Fonds im Fokus


31.05.2017 | 11:33

UBS: The Mechanics of the Fed's Balance Sheet

The Federal Reserve's plan to runoff its balance sheet is not yet final, nor is its long-run framework for policy implementation. But tentative assumptions allow us to make some projections. A baseline is three years of contraction to just under $3.4 trillion.

The Federal Reserve's balance sheet is in the spotlight. It is too soon to be definitive, but the balance sheet could runoff for less time and end up larger than many commentators suspect. Under plausible assumptions, the balance sheet might contract for only three years to a size of nearly $3.4 trillion. We stress, however, that there is much uncertainty to these projections, because many of the critical details of the balance sheet runoff and the long-run framework for monetary policy are as yet unclear. The details and the mechanics can be a bit dull, but taking them seriously is necessary to understand the broader implications. The market and macro effects of the balance sheet rely on a host of assumptions that we will lay bare here.

The first point to make is that balance sheets balance. Stated that way, the tautology is clear, but the fact is critical to projecting the balance sheet. We will model assets, liabilities, and capital. The Fed's Large Scale Asset Purchases (LSAPs) boosted the size of the balance sheet. In sharp contrast, historically, the Fed's balance sheet was liability driven; currency growth determined the growth of the balance sheet, and purchases of Treasuries matched that growth. Liabilities determined the size of the balance sheet. Perhaps unintuitively, that statement is a useful definition for a "normalized" balance sheet. As will be clear in the projections, at some point in the future, the size of the Fed's balance sheet will again be determined by liabilities; we will identify that point as normalization.


The main assets on the Fed's current (post-crisis) balance sheet to consider are Treasury securities and MBS. As long as the Federal Reserve decides to roll over these securities, total assets on the balance sheet will not change much. When the FOMC decides to let Treasury securities mature, the effect is deterministic. The FRBNY publishes the CUSIP of every Treasury security held by the Fed, so the maturity profile is known. In contrast, the potential runoff of the MBS portfolio is uncertain. Without reinvestment, when payments of principal on the underlying securities are made, the face value of the MBS holdings would decline. The amount of prepayment, however, varies, in part because of rising or falling interest rates. Falling interest rates tend to lead to more mortgage refinancings, accelerating MBS prepayments. Rising interest rates tend to slow prepayments.


Liabilities are a bit more complicated. We focus on four key liability items: currency, reserves, and reverse repurchase agreements (RRPs), and the Treasury's account at the Fed (the TGA). Prior to the Crisis, the only liability of note was currency. A critical fact is that the Federal Reserve supplies currency completely elastically, fully satisfying any demand from the public. When a bank requests currency (for its vaults, ATMs or customers), the Fed ships the currency to the bank and debits the reserves in the bank's account, dollar for dollar. An increase in currency reduces reserves. A standard, but sadly wrong, assumption is that currency should grow in line with U.S. nominal GDP. The Fed has estimated that over half of currency is held outside of the US. Explaining exactly what does drive currency growth is difficult, whether inside the US or outside. Domestically, as a quick check, even half of the nearly $1.5 trillion of outstanding currency amounts to roughly $2,250 for each man, woman, and child in the country—a surprisingly large amount—so the standard view of currency is likely incorrect. Outside the US, financial and political turmoil abroad often boosts demand for a safe store of wealth, but forecasting such developments is not possible. Reserves are the deposits of money that banks have with Federal Reserve Banks and require two critical points to be made. First, the banking system cannot by itself change the quantity of reserves. Banks do not "put money back to the Fed," nor can those reserves "flow out into the economy." A bank may wire funds to a different bank, but the aggregate quantity of reserves will not change. Second, it is transactions by the Fed (buying or selling) that change the aggregate quantity of reserves. When the Fed unwinds its balance sheet, reserves will decline.

The RRP facility is a relatively new tool for controlling short-term interest rates. Prior to the Crisis, the Fed allowed foreign official institutions to invest cash in repo transactions, but the tool was small and inconsequential. With the use of RRPs to control short-term rates, the volume of RRPs on the Fed's books has increased significantly and become much more volatile. The Treasury General Account—or TGA—is the account the US Treasury holds at the Fed. Since 2015, the Treasury has been increasing this account in order to maintain liquidity to cover at least one week's worth of outflows. Prior to the financial crisis, this account was only $5 billion most days.


To round out the balance sheet, we have Reserve Bank capital. In contrast to real banks, Reserve Bank capital does not convey any controlling authority, regardless of conspiracy theories. The history and evolution of Reserve Bank capital is the stuff of storytelling but not analytics.

You will find the full report here.