Notes from the Vault

Larry D. Wall
June 2016

The Federal Reserve's discount window offers collateralized loans to banks to assist the Fed in executing monetary policy and maintaining financial stability. However, for the discount window to achieve these goals, banks must be willing to borrow from it, which is not always the case because of a problem known as stigma. Stigma arises because banks are concerned that if their use of the discount window is detected, other market participants will view the usage as a sign of financial weakness.

Stigma was a particularly large problem at the start of the recent financial crisis in the fall of 2007 when some short-term bank borrowing rates soared. The Fed's initial response was to lower the rate it charged to encourage banks facing temporary liquidity problems to borrow at the discount window. However, banks' use of the discount window remained subdued and they continued to show a preference for buying funds in the market at elevated rates. In an attempt to mitigate stigma and reduce short-term market rates, the Fed created the Term Auction Facility (TAF) in December 2007. The TAF was far more successful than the discount window at providing loans during the 2007–09 crisis as shown in the chart.1


As the crisis abated, banks' urgent demand for liquidity decreased. The Federal Reserve judged the TAF was no longer needed and held the last TAF auction in March 2010. However, given the TAF's success during the crisis, the facility could be reopened in the future. Should we expect the TAF would work as well in a future crisis? Could a TAF-like facility possibly prove to be a superior substitute to the discount window for monetary control purposes? This post explains why discount window borrowing may result in stigma, discusses some theories on why the TAF overcame that stigma, and considers the implications for the future of TAF-like facilities.

Discount window and stigma
The discount window was created as a part of the Federal Reserve's original mandate to create an "elastic currency" that could be expanded during financial panics. Banks facing temporary liquidity problems could borrow from the Fed, provided they had good collateral. As the Fed's mission increasingly focused on monetary policy, the discount window's intended role expanded to helping with interest rate control by setting a ceiling on how much banks needed to pay to buy additional reserves.

The ability of the discount window to provide funds to banks with liquidity problems is also the source of its biggest weakness, stigma. When uninsured depositors become concerned about the financial conditions of their banks, they have an incentive to use any available information to identify which banks are weak and in danger of failing.2 One of the reasons why a bank may have a liquidity problem is that some other uninsured depositors have adverse information about the bank and are withdrawing their funds. Thus, uninsured depositors in a bank will be looking carefully for signs a silent run has begun on their bank and may perceive a bank's use of the discount window is a sign of a silent run. In theory, banks' usage of the discount window is confidential information, but in practice other market participants seem to be able to identify which banks are coming to the Fed.3 Rather than risk sending a signal of weakness, a bank may prefer to borrow in private markets even if the rate paid on such borrowings exceeds the rate charged by the Fed—this is what I mean by stigma.

Several studies provide evidence of stigma prior to the crisis. St. Louis Fed economist David Wheelock (1990) found evidence banks became less willing to borrow from the discount window in the Great Depression. New York Fed economist Stavros Peristiani (1998) found a similar reluctance to borrow during the 1980s when banks were failing at the fastest rate since the Great Depression. Bank for International Settlement economist Craig Furfine (2001) found the reluctance to borrow from the Federal Reserve applied even during a period when the need for funds was clearly benign and not a sign of weakness—concern about potential disruptions due to incompletely modified computer programs around Y2K.4

Part of the reason why banks proved reluctant to borrow from the discount window in earlier periods was the way the Fed was managing the program, according to Federal Reserve Board economists Brian Madigan and William Nelson, writing in 2002. The discount window offered loans through its adjustment credit program at rates below the federal funds rate, the rate at which banks could borrow from one another in the market. To address concerns that banks might rely on the cheaper discount window in preference to borrowing in the federal funds market, the Fed took administrative measures to discourage borrowing, such as threatening to deny the bank's future requests for adjustment credit. Thus, banks found it difficult to distinguish what the Fed viewed as appropriate usage of adjustment credit versus excessive reliance. In the face of this regulatory uncertainty, they largely avoided using the discount window.

In January 2003, the Federal Reserve sought to fix the problem and provide some clarity as to when discount window usage was appropriate. As a part of its discount window operations, the Federal Reserve created a new primary credit facility that would charge a rate higher than the federal funds rate. The premium rate on primary credit eliminated the incentive of banks to have excessive reliance on the discount window, which in turn allowed the Federal Reserve to stop using administrative measures to discourage bank borrowing. The hope was banks would rely on the interbank market when market rates were less than the primary credit rate, but they would use the primary credit facility whenever the federal funds rate exceeded the cost of primary credit.

Primary credit, however, did not work as intended and the Federal Reserve began to consider other alternatives in the fall of 2007. The TAF was similar to the discount window because its eligibility criteria was the same as that for primary credit and it took the same sort of collateral. In other respects, however, the TAF was less favorable to borrowers: (1) A qualified bank could apply for primary credit any business day and have its loan settled by the end of the day. In contrast, banks could bid for TAF funds only at an auction (initially held biweekly) and the winners faced a three-day lag between when they submitted a TAF bid and when they would receive the funds. (2) A bank could apply for any amount of funds at the discount window subject only to it having adequate collateral, whereas the amount available to any one bank through the TAF was limited. (3) The maturity of a TAF loan was fixed, while primary credit loan was available with a range of maturities up to the maximum available from the TAF.

Even though the TAF offered less favorable non-price terms, it was more successful in distributing funds, as shown in the chart. In order to better evaluate the role of stigma, a team of researchers including Federal Reserve Bank of New York economist Olivier Armantier (2015) compared the prices banks paid for the TAF relative to the primary credit rate. The researchers found banks often paid more to obtain funds through the TAF than they would have paid had they borrowed from the primary credit facility. This finding suggests that something about the structure of the TAF must have allowed it to overcome the stigma arising from the use of the discount window.

Why did the TAF succeed when the discount window failed?
The New York Fed's Olivier Armantier, Sandra Krieger, and James McAndrews provide some of the original reasons why the Federal Reserve thought the TAF might overcome stigma:

… competitive and well-functioning auctions for term credit could circumvent the stigma associated with the discount window. Auctions require banks to bid simultaneously; the interest rate at which the funds are allocated is determined by the demand for the funds. An auction format would enable banks to approach the Federal Reserve collectively rather than individually and obtain funds at a rate set by auction rather than at a premium set by the Fed. Thus, institutions might attach less of a stigma to auctions than to traditional discount window borrowing.

More directly, the fact that multiple banks obtained funds at the same time using TAF may have reduced stigma by making it more difficult for depositors and markets to identify which banks were accessing Fed liquidity.

Celine Gauthier from the Université du Québec, Alfred Lehar from the University of Calgary, Hector Perez-Saiz from the Bank of Canada, and Moez Souissi from the International Monetary Fund provide another hypothesis. In their model, the TAF was a success precisely because it was less flexible and more expensive. The low frequency of TAF auctions and the lag between bidding and settlement made the TAF an ineffective choice for banks currently being run by uninsured depositors. The relatively higher cost to borrow funds for a fixed term meant that the TAF was an expensive way for a weak bank to stockpile funds in anticipation of a future run. As a result, weaker banks that might be run had a strong incentive to wait until they were actually being run and then borrow from primary credit. Of course, the TAF was also higher cost than primary credit for the stronger banks that were not being run. However, this cost was less than the cost of the stigma in financial markets associated with borrowing from primary credit. Thus, the stronger banks would use the TAF to the maximum extent possible and use primary credit only when necessary.

Consistent with their model, Gauthier, et al. found that banks that relied primarily on the discount window had less capital, more bad assets, and a higher predicted rate of failure. They also found that banks that relied primarily on the discount window had a higher future cost of obtaining market funding than banks that relied on the TAF. All of this evidence supports the hypothesis that the banks participating in the TAF were of higher quality, or had less risk.

Is the TAF the cure for stigma?
The Federal Reserve has long used monetary policy to control overnight interest rates paid by banks as a tool for influencing the level and shape of the yield curve. Looking to the future, the Fed's ability to pay interest on reserves gives it a tool for establishing the minimum rate at which a bank would lend to another bank. Ideally, this would be complemented by an effective tool for controlling the maximum rate paid by banks, thereby establishing a corridor within which overnight rates could fluctuate. The primary credit program would furnish this upper bound provided that banks would be willing to borrow from the discount window whenever the overnight rate being demanded in the market exceeds the rate on primary credit. However, so long as stigma is a problem, banks will typically decide to pay higher rates in the market than borrow from the Fed. Thus, an important issue for monetary policy is whether a TAF-like facility might solve the stigma problem and provide an alternative way to establish the rate corridor.

The original TAF was not designed to set an upper bound on overnight rates. If the Fed wanted to establish a new TAF that would be effective for monetary control, several changes would have to be made in the new facility. First, the new TAF would need to provide funds at a fixed rate if it were to set an effective upper bound on short-term rates. Additionally, the new TAF would have to provide for full allotment, that is, every bank would have to get the full amount of its request—otherwise, it would need to buy funds at the elevated market rate. Finally, if the new TAF were going to set a limit on overnight rates, it would need to be available every day and would have to settle the same day as the auction. In other words, the new TAF would have to operate just like primary credit if it is to set an effective ceiling on short-term rates. But if a new TAF were set up with the same terms as primary credit, it would seem to be just as vulnerable to stigma as primary credit.

If a TAF-like facility is not the monetary policy answer for setting a ceiling on overnight rates, our experience during the last crisis suggests it can overcome the stigma problem in an emergency. But will the TAF be equally effective next time? One change that may lessen the effectiveness of a TAF-like facility is the Dodd-Frank Act requirement that the Federal Reserve disclose the name of borrowers with a two-year lag. The lag should be sufficient to make sure a borrowing bank is not at greater immediate risk of stigma in financial markets. However, a different form of stigma may arise to the extent that the borrowing bank is perceived by the public as having been "subsidized" by the Federal Reserve.

A second potential difference is that the opportunity cost of obtaining precautionary TAF balances by a risky bank may have decreased. One such source of possibly reduced costs is the payment of interest on reserves lowers the cost of precautionary borrowing by weak banks. A bank that borrowed at the primary credit and held the funds as excess reserves in anticipation of a run earned no interest on those reserves prior to October 2008.5 Thus, the cost of holding precautionary balances as reserves was equal to the primary credit rate. However, the Fed now pays interest on reserves, so the cost of borrowing from the Fed and holding the funds as excess reserves is the difference between the primary credit rate and interest rate paid on reserves. A possible second benefit of borrowing from the discount window is it helps meet the liquidity coverage ratio (LCR) by converting assets that are not easily liquidated into reserves that help in meeting this new regulatory requirement. As a result, bad banks may be more willing to obtain precautionary balances through the TAF, which would increase the risk that stigma would apply to good banks that also used the TAF.

The TAF was an important innovation by the Federal Reserve during the financial crisis that helped it overcome the stigma associated with the discount window. However, the TAF does not appear to be a potential cure-all for stigma that could easily be modified to provide the Federal Reserve with a tool for better control over short-term rates. Absent stigma, the best structure the Federal Reserve could have for monetary control has long existed in the form of the discount window. Indeed, a better question is whether various regulatory changes will make a TAF-like facility less effective in allowing good banks to separate themselves from bad banks, resulting in the TAF becoming more vulnerable to stigma.


Armantier, Olivier, Eric Ghysels, Asani Sarkar, and Jeffrey Shrader (2015). "Discount Window Stigma during the 2007–2008 Financial Crisis." Journal of Financial Economics 118, no. 2: 317–335.

Courtois, Renee, and Huberto M. Ennis (2010). "Is There Stigma Associated with Discount Window Borrowing?" Richmond Fed Economic Brief (May).

Furfine, Craig (2001). "The Reluctance to Borrow from the Fed." Economics Letters 72, no. 2: 209–213.

Peristiani, Stavros (1998). "The Growing Reluctance to Borrow at the Discount Window: An Empirical Investigation." Review of Economics and Statistics 80, no. 4: 611–620.

Wheelock, David C. (1990). "Member Bank Borrowing and the Fed's Contractionary Monetary Policy during the Great Depression." Journal of Money, Credit and Banking 22, no. 4: 409–426.

Larry D. Wall is the executive director of the Center for Financial Innovation and Stability at the Atlanta Fed. The author thanks Paula Tkac for helpful comments. The views expressed here are the author's and not necessarily those of the Federal Reserve Bank of Atlanta or the Federal Reserve System. If you wish to comment on this post, please email


1 The discount window usage shown in the chart is for primary credit, the facility available to generally sound banks. The other discount window facilities are secondary credit (for banks that did not qualify for primary credit) and seasonal credit (available to small- and medium-sized banks that experience significant seasonal variation in their loans and deposits).

2 In order to simplify the discussion, this post uses the term "uninsured deposits" to signify all uninsured liabilities issued by a bank.

3 The Federal Reserve's policy prior to the Dodd-Frank Act was that it would not disclose the names of banks that borrowed from the discount window, but it would publish a weekly total of borrowing by Reserve District. Nevertheless, considerable anecdotal evidence suggests banks believed that other banks would be able to identify which banks borrowed from the discount window. The Richmond Fed's Renee Courtoisis and Huberto M. Ennis suggest that given knowledge of borrowing in a district, "it would not be hard to infer" which bank is doing the borrowing based on the close relationships banks establish with one another in the interbank market.

4 Many of the programs of banks and other firms were originally programmed to handle only the last three digits of the year, assuming the first significant digit would be 1 (as in 1999). This was done to reduce the cost associated with expensive computer storage. Despite a massive effort to fix this date problem, there was considerable concern computer programs in a wide variety of industries would malfunction on January 1, 2000, and banks would need extra liquidity.

5 The borrower could invest its funds overnight and earn some interest in the market. However, this created two risks for the borrower. First, the borrower would want to hold the funds until late in the business day in case a run started late in the day. Second, even though the investment was only overnight there was a chance the bank would not be repaid on time.