Author: Zhao Ying , Wall Street Insights

At 10 p.m. Beijing time on Tuesday, the U.S. Senate Banking Committee will hold a hearing on Kevin Warsh's nomination as Chairman of the Federal Reserve. This will be Warsh's first formal presentation of his monetary policy views on Capitol Hill. Notably, Warsh has long been critical of the Fed's massive balance sheet, and this hearing may serve as a crucial platform for expressing his views.

In fact, since the end of 2025, the trajectory of the Federal Reserve's balance sheet has been a core issue of great concern to global financial markets. Against this backdrop, Federal Reserve Governor Stephen Milan, along with three other Fed economists, recently published a working paper, "A User's Guide to Reducing the Federal Reserve's Balance Sheet," and on March 26, 2026, at a keynote speech at the Economic Club of Miami, systematically explained the strategic logic and potential path of the Fed's balance sheet reduction.

The core value of this paper lies in breaking the market's conventional wisdom. In the past, the market generally believed that "the ceiling for the Fed's balance sheet reduction is the depletion of reserves." However, the paper points out that reserve demand itself can be shaped by policy— through a series of adjustments to the regulatory and operational frameworks, the Fed can very well achieve a significant reduction in its balance sheet while maintaining an "ample reserve" framework.

In response, the CITIC Securities research team provided an in-depth analysis. Their assessment is that technical options such as relaxing LCR standards, reforming SRPs, and upgrading Fedwire are feasible to some extent; however, proposals such as reserve tiering, reforming TGAs, and foreign reverse repurchase pools are relatively idealistic. Overall, the balance sheet reduction process is unlikely to change the underlying logic of global central bank gold purchases, and CITIC Securities maintains its prediction that the Federal Reserve will cut interest rates by 25 basis points in the second half of this year.

Why reduce the table: A list of reasons for Milan

In his speech in Miami, Milan laid out the multiple reasons for reducing the Federal Reserve's balance sheet.

First, it reduces market distortions. The Fed's excessively large balance sheet leads to unnecessary intervention in the money market, exacerbating the problem of financial disintermediation. Minimizing the Fed's "footprint" in the market is a fundamental requirement for maintaining the market's price discovery function.

Secondly, controlling financial risk. Large-scale asset holdings mean greater exposure to market value losses, while also increasing the volatility of Treasury profit payments (remittances). In recent years, the Federal Reserve has faced unrealized losses due to its large holdings of long-term securities, a problem that can no longer be ignored.

Third, safeguarding the boundaries between monetary and fiscal policies. The massive balance sheet objectively allows the Federal Reserve to intervene in credit resource allocation, blurring the lines between monetary and fiscal policy. Furthermore, the large-scale interest payments on bank reserves have been viewed by some members of Congress as a form of implicit subsidy to financial institutions.

Fourth, preserving policy ammunition. If another zero-interest-rate floor crisis were to occur, the Fed would need to provide easing space by expanding its balance sheet. Compressing the balance sheet to a reasonable size now is to preserve necessary room for future policy maneuvering.

Milan frankly admitted that the general consensus is that a significant reduction in the balance sheet is "simply impossible." However, his assessment is quite different: "Reducing the balance sheet is a solvable challenge; those who categorically deny it simply lack imagination."

Key diagnosis: The obstacle to quantitative tightening is "demand," not "supply."

To understand this discussion, we must first clarify a logical structure that has been misunderstood for a long time.

The traditional framework holds that the constraint on the Fed's balance sheet reduction stems from "reserve supply reaching the steep end of the demand curve"—once supply tightens to a critical point, overnight interest rates will spiral out of control. Therefore, the Fed can only passively stop balance sheet reduction once reserves fall to a "scarce" state. The "repo market earthquake" in September 2019 was a real-world manifestation of this logic.

The paper's breakthrough lies in shifting the perspective from the "supply side" to the "demand side." The paper argues that reserve requirements are not an exogenous constraint "naturally determined" by payment and settlement activities, but rather artificially inflated by regulatory rules, oversight standards, and the Fed's own operational framework—a phenomenon Milan refers to as "regulatory dominance" of the Fed's balance sheet.

Specifically, the following three mechanisms collectively pushed up the baseline for reserve requirements:

1. Interest rate spreads have turned reserves into "passive income assets." After the Federal Reserve began paying interest on reserves in 2008, reserves transformed from mere settlement necessities into assets that could compete with Treasury bills. Historically, during periods when the interest rate on reserves (IORB) was higher than the yield on 1-month/3-month Treasury bonds, banks were more inclined to accumulate reserves from a risk-return perspective.
2. The overlapping of multiple liquidity regulations creates a "ratchet effect." Rules such as LCR (Liquidity Coverage Ratio), ILST (Internal Liquidity Stress Test), RLEN (Liquidity Resolution Assumption), NSFR (Net Stable Funding Ratio), and SLR (Supplemental Leverage Ratio) are intertwined, creating a "robbing Peter to pay Paul" dilemma—changing one rule immediately leads to another becoming the new binding constraint.
3. The discount window has long been "stigmatized." High discount window rates, historical association with "problem banks," and the risk of information disclosure and regulatory scrutiny regarding its usage history have led banks to prefer accumulating large reserves rather than utilizing policy tools during periods of liquidity stress. This same stigmatization logic has also extended to the Standing Repurchase Agreement (SRP).

This diagnosis implies a fundamental policy path: instead of waiting for reserves to return to a scarce state, the "scarcity-sufficiency" threshold can be lowered to allow the adequate reserve framework to continue functioning properly with a smaller balance sheet.

How much can it be reduced: a quantitative estimate of $1.2 trillion to $2.1 trillion.

The paper uses the Fed's H.4.1 statement data as of March 11, 2026, as a benchmark, at which time the Fed's total assets were approximately $6.646 trillion. The liability structure is broken down as follows: reserves of approximately $3.073 trillion, cash in circulation of $2.390 trillion, Treasury General Account (TGA) of approximately $806 billion, and foreign reverse repurchase pool of approximately $325 billion.

The paper quantitatively estimates 15 policy options across two main directions, but its key feature is that it avoids simple summation. Due to the correlations and substitutability among different policies, the paper employs the Monte Carlo aggregation method within the OMB A-4 framework to derive the following confidence intervals:

Dimension	95% confidence interval	Median
Reserve requirements could decrease	$825 billion - $1.75 trillion	Approximately US$1.287 trillion
Total balance sheet can be reduced	$1.15 trillion - $2.125 trillion	Approximately US$1.637 trillion

In his speech, Milan compared the aforementioned period with historical reference points:

15% of GDP: Balance sheet level after the first round of QE ended in 2009, when the banking system was still functioning normally;
18% of GDP (2012 or 2019 level): Reflects the real liquidity needs of the banking system after the Basel reforms and the Dodd-Frank Act requirements become clearer.

The Fed's current balance sheet represents approximately 21% of GDP. According to the median estimate in the paper, if reforms proceed smoothly, the balance sheet could potentially fall back to levels close to those of 2012 or 2019. As for whether it can return to the pre-crisis level of less than 10% of GDP—Milan clearly stated: "Unrealistic and unnecessary."

How to summarize: A "menu-style" analysis of 15 options

The paper divides the 15 policy tools into two categories, and provides an estimate of the effect range and the prerequisites for implementation for each.

Category 1: Reducing the demand for equalization reserves

(I) Regulatory Reform Level

The LCR (Liquidity Coverage Ratio) reform's core measure is to allow banks to include the funding capacity corresponding to non-HQLA loans pre-collateralized at the discount window in their HQLA calculations, with a cap set. The paper estimates the impact on reserve requirements to be between $50 billion and $450 billion. The paper also cautions that if only the LCR is changed, the NSFR (Non-Standardized Receivables Ratio) may immediately take over as the new binding constraint, requiring comprehensive consideration.
ILST and the Liquidity Relief Assumption (RLEN): If regulators approve the discount window capacity and short-term liquidity sources, ILST reform could lead to a $50 billion to $200 billion decrease in reserve requirements; if the RLEN assumption extends the available time window of the discount window, the estimated range is $0 to $100 billion.

(II) Supervision Scope

If banks hold excessive reserves to cater to inspectors' preferences (i.e., T-bills and reserves are not "equal" in terms of supervisory authority), the estimated magnitude of the adjustment would be between $25 billion and $50 billion. This is a reform that can be achieved without amending regulations, relying solely on a shift in supervisory culture, but its implementation should not be underestimated.

(iii) Reduce the returns on reserve holdings

Allowing the effective federal funds rate (EFFR) to be higher than the international interest rate (IORB) would break the current situation where the EFFR has been lower than the IORB for a long time. The paper cites the Lopez-Salido and Vissing-Jorgensen (2025) framework to estimate that if "EFFR-IORB = +2bp" is used as a reference (close to the pressure level in September 2019), the corresponding decrease in reserve demand would be between $150 billion and $550 billion.

However, this path comes at a significant cost: volatility in overnight and repo rates will increase substantially, and if the market engages in precautionary stockpiling as a result, the decline in demand may be partially offset. This approach requires supporting mechanisms such as SRPs and Temporary Open Market Operations (TOMO).

(iv) Enhance the attractiveness of alternative assets

These measures, including upgrading the Fedwire system, improving liquidity in the Treasury market, and advancing central clearing, aim to make alternative assets such as Treasury bonds more attractive to banks, approaching the level of required reserves. These measures will also help enhance the private sector's ability to absorb securities released during the Fed's balance sheet reduction process.

(v) Destigmatize the Fed's liquidity tools

By addressing concerns about the use of tools such as the discount window, standing repo facility, and intraday overdrafts, the Fed can reduce banks' precautionary reserve requirements. This requires systematic cooperation from the Fed in terms of transparency, pricing mechanisms, and regulatory communication.

Category 2: Directly reduce non-reserve liabilities

(a) TGA Management Recalibration

The paper proposes reducing the Treasury's cash buffer in the Federal Reserve's account from "approximately five days of operating funds" to "approximately two days," with any excess being transferred back to the commercial banking system (similar to historical TT&L arrangements). The estimated reduction in the Fed's balance sheet is between $200 billion and $400 billion. The paper also acknowledges that the net effect is not one-to-one, as the return of deposits to banks will correspondingly increase banks' demand for reserves.

(ii) Reduce the attractiveness of foreign reverse repurchase pools

By lowering interest payments and setting limits on the scale of reverse repurchase agreements, foreign central banks and sovereign wealth funds can be guided to shift funds from the Federal Reserve's reverse repurchase pool to the US Treasury market. The estimated range is $0 to $100 billion, but the effect is relatively limited and depends on the willingness of external institutions to cooperate.

Walsh's Signals: From Technical Papers to Policy Expectations

To understand this paper, one cannot ignore the context of the Fed's personnel appointments. The market widely expects Warsh to succeed him as Fed Chairman. Warsh has long been critical of the Fed's balance sheet expansion policy since QE and has repeatedly expressed his preference for reducing the balance sheet publicly.

This working paper, led by Milan, is seen by outsiders as a forward-looking signal of the Federal Reserve's policy orientation in the "Wash era." The CITIC Securities research team points out that, given Warsh's stance and the potential scope revealed in this paper, there is indeed a possibility that the Federal Reserve in the "Wash era" might gradually explore restarting balance sheet reduction.

However, both the paper and the speech repeatedly emphasized that speed and pace are the most important constraints at the implementation level. Milan clearly pointed out in his speech: "Once the preparations for reform begin, according to the usual pace of the government passing the Administrative Procedure Law (APA), it is likely to take more than a year, or even several years." He cited the SLR (Supplementary Leverage Ratio) reform as a reference—it took nearly six years from temporary relaxation to the formal implementation of regulations.

This means that the Fed will not immediately restart balance sheet reduction in the short term because of the release of this paper. A more likely path is to start researching options that are less controversial and technically feasible, while providing the market with forward-looking guidance on how the new mechanism will work.

CITIC Securities' Interpretation: Which are feasible, and which are more idealistic?

The CITIC Securities research team conducted a systematic evaluation of 15 policy options from a practical feasibility perspective and reached the following core conclusions:

Options that are realistically feasible:

Relaxing LCR standards: This is a technical regulatory reform, with relatively controllable variables, giving the Fed more initiative in the reform;
Reforming the Standing Repurchase Tool (SRP): The destigmatization process is relatively direct and does not involve external legislation;
Upgrading payment systems like Fedwire: This is a long-term improvement at the infrastructure level, with a clear direction.
ILST's oversight approach has been adjusted: some reforms do not require legislative amendments and can be promoted through a shift in oversight culture.

More radical options or those requiring external cooperation:

Tiered interest payments on reserves: This may trigger non-linear reactions in the banking system and is complex to implement.
TGA management reform: involves a coordination mechanism between the Treasury Department and the Federal Reserve, and requires political consensus;
Reducing the foreign reverse repurchase pool: It is highly dependent on the willingness of external institutions to cooperate, and the effectiveness is difficult to guarantee.

Overall, CITIC Securities believes this is "a relatively pragmatic reform menu worth referencing," but the actual implementation will be much slower than the potential upper limit described in the paper. It should be regarded as a directional guide rather than a recent policy commitment.

Market impact: Increased volatility, but the logic for interest rate cuts remains unchanged.

Regarding the impact on the bond market, the Fed's balance sheet reduction essentially means reducing the issuance of base money, which will inevitably increase the amount of US Treasury bonds that the private sector needs to absorb. CITIC Securities believes that this will amplify market volatility and increase tail risks —although some deregulation measures (such as the easing of the SLR) will help expand dealers' absorption capacity.

In terms of pacing, the paper explicitly opposes accelerating balance sheet reduction through direct securities sales. A more feasible approach is to allow maturing securities to roll off the balance sheet naturally, while simultaneously providing dealers and the repurchase market with a higher reserve of absorption capacity. This objectively limits the short-term impact of balance sheet reduction.

CITIC Securities believes that US Treasuries are currently more suitable for trading opportunities, and short-term bonds may be better than long-term bonds.

Regarding the impact on the stock market, quantitative tightening will have a contractionary effect on the real economy through two paths: money supply and portfolio balancing effects. However, this can be offset by lowering the federal funds rate. CITIC Securities believes that if quantitative tightening reforms proceed, the necessity for corresponding adjustments to the interest rate path will increase, but this has limited direct connection to the current pace of monetary policy. US stocks may need to wait for a correction window to find a thicker safety margin.

Regarding the impact on the gold market, the balance sheet reduction reform is unlikely to substantially change the strategic logic behind global central banks' increased gold holdings, which is driven more by geopolitical restructuring and the trend of diversifying dollar reserves. Gold still possesses medium- to long-term investment value.

In his speech, Milan explicitly pointed out that the contractionary effect of balance sheet reduction can be offset by interest rate cuts, and that "balance sheet reduction may lead to a larger reduction in the federal funds rate relative to the baseline scenario." CITIC Securities predicts that the US CPI year-on-year growth rate may fluctuate between 3.0% and 3.5% this year, maintaining its judgment that the Fed will cut interest rates by 25 basis points in the second half of this year. The balance sheet reduction reform and interest rate cut decisions are not directly linked.

Hassett: The Fed needs to cut interest rates further.

Popular Articles