What Should the Fed Own?

“[T]he Fed might accommodate a sudden spike in [demand for safe assets] by selling some of its T-bill holdings and using the proceeds to buy longer-term Treasuries.” Greenwood, Hanson and Stein, The Journal of Finance, August 2015.

 “[T]he Committee would be prepared to use its full range of tools, including altering the size and composition of the balance sheet, if future economic conditions were to warrant a more accommodative monetary policy than can be achieved solely by reducing the federal funds rate.” Federal Open Market Committee, “Addendum to the Policy Normalization Principles and Plans,” June 14, 2017.

The Federal Reserve began to consider just how far its balance sheet consolidation should go well before the tapering actually began nearly a year ago. Earlier staff analyses pointed to a gradual runoff of long-term debt that could take years to reduce Fed assets to a new long-run equilibrium. More recently, market observers have speculated about an early end to consolidation that would result in a higher steady-state level.

Yet, as a recent Wall Street Journal article highlights, policymakers and analysts have devoted less attention to the mix of assets that the Fed should select once the balance sheet shrinks to its long-run equilibrium and policymakers allow it to expand slowly—say, in line with the increase of demand for currency.

In this post, we argue that the Fed should aim in normal times—when the economy is expanding and absent any financial strains—for a portfolio that has minimal liquidity, maturity and credit risk. In practical terms, this means that their portfolio should be composed largely of Treasury bills and short-term notes, with an average maturity that is very short.

Let’s take a step back and ask where things stand today. The chart below depicts the evolution of the Fed’s domestic securities holdings over the past 15 years. Three trends stand out. The first is the massive expansion associated with the Fed’s large-scale asset purchases during and after the financial crisis. Second is the shift away from Treasury bills (shaded in black) and toward holdings of long-term Treasury debt (in red) and mortgage-backed securities (MBS, in blue). While these expanding instruments all bear a federal guarantee—virtually eliminating credit risk—they add materially to the liquidity and maturity risk of the Fed’s portfolio. Third is the shrinkage over the past year. What is difficult to see here is that, as of the week ending August 22, 2018 (and for the first time since 2012), the Fed once again started purchasing Treasury bills.

 Federal Reserve holdings of domestic securities (weekly, billions of dollars), 2004-August 29, 2018

 Source: FRBNY  SOMA Historical Data .

Source: FRBNY SOMA Historical Data.

The Fed’s changing role in maturity transformation is summarized in the next chart where we plot the average duration of its Treasury holdings (black line). Duration, which measures the average time until a bond’s owner receives payments weighted by the size of those payments, is a more accurate indicator of a bond portfolio’s interest rate risk than the portfolio’s weighted-average maturity (which treats all the payments as equal). The surges in portfolio duration in 2008 and 2012 have only partly reversed in subsequent years. As a result, the average duration at the end of 2017 is still nearly six years, more than twice what it was in 2007, prior to the financial crisis and the Great Recession.

Duration in years of Fed holdings of Treasury debt (monthly) and the ratio of marketable Treasury debt to GDP (annual, percent), 2002-2017

So, what should be the composition of the Fed’s portfolio in normal times?  To answer this question, we start with two high-level objectives. First, the holdings should be as riskless as possible. And second, the actions of the Fed should distort financial markets as little as possible. Combined, this means holding assets with virtually no liquidity, maturity, or credit risk. Because the federal government issues or guarantees the domestic securities held by the Fed, credit risk already is minimal. But there is plenty of scope to reduce liquidity and maturity risk.

Our conclusion is that, if feasible, the Fed’s portfolio should have an average duration of no more than one year. Short-maturity instruments also tend to be more liquid. Consequently, an “asset-neutral” portfolio composed largely of Treasury bills and short-term notes would achieve the objectives we set. And, it would still permit the Federal Reserve’s System Open Market Account manager to “test the waters” from time to time in other instruments (like long-term Treasuries, inflation-indexed bonds, or agency mortgage-backed securities), both to monitor liquidity conditions and to maintain the infrastructure needed to alter the portfolio quickly should the need arise.

We see three key advantages of such a low-risk portfolio in normal times.

Political economy. First, holding entirely short-term Treasury debt addresses the frequent criticism from Congress and others that the Fed’s financial market interventions overstep its mandate. When the Fed announced large-scale purchases of MBS in 2008, it helped arrest the plunge of housing prices by substituting for the absence of effective private intermediation. These actions increased the portfolio’s maturity and liquidity risk, but the need to halt the crisis and promote recovery warranted aggressive use of the Fed’s balance sheet.

Nearly a decade later, with U.S. real housing prices on average 35 percent above their 2012 trough, there is little justification for continued Fed intervention in the market for housing finance. By setting an asset-neutral standard in normal times, a simple, low-risk portfolio would buttress the independence that is needed for the Fed to secure economic stability. President’s Trump’s objections to the FOMC’s interest rate normalization plans make preserving Fed independence all the more important.

Market determination of risk premia. Second, a low-risk portfolio removes the Fed from interference in the determination of liquidity and maturity risk premia, in line with its approach to credit risk. The Federal Reserve Act precludes the Fed from acquiring corporate liabilities; and its holding of foreign government bonds are limited to the safest sovereign issuers, so the central bank does not interfere in the market determination of credit risk. Why not also allow markets to determine the liquidity and maturity risk premia along the Treasury yield curve in normal times? To be sure, investors would still anticipate a Fed portfolio shift to less liquid, longer-maturity instruments in crises or even in recessions, but this implicit central bank backstop would surely do less to distort market prices in normal times than would a Fed portfolio that is always more closely aligned with the maturity distribution of the outstanding Treasury debt.

Maximizing Policy’s Fire Power. Third, by sharply dialing back the risk setting on its portfolio in normal times, the Fed will maximize its potential to counter threats to financial and economic stability as they arise. For example, as the opening citation from Greenwood, Hanson and Stein suggests, were there a sudden surge in demand for safe assets by banks and other intermediaries—as typically occurs in a bank run—the central bank could rapidly increase the supply by exchanging its short-term Treasury bills and notes for less liquid, longer-term instruments. If the shock were to include a threat to housing finance, the acquisitions could include agency MBS (in addition to long-maturity Treasurys).

The same goal of maximizing fire power applies to the use of the Fed’s balance sheet for traditional monetary policy aimed at price and economic stability. Were policy rates to again hit the effective lower bound—something that many FOMC members view as a significant probability in the next downturn—the central bank needs to be prepared to redeploy unconventional policies, including changes to the composition of its balance sheet (see here).

Whether and how the balance sheet mix affects financial conditions and the economy remains a subject of debate (see here, here and here). As mentioned, by substituting for the absence of private intermediation in the midst of a crisis, interventions such as targeted asset purchases of MBS in exchange for liquid Treasury instruments can have a large impact on risk premia. This is an extreme version of the “portfolio balance” effect, the scale of which depends on the substitutability between the two assets exchanged. When the Fed exchanges assets that are closer substitutes—say, a T-bill for a T-bond in normal times—the portfolio balance effect is likely to be weaker.

At the same time, a simple maturity extension of Treasury holdings may alter maturity risk premia by reinforcing the credibility of a Fed commitment to keep policy rates low once they have reached the effective lower bound. Such forward guidance lowers long-term yields (and stimulates economic activity) both by reducing expected future short-term rates and by depressing volatility.

Role of Treasury Debt Management. We see few arguments against the conclusion that the Fed should keep a low-risk portfolio in normal times. However, the feasibility of such an approach depends on the willingness of the Treasury to provide an adequate supply of bills and short-term notes. Of the $15.4 trillion in marketable Treasury debt outstanding as of March 2018, bills account for $2.3 trillion―nearly all of which is being held willingly outside the Federal Reserve. And, given the existing premium for safe, short-term assets—partly reflecting the impact of newly implemented liquidity regulation of financial institutions—the Fed’s need for T-bills would only add to the total demand.

Put differently, we cannot fully separate the Fed’s optimal portfolio mix from the Treasury’s debt management practices. This is not a new issue: Greenwood et al highlight that, as the debt-to-GDP ratio began to surge in 2008, the Treasury sharply extended the average weighted maturity of marketable debt outstanding (see chart above). This shift substantially offset the Fed’s use of its balance sheet to lower the supply of long-term Treasurys to the public (and, thereby, lower their yields). Greenwood, Hanson and Stein show that the recent positive relationship between the volume of outstanding debt and its average maturity is a persistent feature of Treasury debt management since the 1950s (see chart below).

U.S. Treasury debt: weighted-average maturity in years (monthly) of marketable debt and ratio to GDP of debt held by the public (annual, percent), 1952-2017

 Source: Average maturity as in Chart 2, Greenwood, Hanson and Stein,  The Federal Reserve’s Balance Sheet as a Financial-Stability Tool  (original data and update through 2017 kindly provided by these authors); FRED for  ratio to GDP of gross federal debt held by the public .

Source: Average maturity as in Chart 2, Greenwood, Hanson and Stein, The Federal Reserve’s Balance Sheet as a Financial-Stability Tool (original data and update through 2017 kindly provided by these authors); FRED for ratio to GDP of gross federal debt held by the public.

Fortunately, meeting the Fed’s additional need for short-term Treasury instruments in normal times, when the yield curve typically is upward sloping, is likely to be consistent with Treasury’s mandate to minimize the cost of financing the federal government. Still, to maximize its clout, the Fed needs to be the last actor in the game of determining the relative supply of short- and long-term Treasury debt held by the public. For this to occur, Treasury debt management needs to be steady and predictable, with Treasury refraining in the short run—say, for periods of a few years—from large shifts that would offset the stabilization efforts of the Fed. Put differently, Treasury ought not respond quickly to shifts in maturity risk premia along the yield curve. Its “policy reaction” should be far slower than the Fed’s.

Such limited cooperation between the Fed and the Treasury in no way violates the Fed’s independence. Quite the opposite: a framework that commits the Treasury to supply an adequate volume of short-term debt in normal times makes it feasible for the central bank when appropriate to counter financial strains and to stabilize the economy at the effective lower bound. With the Fed starting to consider what its normal balance sheet mix should be, now is a good time to make such long-run arrangements.

To conclude, the arguments are compelling for a Fed portfolio composed of near-riskless assets: not only in terms of credit, but liquidity and maturity, too. To the extent that banks are indifferent between holding reserves at the Fed and T-bills, a low-risk portfolio in normal times makes the magnitude of the Fed’s balance sheet less salient. However, the desirable state-steady size of the Fed’s balance sheet depends on a range of factors—including the potential impact on financial stability. We look forward to learning how the Fed will balance these considerations in judging where to halt the ongoing consolidation.