ETR-2026-07
Accelerating the Dystopia
Why Artificial Intelligence Cannot Save a System We Refuse to Fix
Money That Few People Can See Moving
Monetary policy is where technical precision most easily becomes a wall between the public and the thing being described.
When ordinary people say “the government printed money,” economists can correctly explain that contemporary operations involve reserve balances, asset purchases, interest on reserves, Treasury issuance, central-bank liabilities, bank balance sheets, and mechanisms more complicated than a printing press.
Fine. Now answer the question the person actually cares about:
What changed in the financial system, who received the new claims, what risks moved, what asset prices responded, and who became richer or poorer as a result?
The Treasury buyback argument is a good example.
In August 2026, Treasury announced that it would at least double the maximum size of long-end liquidity-support buybacks from $2 billion to $4 billion per operation for the remainder of the refunding quarter. [S128]S128 · Treasury Announces Increased Sizes of Nominal Long-End Liquidity Support Buybacks Beginning September 9U.S. Treasury · 2026official policy announcement · current primary source Its August 5 refunding statement also said Treasury expected to maintain nominal coupon and floating-rate note auction sizes for at least the next several quarters, while meeting financing needs through the normal mix that includes weekly bills and cash-management bills. Treasury projected up to $38 billion of off-the-run liquidity-support buybacks and up to $25 billion of cash-management buybacks in the one-month to two-year bucket for the quarter. [S129]S129 · Quarterly Refunding Statement, August 5 2026U.S. Treasury · 2026official policy statement · current primary source
The announced program is not pandemic-scale QE. Treasury bills are not Federal Reserve reserve balances. Treasury and the Fed are different institutions with different authorities.
Then comes the part that gets lost when the vocabulary becomes the argument: the duration arithmetic does not care which department’s letterhead appears on the press release.
If Treasury removes a long-duration government bond from private portfolios and finances the public sector at a shorter maturity, the private sector holds less long-duration sovereign risk than it otherwise would.
If the Federal Reserve buys a long-duration Treasury and creates short-duration reserve liabilities, the private sector also holds less long-duration sovereign risk than it otherwise would.
The plumbing differs. The duration channel overlaps.
This is not a fringe interpretation. A 2026 Kansas City Fed paper identifies shocks to the level and maturity structure of Treasury debt and finds that debt-management choices affect yields, risk premia, investment, and production. The authors explicitly conclude that Treasury debt-management policy can reinforce or offset Federal Reserve asset-purchase programs. [S130]S130 · Treasury Supply Shocks: Propagation Through Debt Expansion and Maturity AdjustmentHuixin Bi; Maxime Phillot; Sarah Zubairy / Federal Reserve Bank of Kansas City · 2026Fed research working paper · mechanism evidence
Greenwood, Hanson, Rudolph, and Summers made the mirror-image argument years earlier: as the Federal Reserve removed duration through QE, Treasury maturity extension put duration back into private hands, functioning as a partial reverse-QE force. [S131]S131 · Government Debt Management at the Zero Lower BoundRobin Greenwood; Samuel Hanson; Joshua Rudolph; Lawrence Summers · 2014Treasury-hosted academic paper · mechanism precedent
The Federal Reserve’s own explanations of QE emphasize the portfolio-balance mechanism. Asset purchases remove duration risk and compress term premia; investors rebalance into substitutes; financing conditions change. [S132][S133]S132 · The Mechanics of LSAP TransmissionFederal Reserve Bank of New York · 2010central-bank research/speech · QE mechanismS133 · The Federal Reserve's Balance Sheet and Treasury MarketsMichael S. Barr / Federal Reserve Board · 2026official central-bank speech · QE mechanism/current framing Christopher Waller has described the Fed’s QE balance sheet as a maturity transformation in which short reserve liabilities finance longer-duration assets. [S134]S134 · The Case for Removing Mortgage-Backed Securities from the Federal Reserve's Balance SheetChristopher J. Waller / Federal Reserve Board · 2025official central-bank speech · maturity-transform mechanism
My objection is not to technical language. It is to using technical language as a semantic stop sign.
If two operations differ in institutional plumbing but both remove substantial duration from private hands, then a serious analysis has to discuss the shared channel rather than declaring the comparison false because “Treasury cannot print reserves.”
Now follow the channel into society.
Lower term premia can reduce longer-term borrowing costs. They can support mortgage markets, corporate credit, and investment. They can improve employment. They can also increase the present value of assets by lowering discount rates.
Who owns the assets?
The Federal Reserve’s distributional accounts tell us ownership is highly concentrated. [S118]S118 · Distributional Financial AccountsFederal Reserve Board · 2026official statistics · wealth-distribution baseline
That does not mean QE is simply a gift to billionaires. A New York Fed analysis finds unconventional monetary policy supported activity and could reduce inequality within much of the distribution through labor-market effects while widening the income gap between the top tenth and the rest through profits and equity values. [S135]S135 · Unconventional Monetary Policies and InequalityFederal Reserve Bank of New York · 2024central-bank staff report · distributional mechanism
That is precisely why the distributional analysis matters.
A policy can save jobs and inflate assets.
A homeowner can benefit through refinancing and home appreciation.
A renter can benefit through employment while falling farther behind the down payment required to enter the housing market.
A pension fund can benefit from rising asset prices while future expected returns fall.
A leveraged firm can refinance cheaply while a saver receives less interest.
There is no single scalar called “who won QE.”
There are balance sheets.
Ordinary citizens do not carry a consolidated public-sector balance sheet around in their heads. They experience the rent, the mortgage rate, the grocery bill, the retirement account, the job market, and the taxes.
That gap is a democratic-legibility problem. Treasury issuance, reserves, bank balance sheets, money-market funds, repo, collateral demand, term premia, inflation expectations, and global dollar funding are genuinely difficult subjects. That does not make the public consequences less real. It makes the people who translate those consequences unusually powerful.
We have built a financial vocabulary in which a person can be materially affected by a mechanism they cannot realistically audit, then be told the mechanism is being misunderstood because they used the wrong noun for it. I have very little patience for that maneuver.
The fiscal baseline can be stated more directly. CBO projects federal debt held by the public rising from 101% of GDP in 2026 to 120% in 2036. It projects net interest outlays increasing from about $1.0 trillion and 3.3% of GDP to about $2.1 trillion and 4.6% of GDP. These are conditional baseline projections, not outcomes already observed. [S136]S136 · The Budget and Economic Outlook: 2026 to 2036Congressional Budget Office · 2026official fiscal projection · current fiscal baseline
Interest competes with everything else.
The United States does not need to be presently unable to pay its interest for the constraint to matter. Interest increasingly competes with everything else.
Every dollar devoted to debt service is a dollar that has to be raised, borrowed, or not spent somewhere else. The political choices remain choices, but the feasible set narrows as the interest burden grows.
This is where my criticism of the larger architecture becomes normative.
Credit systems depend on refinancing, asset values, income, and confidence. Interventions during a crisis can support employment and credit availability while also changing the value of assets and the distribution of risk. Whether a particular intervention reinforces concentration depends on those channels and on the counterfactual without intervention. Neither an official purpose nor an increase in asset prices settles the whole distributional question. [S135]S135 · Unconventional Monetary Policies and InequalityFederal Reserve Bank of New York · 2024central-bank staff report · distributional mechanism
The resulting distribution reflects institutions as well as market prices.
The dollar adds another layer. IMF COFER shows the dollar’s reserve share rose to 57.13% in Q1 2026 from 56.42% in the previous quarter and describes the share as broadly stable. That does not support an imminent reserve-currency collapse narrative. [S137]S137 · COFER Data Brief, First Quarter 2026International Monetary Fund · 2026official reserve statistics · current reserve-currency baseline
Long-run diversification, geopolitical payment systems, sanctions risk, reserve composition, and the fiscal advantages of deep dollar markets remain consequential. The domestic distributional questions do not depend on predicting a collapse.
We have a financial system whose interventions can alter trillions of dollars of asset value, a wealth distribution in which access to those assets is extremely unequal, and a fiscal path on which debt service increasingly constrains public choice.
Now imagine AI systems managing more of the forecasting, trading, issuance strategy, risk modeling, tax optimization, policy simulation, lobbying, and public explanation around those institutions.
If the objective remains “keep the existing machine stable and the numbers going up,” more intelligence can make the machine harder to challenge rather than easier to reform.
Follow the balance sheets
Two operations can differ in authority, funding, scale, timing, and legal machinery while still pushing on the same economic channel. That is the whole point of comparing them.
The Treasury manages the federal government’s financing. The Federal Reserve conducts monetary policy and provides reserve accounts to eligible institutions. A Treasury security held by an investor is a claim on Treasury. A reserve balance is a liability of the central bank held by an eligible account holder. A household does not hold a reserve account in the way a bank does. Those distinctions affect what happens when the ownership of an asset changes. [S129][S132][S134]S129 · Quarterly Refunding Statement, August 5 2026U.S. Treasury · 2026official policy statement · current primary sourceS132 · The Mechanics of LSAP TransmissionFederal Reserve Bank of New York · 2010central-bank research/speech · QE mechanismS134 · The Case for Removing Mortgage-Backed Securities from the Federal Reserve's Balance SheetChristopher J. Waller / Federal Reserve Board · 2025official central-bank speech · maturity-transform mechanism
For an illustrative Treasury buyback, suppose Treasury repurchases an outstanding long-term bond. The investor gives up that security and receives payment. Treasury has to finance its overall cash needs through its available cash and financing operations. To know the net effect on the maturity of debt held outside the public sector, we need to examine the issuance that accompanies the buyback, not merely the bond that was purchased.
If the transaction is paired with an equal amount of short-term issuance, the private sector’s exposure to long-duration government debt may decline. If it is paired with comparable long-term issuance, much of that duration can return. If the operation mainly replaces a less liquid security with a more liquid one, its most important effect may concern market functioning rather than a large change in aggregate duration.
These are simplified cases, but they expose the structure. Gross purchases are not net debt retirement. A maturity bucket is not a complete measure of interest-rate risk. A scheduled maximum is not the amount actually accepted in an operation.
Similar channel. Different machinery.
The seller receives payment. Treasury finances its overall cash needs through available cash and financing operations.
What accompanies the purchase?
Short-term issuance can reduce private-sector duration exposure. Comparable long-term issuance can put duration back. Gross purchases alone do not establish net debt retirement.
Settlement involves bank reserves; when a nonbank sells, its bank deposit is also part of the chain.
Is the purchase price a gift?
No. The seller exchanges an asset for another claim. Any gain in wealth requires a separate account of price, risk or other financial effects.
Both can affect the duration held in private portfolios.
For a simplified Federal Reserve purchase, the central bank acquires a security and the banking system receives reserve balances as settlement occurs. When a nonbank sells, the seller’s bank deposit and the bank’s reserve position are part of the settlement chain. The details matter for liquidity and portfolio choice. They do not mean that the seller receives a gift equal to the purchase price: the seller has exchanged an asset for another claim. [S132][S134]S132 · The Mechanics of LSAP TransmissionFederal Reserve Bank of New York · 2010central-bank research/speech · QE mechanismS134 · The Case for Removing Mortgage-Backed Securities from the Federal Reserve's Balance SheetChristopher J. Waller / Federal Reserve Board · 2025official central-bank speech · maturity-transform mechanism
The possible gain to an asset holder may arise from price effects, reduced risk, or other changes in financial conditions. It is not established by counting the entire purchase price as newly created wealth. That would confuse an asset exchange with a transfer of net worth.
Duration is the channel I care about
A bond’s maturity tells us when its principal is due. Duration summarizes sensitivity to changes in yields and also depends on the timing of cash flows. Two bonds with the same maturity can have different duration. A comparison that uses maturity alone can therefore be suggestive without being complete. The portfolio-balance explanation of asset purchases concerns the risks investors are left holding, not only the face value of securities exchanged. [S132][S134]S132 · The Mechanics of LSAP TransmissionFederal Reserve Bank of New York · 2010central-bank research/speech · QE mechanismS134 · The Case for Removing Mortgage-Backed Securities from the Federal Reserve's Balance SheetChristopher J. Waller / Federal Reserve Board · 2025official central-bank speech · maturity-transform mechanism
Imagine investors collectively hold a fixed amount of interest-rate risk. An official operation removes some of that risk. Investors may seek substitutes, changing the prices and yields of other assets. That is the broad logic of the duration channel. Its strength depends on market conditions and on how readily investors substitute among assets. It is not a promise that a given purchase will move every asset price in the same direction.
The Kansas City Fed working paper distinguishes debt expansion from maturity extension rather than treating all Treasury supply changes as one shock. Its identified shocks have different effects on yields, risk premia, and activity, and the authors find that debt-management choices can reinforce or offset Federal Reserve purchases. [S130]S130 · Treasury Supply Shocks: Propagation Through Debt Expansion and Maturity AdjustmentHuixin Bi; Maxime Phillot; Sarah Zubairy / Federal Reserve Bank of Kansas City · 2026Fed research working paper · mechanism evidence
That is the comparison I was making. I do not need Treasury buybacks and QE to be identical. I need us to stop pretending that institutional differences erase an overlapping duration channel.
The August 19, 2026 announcement raised the maximum size of certain long-end liquidity-support operations from $2 billion to at least $4 billion. The specified operations ran from September 9 through November 4. Treasury described the change in terms of liquidity support and the strength of offers. That statement identifies the program and its stated purpose. It does not, by itself, establish the magnitude of a macroeconomic stimulus or the motive of any official beyond what the announcement says. [S128]S128 · Treasury Announces Increased Sizes of Nominal Long-End Liquidity Support Buybacks Beginning September 9U.S. Treasury · 2026official policy announcement · current primary source
The August 5 quarterly plan preceded that change. Its purchase caps are therefore a dated plan, not an immutable total for the later quarter. Keeping the dates visible prevents two individually correct statements from being combined into a misleading account. [S129]S129 · Quarterly Refunding Statement, August 5 2026U.S. Treasury · 2026official policy statement · current primary source
Liquidity is useful, and not everything useful is neutral in effect
A liquid market allows participants to buy and sell without unusually large price concessions. Better market functioning can reduce costs and improve the reliability of financing. Those benefits are not imaginary merely because financial institutions participate in the market. Treasury’s buyback materials describe liquidity support and cash management as program functions. [S128][S129]S128 · Treasury Announces Increased Sizes of Nominal Long-End Liquidity Support Buybacks Beginning September 9U.S. Treasury · 2026official policy announcement · current primary sourceS129 · Quarterly Refunding Statement, August 5 2026U.S. Treasury · 2026official policy statement · current primary source
Market functioning and distribution are different accounts of the same intervention. A change can improve the first while affecting balance sheets unequally. That does not make the stated purpose false. It means one outcome measure is not enough.
Consider a stylized household comparison. One household owns a home and has a mortgage it can refinance. Another rents and is trying to save a down payment. Lower long-term rates may reduce the first household’s financing costs. They may also contribute to higher asset prices, depending on supply, demand, and other conditions. The second household could gain through employment while facing a different entry price for housing. No single sign on an interest-rate change describes both households’ whole experience.
A retiree holding interest-bearing assets faces a different set of channels. A business considering an investment faces another. A worker whose job depends on demand may benefit even without owning a substantial portfolio. The categories overlap because people have several economic roles at once.
This is why the New York Fed’s analysis of QE and inequality is more useful than a simple claim that asset purchases help only the rich or help everyone equally. The study considers labor-market and asset-income channels that can move different parts of the distribution differently. Its findings are specific to its model and empirical setting, not a universal accounting identity about every intervention. [S135]S135 · Unconventional Monetary Policies and InequalityFederal Reserve Bank of New York · 2024central-bank staff report · distributional mechanism
The key lesson is methodological: trace the channels through the relevant balance sheets and income flows. Do not infer a person’s total experience from one asset price or one aggregate measure. A distributional analysis is not complete because it has found one group that benefits or one group that loses.
What the public should be able to inspect
Evaluating an intervention still requires a counterfactual. A policy can have distributional costs and prevent larger losses. It can help in the short run and create vulnerabilities later. None of that excuses turning the explanation into a priesthood.
I want ordinary readers to be able to tell which result is observed, which is estimated, which is a model scenario, and which is a judgment about acceptable trade-offs. That is not dumbing the subject down. It is the minimum required for democratic scrutiny.
An AI system could help maintain those distinctions. It could compare dated documents, calculate a net maturity change, and explain assumptions in plain language. It could also produce an eloquent account that blurs them. The quality of the explanation depends on the evidence and the task, not merely on the confidence of the voice.
Debt service and the range of future choices
CBO’s baseline is a conditional projection under specified assumptions and law, not a statement that the future is already fixed. Its projected debt and interest figures describe a path that can be altered by economic developments and subsequent decisions. The baseline’s value is that it makes a set of assumptions and their implications inspectable. [S136]S136 · The Budget and Economic Outlook: 2026 to 2036Congressional Budget Office · 2026official fiscal projection · current fiscal baseline
Net interest is also not the same as the primary deficit. The primary balance excludes net interest. Separating the two helps show whether a change in total borrowing comes from the gap between noninterest spending and revenue, from the cost of servicing debt, or from both. CBO’s 2026–2036 baseline has different paths for those components. [S136]S136 · The Budget and Economic Outlook: 2026 to 2036Congressional Budget Office · 2026official fiscal projection · current fiscal baseline
A growing interest bill creates a resource constraint, but the constraint does not dictate one political response. Revenue, spending, growth, inflation, interest rates, and the maturity of financing can affect the path in different ways. Their trade-offs are matters for evidence and public decision, not a conclusion that follows mechanically from one debt ratio.
The maturity structure also changes exposure to future rates. Shorter-term financing can reprice sooner. Longer-term financing can lock in a rate for longer while carrying a different current cost. There is no universally cheapest maturity independent of future rates and risk. A strategy that appears inexpensive under one path can look costly under another.
This is where the phrase “optimization” can conceal a choice about risk. Minimize expected cost over what period? Subject to which refinancing risks? How are low-probability but costly outcomes treated? Which uncertainty is included in the model, and which is left outside it? A more capable optimizer can answer the specified problem more effectively while leaving those choices unresolved.
The issue is not unique to government financing. A corporation or household can also lower an immediate payment by accepting more future risk. The immediate saving is real. So is the exposure. A useful explanation keeps both in the same frame.
The dollar question
Reserve composition is another place where a striking narrative can outrun a narrow statistic. The IMF’s COFER data describe the currency composition of reported foreign-exchange reserves. They do not include every international use of a currency, and gold is not part of that currency-share denominator. A change in share can also reflect exchange-rate valuation, not only active purchases and sales. [S137]S137 · COFER Data Brief, First Quarter 2026International Monetary Fund · 2026official reserve statistics · current reserve-currency baseline
In the first quarter of 2026, the reported dollar share rose from 56.42% to 57.13%. The IMF attributed a substantial part of the change to valuation effects. That quarter does not support an immediate-collapse story. It also does not settle the longer questions about diversification, invoicing, cross-border finance, sanctions exposure, or payment infrastructure. [S137]S137 · COFER Data Brief, First Quarter 2026International Monetary Fund · 2026official reserve statistics · current reserve-currency baseline
Financial intelligence and public understanding
AI could make financial complexity easier to inspect. A capable system could reconcile announcements with executed operations, distinguish gross from net flows, identify revisions, and show how a conclusion changes under different assumptions. These are plausible uses, not claims that such an audit has been completed for every operation discussed here.
The same capability could be used to optimize a narrow target while making the resulting explanation more persuasive. A model might recommend a strategy because it performs well under a chosen forecast and then generate language that makes the strategy sound inevitable. The danger is the promotion of a conditional result into an unconditional conclusion.
That promotion is not prevented by adding more intelligence to the forecasting stage. It requires preserving the assumptions, alternatives, uncertainty, and authority behind the decision. A reader should be able to distinguish “this is what the model estimates under these conditions” from “this is the only responsible choice.” The first is an analytical statement. The second contains a judgment that the model does not independently authorize.
This is the bridge to the rest of the essay. Complex systems allocate resources through decisions whose effects are difficult to see from an ordinary household’s position. Better intelligence can illuminate the machinery. It can also give the institution describing that machinery a faster, cleaner, more persuasive way to tell you what it means.
The difference lives in two boundaries: explanation versus persuasion, and the institution’s objective versus the lives affected by it. Lose either one and a very intelligent explanation can become a very efficient way to make somebody else’s priorities sound inevitable.