William Bergman, who was an economist at the Chicago Fed for 14 years, has published a bombshell analysis for the Institute for New Economic Thinking, which describes in depth how the central bank has been hiding large losses and what should be shown as negative equity via its own irregular accounting. Bergman’s finding of a loss rate of $100 billion a year is based on the subsidy to banks provided by the payment of interest on reserves, whose costs has risen considerably due to comparatively high interest rates in recent years. We’ll later unpack a second unaccounted-for source of losses, which Bergman mentions but does not attempt to estimate, which is unrecognized losses on securities purchased during QE, whose value has fallen in a higher interest rate environment. As we’ll explain, QE by design was a “buy high, sell low” scheme, so this outcome should not be surprising.
Due to the size of the full article, which includes the 2002 paper, we are embedding only the 2025 portion and the abstract of the 2002 paper. Both are highly readable. We encourage you to read them in full (the entire article is here).
Bergman documents that the central bank has been hiding its negative equity status by engaging in accounting that is out of conformity with any form of reporting to which the Fed is required to adhere. These losses, now in excess $100 billion a year due to the Fed’s early implementation, in 2008, of paying interest on reserves banks keep with the Fed.
Adding to the weight of Bergman’s charges is a 2002 Bergman draft paper that the Fed suppressed1 and is included with the document published by INET. It documents how the Fed was already engaged in an impermissible subsidy, by not recovering its fully loaded costs for Fedwire, the essential interbank payment system, as required by statute. Even though the Fed’s own reports show Fedwire as paying its own way, repeated statement by Fed official contradict those figures. As Bergman pointed out:
If a 1980 law directs the Federal Reserve to charge prices for its payments services so as to achieve full cost recovery in the long run, why did members of the Federal Reserve Board of Governors testify in the late 1990s that guaranteed final settlement through Fedwire comprises a subsidy?
We have explained before the absolutely central role that Fedwire plays in the dollar payments system. When a bank makes a payment to another bank, the receiving bank treats it as money good because the Fed stands behind the transaction. That means banks run large intraday balances with Fedwire and are expected to settle up at the end of day. That entailed risk of a bank that was in a deficit position not being able to make good and imposing a loss on Fedwire.
The Fed in 2008 changed to a system of paying interest on reserves that banks keep with the Fed. That incentivized banks to keep large reserves, theoretically eliminating the payment risk issue (although that was not the reason for implementing this system2). The Fed attempted to justify this new subsidy to banks as a way to manage interest rates, when there was nothing wrong with the established system of having the New York Fed’s money market desk trade so as to keep market rates in line with the central bank’s policy rate. And in fact this new scheme performed poorly in rising interest rate cycle. The repo panic of 2017 was the direct result, as we pointed out repeatedly at the time, of the Fed having gotten out of the habit of managing rates via at times very active intervention by the New York Fed and stumbling about as it tried to use atrophied muscles.3
Bergman explains what resulted:
So, today, we have the Fed incurring massive losses driven by the Fed paying interest to banks for the privilege of reducing the risk they pose to the Fed, instead of charging banks to fully recover the cost of guaranteeing daylight-overdraft funded Fedwire payments. And the Fed is accumulating losses in a dubious asset that helps it keep from reporting a negative capital position on its balance sheet. Yet the Fed’s balance sheet has significant consequences for the federal government’s fiscal condition, and in turn, taxpayers. The Fed is effectively masking its true net position, keeping it from showing a negative number like the “zombie banks” that Edward Kane identified. Ironically, the Fed is able to do so using accounting policies it drafts for itself while asserting the value of central bank “independence.”
The Fed asserts that its losses need not impair its ability to conduct monetary policy. If they are indeed irrelevant, however, why does the Fed choose to implement strange accounting policies that keep it from reporting red numbers in the net position for the Reserve Banks?
he article explains in detail how the Fed does not adhere to any of the arguably applicable accounting standards: FASB, GASB, or FASAB.
To turn to the QE matter, which only adds to this sorry picture: As Bernanke explained at the time, the objective of QE was to lower mortgage interest rates and spreads by buying Treasuries and other high quality securities, as in Federally guaranteed mortgage securities. The central bank never purchased subprime or less than pristine bonds (from a credit standpoint). The purpose was to lower mortgage rates, which are much longer in maturity than the short end, where the Fed normally operates. The goal was to goose the very sick housing market.
Success of this scheme would inherently produce interest rate losses on the instruments the Fed purchased. QE represented an artificially large level of demand destined to lower rates, as in bid up prices. Prices would be lower absent the Fed operation, hence the central bank was, by design, “buying high”.
So when the economy recovered, which a more robust housing market was expected to achieve, interest rates would increase from their abnormally low level during the crisis. That would produce interest rate losses on the securities the Fed bought during QE. The Fed has pretended they don’t exist by treating them as “hold to maturity”. But refusing to mark to market does not mean the losses do not exist.
Even though this bill of particulars would make a great Trump Administration bludgeon to use against the Fed, I doubt they will go there, save at most for criticizing Powell for making this sorry situation worse with continued high rates. First, talking about a >$100 billion annual subsidy to banks should lead to demands to unwind that, or alternatively, recover it via true full pricing for Fed services as mandated by Congress. It seems very unlikely that Team Trump will want to gore the oxen of powerful donors. Second, the issue is complex and Trump and complexity do not co-exist happily.
Below is Bergman’s short recap for INET if you don’t have time to dive into the full article. Hopefully it will whet your appetite to read the underlying piece.
By William (Bill) Bergman, who has four decades of financial market experience in private sector, public sector, and educational roles. He served as an economist and financial markets policy analyst for the Federal Reserve Bank of Chicago from 1990 to 2004. He delivers a daily newsletter on government finance at GovMoneyNews, and he is co-authoring a book (with Larry Feltes) to be titled “Three Keys: Unlocking Prosperity with Ethics, Economics and Excellence.” Originally published at the Institute for New Economic Thinking website
Without transparent accounting practices and proper risk management, the Federal Reserve’s current financial losses—unprecedented in scale—and the questionable accounting practices it uses to downplay their impact threaten public trust, economic stability, and the integrity of fiscal policy.< The Federal Reserve is experiencing something new in its history: sustained and sizable operating losses. These losses—currently running at more than $100 billion a year on an annualized basis—stem largely from the sharp rise in short-term interest rates, which has increased the interest the Fed pays on bank reserves while the income from its long-term securities portfolio remains comparatively low. At the same time, the Fed’s approach to accounting for these losses departs from the norms applied to other public and private institutions. Since 2010, the Fed has recorded them as a “deferred asset”—essentially the expectation of future earnings—rather than reducing its reported capital position. This treatment helps the Fed avoid reporting negative capital, but also makes it harder for the public to see the institution’s true financial condition. My new INET Working Paper connects these current developments to a longer history of accounting choices—particularly those related to the Fed’s Fedwire payment system—that have tended to understate the costs and risks the central bank assumes.
Fedwire and Daylight Overdrafts
Fedwire is the Federal Reserve’s wholesale payment system, moving trillions of dollars daily between banks. One of its defining features is that the Fed guarantees each payment to the receiving bank, even if the sending bank does not have sufficient funds in its account at the moment of transfer. This practice can create “daylight overdrafts”—short-term intraday credit extended by the Fed.
The 1980 Monetary Control Act required the Fed to price its payment services, including Fedwire, to recover all direct and indirect costs over the long run, including an imputed return on capital and the cost of credit provided. Yet for decades, the Fed has excluded important components from its cost-recovery calculations, such as the foregone interest on daylight overdrafts and expenses incurred in monitoring and managing payment system risk.
The result is an apparent contradiction. In testimony to Congress, Fed officials have acknowledged that Fedwire access and daylight overdrafts represent part of a “safety net” subsidy to banks. At the same time, the Fed’s cost-recovery reports assert that Fedwire has fully recovered its costs year after year. Both cannot be true in the same sense.
From Payment Services to Balance Sheet Losses
The link to today’s losses lies in how the Fed manages the risks and costs of its own activities. Since 2008, the Fed has paid interest on reserves, encouraging banks to hold large balances at the central bank. This has helped reduce daylight overdraft exposures—but it also means that the Fed now pays substantial interest to banks, particularly in a higher-rate environment.
With the level of reserves now above $3 trillion, the cost of these interest payments has risen sharply, contributing to today’s operating losses. Those losses are then absorbed into the “deferred asset” account, avoiding an immediate impact on reported capital. This approach resembles the way some banks accounted for securities on a “held-to-maturity” basis—a method that came under scrutiny during the 2023 banking turmoil.
Why It Matters
Central banks can, in principle, operate with negative capital without immediate operational disruption. But the way losses are measured and reported still matters for transparency, fiscal policy, and public trust. The Fed’s accounting choices for both its payment system and its broader balance sheet have tended to minimize reported costs and risks. Over time, this can obscure the real economic implications for the U.S. Treasury and, ultimately, taxpayers.
The experience with Fedwire offers a reminder that the details of cost measurement, risk assessment, and accounting policy are not mere technicalities. They shape perceptions of central bank performance, influence incentives in the financial system, and determine how the burdens and benefits of central bank activities are distributed.
If we want to understand the Fed’s current losses—and their implications—it helps to look at how the institution has long handled the costs of the services it provides. Fedwire’s history shows that the question is not just how much the Fed earns or loses in a given year, but how it defines and measures those outcomes in the first place.
____
1 Bergman had been invited to present it in Washington. He was told by a senior Chicago Fed officer, “You can drop this and move on, or do this, and really move on.”
2 From Bergman’s article:
A comprehensive piece of legislation, the Monetary Control Act was preceded by a lengthy deliberative process. During the 1970s, high and rising inflation and interest rates were coupled with a much-discussed Federal Reserve ‘membership problem,’ a problem identified as a source of difficulty in conducting monetary policy. Federal Reserve member banks were increasingly opting out of membership as the opportunity cost of maintaining non-interest bearing reserves rose along with market interest rates, even though payments services were offered to member banks for free.
In April 1978, G. William Miller, the Chairman of the Federal Reserve Board of Governors, testified to Congress that the Federal Reserve was considering the possibility of paying interest on reserves held at Federal Reserve banks as a means of addressing the membership problem. The legal foundation for an independent decision on this action was questioned by Congressional leadership. Comprehensive legislation was developed to address the membership issue by requiring universal yet simpler and less onerous reserve requirements, together with a direction that the Federal Reserve begin offering its payments services to all depository institutions on a priced basis. The Federal Reserve previously provided these services exclusively to member banks, and for free.
3 It did not help that the two most senior traders on the money market desk had resigned a few months before the repo panic. One has to wonder why.
00 -Bergman-Fed 2025 section-compressed
33 comments
Comments are closed.
Holy samolians! (To use American slang.)
Thanks for this post.
adding: why should I ever trust a CBDC?
Is the Fed really hiding anything? Or is this sensationalism? The Deferred Asset item is reported on the Balance Sheet, and it means the Fed will not be sending any future profits to the Treasury until the deferred asset balances are all cleared up. And they will clear up when the securities it is meant to hold to maturity do reach maturity. “Bank capital” doesn’t mean anything for the Fed’s ability to operate, so it is always going to operate on special accounting no matter what.
Having said that, it is always a good thing to re-examine whether paying banks interest on reserves is still the sensible thing to do.
Yes, that was my reaction too…..this isn’t a secret. It’s a feature, not a bug—-and buried in the news cycle because peeps are innumerate, accounting jargon makes their eyes glaze over, and screeching about culture war issues is easier.
You are completely skipping over the issue of the >$100 billion a year losses at current run rates due to the unwarranted interest on reserves subsidy. That will not be recovered. That is completely separate from the QE losses, which is what you are focusing on to the exclusion of the matter to which Bergman devoted the overwhelming amount of analysis and support.
In other words, this is a straw man.
Second the “deferred asset” treatment is not permissible. so the Fed is indeed hiding its negative equity
Third. central banks with negative equity are not a nothingburger. None other than the BIS has warned that central bank negative can, and even has, undermined trust in money: https://www.bis.org/publ/bppdf/bispap146.pdf
The IMF made similar warnings earlier, that negative equity can compromise central bank independence:
https://www.imf.org/external/pubs/ft/wp/wp9783.pdf
Recapitalize = getting more equity from taxpayers.
Cyprus a few years ago got to experience a bail-in. Wonder what is the over/under on that happening in the US?
A current example of a central bank with negative equity is the Argentinian one.
Not skipping over anything at all. The losses are there and they are not being hidden by the “special accounting.” The special accounting simply tells us that the Fed is not like a commercial establishment, where insolvency is death. As the IMF passage you quoted says “Central banks may operate perfectly well without capital as conventionally defined.” This is basically acknowledging that central bank balance sheets have “special accounting”.
The true issue, again as your IMF quote suggests, is whether Congress or Trump will make hay about these losses and thereby place restrictions on Fed operations. The issue is not whether the losses themselves are sustainable or not (they are).
Now it is definitely worth debating whether the Fed should be paying interest on reserves. It is just that this debate should be had irrespective of whether the Fed is making losses or not.
Totally agree.
It’s sensationalism that serves to undermine the most important democratic/public-connected banking institution in the financial system–which is not to say that it is especially democratic or public, just that it is much more so than any other banking institution. Any of us who have followed the Fed have known for some time about the losses, and they exist because the Fed is a tool of policy, not a for-profit bank.
But 100%, bring back reserve requirements and stop paying interest on said required reserves.
No, you really have this wrong. Read my later comments.
1. The losses mean no more Fed remittances to Treasury, which historically reduced Federal fundraising needs (remember, politically we do not operate according to MMT principles). So the public is already losing in a tangible manner. Even the CBO acknowledges that.
2. Second, if a central bank gets too far in negative equity terrain, its currency starts to be questioned. Economists have warned that recapitalization (as in some sort of levy on the public) may be necessary. This is not theoretical:
https://www.socialeurope.eu/central-bank-independence-the-beginning-of-the-end
The subhead to a 2025 article in Central Banking (sadly paywalled) similarly says that 10 banks were recapitalized in the last year: https://www.centralbanking.com/benchmarking/governance/7972618/over-40-of-central-banks-lack-formal-recapitalisation-agreement
I am trying to make sure that I understand this correctly — especially the parenthetical bit. Is this to say that the US Government (and therefore the Fed) “operates” under MMT principles, but does not admit so publicly? Is that what “politically we do not operate according to MMT principles” implies or means?
That’s my instinctive interpretation, but maybe I am drawing inferences not intended.
Correct me if i’m wrong.
Technically, recapitalizing a central bank should not be a particular problem. For example, the Treasury can issue special non marketable zero coupon perpetual bonds and swap them for shares at the CB refilling the capital buffer, the accounting should square off. This is a solution for the negative capital “bad optics”.
This would make the problem more visible, which is the last thing the Fed wants. It would feed Trump demands for big changes at the Fed.
First, there is no numerical evidence/analysis that the losses so far and for the next couple years create risk of needing recapitalization for the Fed. The Fed is not the Swedish CB.
Second, I think we are tying up the IOR issue with the losses issue. Both myself and prior commenter agree the IOR issue should change. I analytically understand that the IOR policy is causing losses that reduce remittances. But note, the IOR policy reduces remittances regardless of whether there are accounting profits or losses. On the other hand, losses are a feature of monetary policy and can exist during periods of high short rates regardless of IOR policy. In fact, losses may be the necessary pain of ending QE, which I evaluate negatively. Further, I have a keen awareness that some reactionary Republicans have already and will continue claiming the Fed is a failed institution because of the losses, and their remedies will be a combination of further privatization and political control by the present administration.
Hence, I have to ask whether you have separated IOR and accounting losses as conceptual issues, putting to the side the obvious analytical reality that the current IOR policy is making the accounting losses larger than they need to be.
Is this serious? This is a combination of straw manning and diversion.
First, whether the Fed will be recapitalized will be a political decision in the US. And it will be a function of the level of losses, which would in turn be based on the Fed’s interest rate actions, which in turn will be based significantly on the inflation, which in turn will largely be based on Trump’s tariffs, when Trump keeps changing his plans on the fly. So your demand for a numerical analysis is in bad faith.
Second, as the papers I cited pointed out, recpaitalization is not the only possible downside a central bank with significant negative equity. The other is loss of independence. You ignore that risk.
Third, the losses ARE due to IOR in combination with short rates that are not even abnormally high by historical standards. How can you carry on otherwise? People don’t get upset about unwarranted bank subsidies, which are rife. How much ire is there about underpriced FDIC insurance? About the unwarranted bailout to SVB? Voters might wake up a bit when they realize every dollar in IOR is regarded by Congress as less money to fund the government, and the current losses and negative equity put the situation in spotlights.
This whole angle that you allegedly agree that the Fed shouldn’t be making these payments yet you believe they should be allowed to do it for free on their balance sheet is just bizarre.
Even if we are just looking at remittances, your statement is misleading because the lack of Fed remittance has budgetary impact. Even though it is not operationally necessary (the Treasury could just “net spend”), the fact that the Fed is generating losses rather than income means that the US must either tax or borrow more.
See this CBO discussion, for instance;
https://www.cbo.gov/system/files/2023-02/58913_Fed_Res_Remittance.pdf
And the results have been worse than the CBO anticipated. From a St. Louis Fed paper:
For most of the past decade, the Fed remitted between $5 billion and $10 billion per month to the Treasury. In fact, between 2011 and 2021, the Fed’s remittances totaled over $920 billion. However, beginning in September 2022, remittances due became negative. Since then, the Fed has experienced a shortfall in earnings that is generally between about $5 billion and $11 billion per month.
https://www.stlouisfed.org/on-the-economy/2023/nov/fed-remittances-treasury-explaining-deferred-asset
A more recent St. Louis Fed table suggests the shortfall is now $20 billion a month.
https://fred.stlouisfed.org/series/RESPPLLOPNWW
Sure, that the Fed no longer remits profits to the Treasury has an impact on the Federal government’s budget. But the Federal government should not be relying on central bank profits in the first place. Fed policy should be concerned with financial stability, employment and inflation. Profits should not even be a consideration!
Yves, that you for this. Fascinating.
One question, there is a rather intriguing footnote [1] but I can’t find the antecedent reference in the text–please excuse if I am mistaken. It also seems slightly mangled:
“after I was” => afterwards?
Will clean up. The “after I was” is indeed a mangle and extra.
The text reference was also goofed up and is now: “Adding to the weight of Bergman’s charges is a 2002 Bergman draft paper that the Fed suppressed1“
Pheww!! What to do? Is gold really the only solution? Can I expect the Treasury to renege on my government bonds? Seems the dollar as a reserve currency is doomed. Who the hell is going to trust the greenback? Owning products that help you survive alive seem to have the greatest value now!
I de-myrrh.
I am frankincensed by both of your comments.
Oh my. The Three Weisenheimers from the West!
The recent rumours concerning Trump issuing gold backed Treasury notes on July 4, 2026 in “honour” of America’s 250th birthday looks to be par for this course. This course being, naturally, a Par Three Wise Guys one.
The Fed seems to have messed up in not collecting the “vig” on the latest economic “loans.”
They just started tariffing some Swiss gold bars of smaller dimensions 39%, this is going to upset the apple cart. Suspicion of gold revaluation incoming to take the pressure off bond issuance. Are those bars really there in Fort Knox or is the revaluation of fresh air coming.
As fewer and fewer world investors in different nations are investing in the US $$, will we expect that The Fed has and WILL refuse to show those losses?
The issue for me is that the term , “deferred asset,” describes at worst a temporary loss followed by an eventual recovery that turns it into an asset. Hence it is “deferred” and an asset.
But payment of dividends to Federal Reserve Banks and the payment of interest on reserves to banks are one-off irreversible payments. They are permanent losses. They do not contain some hidden “asset.” There is also nothing temporary or two -sided about them. There is no “deferred asset” on the other side. It is simply a loss.
So the use of this “deferred asset” accounting device is misleading and tawdry not only because it is unconventional, but also because it affirmatively misrepresents the underlying transactions and economic reality.
It’s shocking that the keeper of the world’s reserve currency engages in accounting practices that would be considered fraudulent if used by any other company or institution. It’s not normal at all.
It is not irregular to use odd deferred accounts in special regulatory contexts. Not everything is GAAP. Take it from someone who has worked a lot with utility regulatory accounting, tilting at wind turbines and all that.
Unless Cervates’ job was to engage in accounting fraud for said utilities then I doubt he was doing stuff like this. A deferred asset as I understand it in a utilities context would be connected to some specific future income flow. Whereas the Fed is just taking a loss and calling it a nothing because it helps to enrich the owners of member banks at everyone else’s expense. The idea that basically just switching around positive and negative signs without any identifiable basis for doing so is standard practice in eg the utility industry… Im not sure whether to be amused or horrified by this assertion.
Enron used something called a Special Purpose Vehicle (SPV) to hide its losses. Its former CEO Ken Lay was convicted on six counts of fraud and conspiracy and four counts of bank fraud.
https://www.investopedia.com/updates/enron-scandal-summary/
Wolf Richter provides some context on this:
https://wolfstreet.com/2025/03/21/feds-operating-losses-declined-to-78-billion-in-2024-unrealized-losses-rose-to-1-06-trillion/
Accounting appropriateness aside, no surprise to anyone as stated earlier in the comments.
If the losses are no surprise then why use a misleading accounting device to hide their effect on the balance sheet?
You’ll notice that most of Wolf Richter’s analysis is drawn from footnotes to the financial statements, not the financial statements themselves.
Good point. Is this a stealth way for the Fed to recapitalize the big banks holding big reserves at the Fed? If so, why do the big banks need recapitalization? Another bailout by stealth?