There is an important U.S. government-sponsored banking system that most people know nothing about. Created by an act of Congress in 1932, the Federal Home Loan Banks (FHLBs) issue bonds that investors perceive as having government backing, and then use the proceeds to make loans to their members: namely, 6,800 commercial banks, credit unions, insurance companies and savings associations. As the name suggests, the mission of the (currently 11) regional, cooperatively owned FHLBs is “to support mortgage lending and related community investment.” But, since the system was founded, its role as an intermediary has changed dramatically.
With assets of roughly $1 trillion, it turns out that the FHLBs—which operate mostly out of the public eye—have been an important source of regulatory arbitrage twice over the past decade. In the first episode—the 2007-09 financial crisis—they partly supplanted the role of the Federal Reserve as the lender of last resort. In the second, the FHLBs became intermediaries between a class of lenders (money market mutual funds) and borrowers (banks), following regulatory changes designed in part to alter the original relationship between these lenders and borrowers. The FHLBs’ new role creates an implicit federal guarantee that increases taxpayers’ risk of loss.
In this post, we highlight these episodes of regulatory arbitrage as unforeseen consequences of a complex financial system and regulatory framework, in combination with the malleability and opaqueness of the FHLB system.
Let’s start with a view of the evolving FHLB balance sheet. The following chart displays the size and composition of the FHLB system’s assets. (Because the individual institutions are jointly and severally liable, we focus on the consolidated financial statement of the system; see Frame and White for a detailed discussion.) “Advances” (in red) are loans to member institutions that are anywhere from overnight to 20 years. Today, these account for roughly two-thirds of total assets. The remainder of the assets are a combination of debt securities of various types (20%), mortgages (6%), fed funds lending (6%), and a variety of miscellaneous categories (6%).
Federal Home Loan Bank System assets (Billions of dollars), 2001-2Q 2019
Source: FRED.
The primary source of FHLB funding is the issuance of short-term notes and of medium-to-long-term bonds. As Gissler and Narajabad describe, the assets are generally longer term than the liabilities, so this entails some maturity transformation. And, like any normal financial intermediary, the FHLBs turn a profit by lending at a rate higher than their cost of funds. In 2018, they report a net interest margin of 0.48%, which yielded a return on assets of 0.32% and a return on equity of 6.18%. Since this is a cooperative, it is the members (banks and the like) who are the shareholders. In 2018, FHLB member institutions received an average dividend yield exceeding 5.5%.
The FHLBs’ assets and liabilities have some extremely important characteristics. First, as a lender, the FHLBs have priority over the claims of virtually all creditors: this includes not only a borrowing bank’s depositors, but the FDIC and the Federal Reserve as well! Second, while their debt is not explicitly government guaranteed, it is pretty close. It not only carries a AAA rating from Moody’s and AA+ from S&P, but the FHLBs also describe 10 reasons to conclude that they have “Strong U.S. government support” (see here). In sum, they are a government-sponsored enterprise (GSE) (although less prominent than Fannie Mae and Freddie Mac); and as a result, for a number of purposes, FHLB liabilities function very similarly to U.S. Treasury debt.
Returning to the chart, we focus on the developments in FHLB advances that correspond to our episodes of interest. First, there is a dramatic rise during the first stages of the crisis in 2007. Second, from 2014 to 2016, the level nearly doubled. Each of these is worth a closer examination.
Starting with the crisis period, Ashcraft, Bech and Frame describe how, when funding strains began to arise in mid-2007, banks turned to the FHLBs for advances. In effect, they were serving as lenders of first resort, diminishing the market discipline typically associated with the penalty rate charged by a central bank lender of last resort. Indeed, FHLB funding initially was cheaper than discount borrowing from the Federal Reserve. For example, prior to August 2007, the rate on overnight advances averaged 69 basis points less than the primary discount rate. Also, many banks believed that both investors and authorities would frown upon borrowing from the Fed. These cost differences changed later in 2007 as the Federal Reserve initiated programs aimed at providing reserves more broadly. Even then, however, the FHLBs were willing to offer longer-term loans, keeping them attractive as a source of funding.
For the more recent period, the new role of the FHLBs is more complex and requires an explanation of two major regulatory shifts: the SEC’s reform of money market mutual funds (MMMFs), and the introduction of new liquidity requirements for banks.
Recall that following Lehman’s failure in in 2008, investors began to flee from MMMFs. To halt the run, the U.S. Treasury guaranteed all $3.8 trillion in outstanding MMMF liabilities. As we discussed previously (see here), MMMFs functioned much like banks engaged in the transformation of liquidity, credit and (to some extent) maturity. Similar to banks that redeem deposits at face value, they promised investors a fixed share value of $1 (a “buck”) on demand. Unlike depositories, however, MMMFs had no capital, no deposit insurance, and—at least officially—no access to the lender of last resort. So, when the Reserve Primary Fund “broke the buck” (by failing to redeem at the $1 par value) in September 2008, MMMF investors panicked.
In 2014, after substantial prodding from the Financial Stability Oversight Council, the SEC finally proposed a MMMF reform that went into effect in mid-October 2016. It requires institutional prime MMMFs to operate like other mutual funds with a floating net asset value (NAV). Importantly, however, it exempts MMMFs that invest primarily in federal government and agency securities, including those of the FHLBs. Unsurprisingly, investors were not thrilled with the idea of having floating NAVs, so they shifted from prime institutional funds to government funds. In 2014, government MMMFs accounted for roughly one-third of the $3 trillion total. By the end of 2018, that fraction rose to three-quarters.
The second change concerns the implementation of the liquidity coverage ratio (LCR) for banks. Announced in 2014, with full effect in 2016, the LCR requires banks to hold reserves and government securities sufficient to cover outflows in a 30-day stress scenario (see our earlier post). The stress presumes certain things about the likelihood of various liabilities running―retail deposits are unlikely to run while short-run interbank loans are very likely to run.
As it turns out, the LCR treats banks’ relationship with the FHLBs very favorably. On the asset side, FHLB bonds count toward the LCR with only a 15% haircut (see here). On the liability side, things are even better. When a bank gets an advance of more than 30 days, it doesn’t count at all (since it is longer than the LCR stress period). When the advance has less than 30 days, the bank must hold government bonds or reserves equal to only 25% of the loan amount. So, in principle, a bank could improve its LCR by borrowing from the FHLBs to purchase FHLB bonds. (To be sure, such a transaction does expand the bank’s balance sheet in a way that could increase its capital requirement.)
How did these two changes alter the intermediation between MMMFs and banks? Prior to the crisis, MMMFs held substantial quantities of bank liabilities: open market paper plus bank deposits accounted for more than 30 percent of MMMF assets. By late 2016, when the SEC rule was in place, that share was less than 10 percent. This decline of about $600 billion in funding is roughly 5 percent of the total liabilities of the U.S. banking system!
Both Gissler and Narajabad and Anadu and Baklanova describe what happened. As a direct result of the regulatory changes, the FHLBs interposed themselves in the intermediation chain between the MMMFs and the banks. In the original chain, funds went directly from prime MMMFs to banks. Now, instead, they go from government MMMFs to the FHLBs to banks. Despite increased complexity and opacity, investors like this because they get an MMMF with a fixed net asset value that is exempt from the SEC regulation. And the banks are happy because they get a liability that is more stable and has favorable treatment in the LCR. They also share in the FHLBs’ profits.
A simple look at the data confirms that this is what is going on. The following chart shows the fraction of all U.S. MMMF assets held in FHLB securities. The line starts at 5.5% in 2011, increasing gradually to 9% by the end of 2015. Starting in early 2016, the fraction rises steeply, reaching 18% several months after the SEC rule went into effect. (As Gissler and Narajabad show in their Figure 4, the fraction of FHLB debt held by MMMFs rose from 25 to 55 percent over this same period.)
Fraction of all U.S. MMMF assets invested in FHLB debt (Percent), 2011-August 2019
Source: OFR Money Market Fund Monitor, and authors’ calculations.
It is more difficult to see what is happening on the banks’ balance sheet. Quarterly disclosure in the call reports provides information on FHLB advances. Anadu and Baklanova go to the trouble of collecting these data for several large banks (see their Figure 7). For both Wells Fargo and JPMorgan Chase, the authors show a rise from very low levels in 2011 to over $70 billion in 2016. A quick check suggests that FHLB borrowing by these banks remains at those relatively high levels.
It is hard to imagine that the designers of these two regulations intended for the result to be a large increase in the demand for FHLB intermediation services. Should we care? Our answer is yes. By interposing a GSE into the intermediation chain, we have taken what was a simple market transaction and introduced an implicit guarantee. Instead of MMMF investors bearing the risks, taxpayers do. Some observers even worry that the FHLBs’ increased maturity transformation and sensitivity to MMMFs could make them vulnerable to rollover risk (see Fischer 2017).
Our bottom line: Leaving aside the benefits from diversification and specialization, relatively simple financial networks that limit the nodes through which funds pass (from provider to user) are likely to be both more transparent and efficient. We suspect that they also are more stable. Complex, opaque systems are more likely to distort incentives and assign risk without proper compensation—especially when a GSE is involved. They also facilitate regulatory arbitrage, thwarting efforts to encourage market discipline and to protect taxpayers.
Acknowledgement: Without implicating him, we thank our friend and Stern colleague, Professor Lawrence J. White, for his very thoughtful suggestions. We also thank Borghan Narajabad for helping us to understand the intricacies of the nexus between MMMFs and the FHLBs.