pre-crisis system there is a way. All that is needed is for the price to be right.
Specifically, the survey asked the banks whether they would decrease their reserve holdings further if comparable short-term interest rates were to rise above IOER at 5, 25, and 50 basis points (bps). The percent of banks that answered yes were 8, 46, and 49 percent, respectively. So somewhere between 5 and 25 bps, there is a threshold were a large number of banks would find it worthwhile to depart with reserves and hold treasury bills. Everyone has a price!1 In this case, the time to a corridor system might take longer.
The big takeaway, for me, is that the Fed may not be that too far away from a corridor system that maintains a moderately-sized balance sheet (owing to the residual demand for bank reserves). For reasons outline in my previous post, however, I would prefer a return to a corridor system with fewer reserves and the Senior Financial Officer Survey suggests it is possible.
1 What is really interesting about this finding is that Dutkowsky and VanHoose (2018) find in a theoretical model that banks would stop holding large amounts of excess reserves once the spread crossed 6 bps. This theoretical results fits nicely with the survey showing their is a threshold between 5 and 25 bps.
As readers of this blog know, I have an interest in the Fed's operating system. This interest has culminated in a new paper where I look at the consequences of the Fed moving from a corridor system to a floor system in 2008. In particular, the paper looks at what the change has meant for bank portfolios and, as a result, financial intermediation provided by banks. The paper concludes with some policy recommendations. I would also note that George Selgin has just released a new book on this topic. My hope is that these projects will help inform the conversation over what operating system the Fed wants as it continues to normalize monetary policy.
Paper Outline So what does my paper have to say? It starts by laying out the standard arguments for a floor system:
The central idea behind this move was to remove the opportunity cost to banks of holding excess reserves by offering the banks a deposit rate at the Fed—the IOER rate—that was equal to or above short-term market interest rates. This favorable return was to sever banks’ incentive to rebalance their portfolios away from excess reserves toward other assets. The IOER rate was also to put a floor under short-term interest rates so as to align them with the Fed’s desired interest rate target. Together, these two facets of the floor system would allow the Fed to use its balance sheet as a tool of monetary policy while still maintaining interest rate control.
In this new operating system, the stance of monetary policy was no longer set by a market interest rate but by an administrative interest rate: the IOER rate. The stance of monetary policy also was no longer tied to the supply of reserves. Instead, it was linked to the quantity of reserves demanded by banks, which the Fed influenced through changes to IOER. Specifically, the Fed set the IOER rate high enough that banks’ demand for reserves became perfectly elastic with respect to the federal funds rate. As a result, changes in the quantity of reserves supplied led to identical changes in the quantity demanded, other things being equal.
The Federal Reserve, in short, went from an operating system in which monetary policy was transmitted through open market operations to one in which it is transmitted through the IOER rate. The Fed’s operating system changed from one in which money, in the form of reserves, mattered for monetary policy to one in which money has been “divorced” from monetary policy.
Okay, so why does this "divorce" matter? For advocates of a floor system the answer is simple:
This divorce from money is seen by many observers as the key advantage... because it gives the Fed the freedom to use its balance sheet independently of its desired interest rate target. The Fed, for example, can now sharply increase the supply of reserves in response to a liquidity crisis without causing a decline in its targeted interest rate.
Skeptics of the floor system, on the other hand, see this divorce as more problematic:
Others, however, see this divorce as creating an operating system that impairs the transmission mechanism of monetary policy... These observers’ understanding starts with the standard assumptions of a floor system. First, a floor system requires the IOER rate to be set at least equal to short-term interest rates. This removes the opportunity costs to banks of holding reserves and thereby keeps their demand perfectly elastic with respect to other short-term interest rates...
[This] can lead to a rebalancing of bank portfolios that causes the supply of loans to be lower than it would have been otherwise. Banks lend as long as the marginal cost of funding is less than the risk-free marginal return on bank lending. In the Fed’s floor system, the IOER rate sets the marginal funding cost. Consequently, by setting the IOER rate higher than other short-term interest rates, the Fed has raised the marginal costs of funding and narrowed the gap between these costs and the risk-free marginal return on bank lending. All else being equal, the narrowing of this gap implies a relative reduction in the supply of loans and therefore a relative decline in the money supply.
Are these worries merited? My paper provides an empirical look at bank portfolios before and after the advent of the floor system to see if (1) there have been big structural shifts in bank balance sheets consistent with the critics claims and (2) whether such shifts can be attributed to the Fed's floor system.
Empirical Evidence On (1) I start with the following figure. It shows the share of bank assets allocated to loans and to safe assets (defined as the sum of cash, treasury, and agency asset holdings). Unsurprisingly, these series are almost mirror images of each other over both cyclical and structural time horizons. They tend to move in opposite directions during recessions--the grey bars--and over longer periods. At the advent of the floor system in late 2008, the loan share began declining as the safe asset share started rising. This change has been sustained and only recently has started reversing:
If we break the safe assets apart into its subcategories, we see that that this tight link has been historically driven by movements in treasury and agency investments corresponding with changes in the loan share. Since 2008, the driving force behind the tight link became cash not treasury and agency investments:
If we zoom in and use two scales, this apparent structural change is even clearer. Cash shares and loan shares become mirror images of each other:
Something big happened in 2008 that continues to the present that caused banks to allocate more of their portfolios to cash assets and less to loans. While the financial crisis surely was a part of the initial rebalancing, it is hard to attribute what appears to be 10-year structural change to the crisis alone. Instead, it seems more consistent with the critics view that the floor system itself has fundamentally changed bank portfolios allocation.
That takes us to question (2). The following figure shows the loan share of bank assets plotted against the spread between the IOER rate and the overnight dollar Libor rate. This IOER-Libor spread is negatively and strongly correlated with the loan share. This suggests banks invested less in loans when the relative return on bank reserves rose and vice versa.
Conversely, banks appear to have allocated more to cash assets with the IOER-Libor spread rose and vice versa:
The paper goes on to more carefully test these relationships using two-stage least squares regressions that control for endogeneity issues and other confounding influences. Moreover, the paper also provides a further breakdown of this relationship among foreign banks, large domestic banks, and small domestic banks. Collectively, the regressions point to a strong causal effect running from the IOER-Libor spread to the allocation of bank assets.
The Fed's move to a floor system, then, does seem to have influenced the amount of financial intermediation provided by banks to the private sector. The end of the paper provides some counterfactual exercises, including the following figure
This counterfactual shows the supply of bank loans would have been, at its peak, as much as $2 trillion higher in 2014. By mid-2018, with the IOER-Libor spread shrinking, it was closer to a $1 trillion shortfall. Now to be fair, other factors such as fintech and new regulations also probably contributed to below trend growth in bank lending since the crisis. Still the evidence strongly suggest that financial intermediation to the private sector through banks declined because of the floor system.
The Fed's floor system, in short, has caused banks to increase their investment in the Fed at the expense of investing in the private sector. The question, then, becomes whether the Fed is any better than banks in allocating this credit. Put differently, is the Fed as a financial intermediary--funding short and lending long--really providing a better financial service than would have been provided by the private sector financial firms? It is not obvious to me that the answer is yes.
What is obvious to me is that the Fed's floor system creates a whole set of other problems. First, as a profitable financial intermediary, the Fed is setting itself up for political shenanigans. Recall Congress taping into the Fed's capital surplus in 2018 and 2015. Though these transfers were relatively small and to some extent accounting gimmicks, they show how tempting a large, profitable Fed balance sheet can be to Congress. Second, the floor system effectively destroyed unsecured interbank lending. This market provided useful interbank monitoring and price discovery that no longer exists. Bringing this market-based monitoring back would be a nice addition to the bank regulator's existing toolbox. Third, the floor system forces the Fed to take safe asset collateral off the market which impairs other parts of the financial system. Only recently has President Trump's budget deficits begun to fill this hole. Fourth, the floor system can create bad optics for the Fed via the appearance of large 'subsidies' to large and foreign banks. In this era of populist politics, the Fed should be worried about the dangers to its independence this image could create. Finally, as noted above, I worry the Fed is a less effective financial intermediary than the private sector.
In short, I see the costs exceeding any benefits from the floor system. That is why I advocate a return to a corridor system, but this time with the IOER explicitly setting the lower boundary. That is, the IOER would still be around, but it would be set lower than overnight market interest rates. For the past decade it has been for the most part above them. More details are in the paper.
P.S. Here is a panel from the recent Cato Monetary Policy Conference on this issue. Great comments from Stephen Williamson, George Selgin, and Peter Ireland. Josh Zumbrum was the moderator.
Andrew Metrick of Yale University interviewed former Fed Chair Janet Yellen today. It was an interesting discussion and one where they talked about, among other things, what changes the Fed could bring about in light of the recently announced strategies, tools, and communication review to be held in 2019. Janet Yellen said her idea for reform "has much in common with NGDP targeting". Her response can be seen in the video below:
A Conversation with Janet Yellen, Former Chair of the Federal Reserve - YouTube
Glad to see her endorse a NGDPLT-like monetary regime.
Tim Duy reports that r-star, which rose to prominence over the past few years, is experiencing a Caesar-like betrayal at the Fed:
The Federal Reserve’s “r-star” has gone full supernova. New York Federal Reserve President John Williams, its key proponent, made clear in a speech late Friday that the neutral interest rate is no longer a guiding star for monetary policy. This means a federal funds rate in the range of what is considered neutral has no special significance as far as policy is concerned...
Williams’s attachment to r-star cannot be overstated. At a professional level, it has been a key element of his research agenda. As recently as May he said that for “the moment, r-star continues to shine brightly, guiding monetary policy, but hold steady, low on the horizon.” The moment quickly passed. Last week, he tossed aside the metric, saying that it has “gotten too much attention in commentary about Fed policy.” A remarkable shift after just two 25-basis-point rate increases since his May comments...
Williams’ speech marks the end of a transition in policy away from explicit forward guidance. It began this past August with Fed Chairman Jerome Powell’s Jackson Hole speech in which he noted the uncertainty surrounding estimates of key variables like the neutral interest rate. Fed Governor Lael Brainard pushed this point further in a subsequent speech, adding further uncertainty by differentiating between short- and long-run neutral. It continued in the September Federal Open Market Committee statement with the removal the description of policy as “accommodative.” And it ends with the primary proponent of the r-star concept — Williams — throwing it into the trash bin of crisis-era policy artifacts.
One is tempted to say "It was good knowing you r-star". However, r-star will still be around in all the models used by the FOMC and Fed staff. Just look at, for example, the policy rules on the Board of Governor's website or in its annual report. The reported change, as I see it, is more a move toward less explicit reliance on it. Implicitly, r-star will still be important to an FOMC that relies on the Phillips curve thinking in making its decisions.
Still, these developments do indicate there is some movement towards looking at other indicators as I noted in recent post. There I suggested one useful metric the FOMC could add to its lists of monetary policy indicators is the gap between a stable benchmark growth path for nominal GDP and its actual value. I outlined in this note several ways to create this metric and note that it is in the spirit of a NGDP level target without actually adopting one.