The Reweight

On February 16, 2026, Michelle W. Bowman, the Federal Reserve's Vice Chair for Supervision, addressed the American Bankers Association Conference for Community Bankers in Orlando, Florida. The Vice Chair for Supervision is the Board's senior official responsible for the prudential regulation of the American banking system. When she previews regulatory proposals at an industry conference, she is not floating ideas. She is setting an agenda.

The speech announced that two proposals would soon be introduced to modify the regulatory capital framework for mortgage activities. The first would remove the requirement to deduct mortgage servicing assets from regulatory capital while maintaining the 250 percent risk weight. The second would introduce loan-to-value sensitive risk weights for residential mortgage exposures, replacing the current uniform treatment. Together, these changes would loosen the post-crisis capital framework for bank mortgage operations — a framework the Federal Reserve itself designed and implemented beginning in 2013.

The case for these changes rested on a specific diagnosis. The 2013 capital treatment, Bowman argued, was an over-calibration that drove banks from the mortgage market, concentrated origination and servicing in under-regulated nonbank institutions, reduced consumer choice, raised costs, and disproportionately harmed low-to-moderate income borrowers and community banks. She described the current risk weights as "disproportionate to risk" and said she was "open to revisiting whether the capital treatment of MSRs and mortgages is appropriately calibrated."

The diagnosis has surface plausibility and institutional weight. Banks originated roughly sixty percent of mortgages in 2008; by 2023, that share had fallen to thirty-five percent. Bank servicing dropped from ninety-five percent to forty-five percent over the same period. Nonbank servicers, as the Financial Stability Oversight Council documented in 2024, operate without the resolution frameworks, deposit funding, or supervisory infrastructure that govern banks. Community banks — the audience in the room — have been squeezed between the fixed costs of servicing technology and a capital framework designed with the largest institutions in mind. The problems are real. The framework is thirteen years old. The speech sounds overdue.

These are testable claims. The Federal Reserve's own documentary record — transcripts of committee deliberations, Tealbook staff analysis, internal memoranda, and public communications spanning the design and implementation of the post-crisis capital framework — contains the institution's contemporaneous assessment of every trade-off Bowman proposes to revisit.

We went into the archives.

I. The Deliberate Design

The foundation of Bowman's case is a word: recalibration. The speech acknowledges that regulators tightened the capital treatment of mortgage servicing rights "for sound reasons" — valuation volatility, model dependence, the propensity of MSR markets to seize during periods of high defaults. It then argues that regulators are now better positioned to reconsider:

The architecture of the speech treats the 2013 capital changes as a well-intentioned overcorrection — an error of calibration, not of judgment — that can now be quietly adjusted.

The documentary record tells a different story. The technical case against MSRs was not a rough estimate made under crisis pressure. It was a carefully constructed analysis, built over years, by officials who understood exactly what these assets were and why they resisted conventional risk measurement.

In August 2005, Patricia Mosser, then Senior Vice President at the Federal Reserve Bank of New York, told the FOMC what made mortgage servicing rights different from ordinary financial assets:

The significance of Mosser's warning was not that MSRs were risky — many assets are risky. It was that the risk was unverifiable. Without proprietary models built on decades of institutional experience, no bank and no regulator could independently assess whether an MSR valuation was sound. Chairman Alan Greenspan reinforced the point in the same meeting, stating that "the prepayment pattern is virtually unknown." The chairman of the Federal Reserve was acknowledging that a core component of mortgage asset valuation was beyond the reach of the institution's own analytical capacity. Mosser added that hedging these risks was "not as straightforward as buying an interest rate option" because MSRs lacked the natural hedges available in standard markets. The asset was unmodelable, unhedgeable by conventional means, and opaque to outside assessment.

By March 2008, the crisis confirmed Mosser's analysis in real time. She described for the committee the mechanism that made MSRs uniquely dangerous during stress:

Duration extension. The asset becomes more volatile precisely when the system is most fragile. When interest rates rise and spreads widen, prepayment speeds slow, stretching the life of the MSR and compounding losses at the moment the institution holding it is least able to absorb them. This is not ordinary market risk. It is risk that amplifies itself.

When the Basel III capital framework was designed in 2010, staff analysis incorporated these warnings. The December 2010 Tealbook was explicit about the expected consequences:

This was not a prediction buried in a footnote. It was staff analysis delivered to the committee during the design phase of the capital framework. The migration of mortgage servicing to the shadow banking sector was identified as a known cost, weighed against the benefits of removing volatile, model-dependent assets from bank balance sheets, and accepted. Governor Elizabeth Duke confirmed the institutional awareness at the committee level two years later, stating that "even though regulatory issues are not, strictly speaking, part of monetary policy," they would "have a strong influence on the structure of the mortgage market, the cost and availability of mortgage credit, and ultimately, the effectiveness of our policies."

The documentary record from this period contains no contemporaneous instance of a Federal Reserve participant — staff, committee member, or supervisory official — framing the bank withdrawal from mortgage servicing as an unintended consequence or a calibration error. In the record we reviewed, the migration is consistently treated as an accepted cost of a deliberate design choice. The framework was built on Mosser's technical analysis, tested against the crisis that validated it, and adopted with full awareness that banks would reduce their mortgage activities.

This does not mean the 2013 framework was beyond revision. It was, and remains, a trade-off — a deliberate balancing of competing risks. On one side: the volatility, model-dependency, and unverifiability of MSR valuations on bank balance sheets. On the other: the costs of driving mortgage activity into a less-regulated sector with weaker resolution frameworks and no deposit funding. The institution weighed these risks and judged, in 2013, that the dangers of the former outweighed the costs of the latter. Trade-offs can legitimately be revisited when conditions change. The question is whether Bowman's speech demonstrates that the risks on the 2013 side of the ledger — the specific vulnerabilities Mosser identified — have diminished enough to justify shifting the balance. The speech asserts that regulators are "much more familiar with MSRs" since the framework was implemented. It does not show that familiarity has resolved the model-dependency, created the natural hedges that were absent, or eliminated the duration extension that amplifies losses during stress. Familiarity with a risk is not the same as reduction of a risk. The word for what Bowman proposes is not recalibration. It is a reversal of the trade-off — and it is being proposed without demonstrating that the original risks have changed.

If the 2013 framework was a deliberate design rather than an accidental overcorrection, the question shifts. Was the framework at least wrong about the cause? Did the capital treatment drive the migration of mortgage activity to nonbank institutions, as Bowman contends?

II. The Structural Shift

Bowman states that "we can clearly see from academic research and industry feedback that the 2013 change in capital treatment of mortgage servicing rights was a factor in the withdrawal of banks from the mortgage market." She is careful with the qualifier — "a factor." But the architecture of the speech treats it as the factor, proposing capital relief as the remedy that will "increase bank incentives to engage in mortgage origination and servicing."

The Federal Reserve's own internal analysis — its staff surveys, its Tealbook briefings, its committee deliberations — tells a more complicated story. In April 2013, staff reported the results of the Senior Loan Officer Opinion Survey to the committee:

Three-quarters of banks said the problem was not the capital charge on MSRs. The problem was that mortgages were not profitable enough compared to other uses of their capital. This is an opportunity-cost argument, not a capital-adequacy argument. Even if the MSR deduction were removed entirely, banks might still avoid mortgages because other assets offer better risk-adjusted returns for less operational burden.

The survey identified a second constraint that was, if anything, more powerful. Almost all banks — on a portfolio-weighted basis — cited the risk of putback of delinquent mortgages by the Government Sponsored Enterprises as a primary restraint on their willingness to originate. Putback risk is a legal and contractual exposure. No adjustment to MSR risk weights can address a bank's fear that Fannie Mae or Freddie Mac will force it to repurchase loans that later default.

Inside the committee, Nellie Liang, Director of the Division of Financial Stability, provided what may be the most revealing explanation of nonbank growth in the documentary record:

The competitive advantage of nonbank mortgage firms was not that they were unburdened by MSR capital charges. It was that they operated under a fundamentally different legal structure — one that permitted leverage levels no bank could match regardless of its capital treatment. The migration was not a regulatory side effect. It was structural arbitrage embedded in the legal framework of American finance.

By 2016, the Federal Reserve's own public assessment had reached a quiet verdict. A Board report acknowledged that mortgage lending was affected by many factors, including technology and GSE practices, and "perhaps to a limited degree, changes in capital and liquidity regulations within the banking sector." The institution that designed the capital framework characterized its own rules as contributing "perhaps to a limited degree" to the market shift. Jerome Powell, then a Governor, observed in 2017 that "almost nine years into this time out, the federal government's domination of the housing sector has grown and is actually greater than it was before the crisis." The dominance of government-backed channels — not the capital treatment of MSRs — had reshaped the mortgage market.

The Federal Reserve's macroeconomic models kept projecting that bank mortgage activity would recover as conditions normalized. It never did. By 2018, residential investment was declining for the first time in the recovery even though mortgage rates were close to staff projections. The models had assumed that if rates were low and standards eased, banks would lend. They did not account for the fact that the infrastructure of lending had migrated permanently. The human capital, the origination technology, the servicing platforms had moved to nonbank firms or vanished entirely. The pipeline had not been temporarily blocked by a capital charge. It had been permanently rerouted by structural forces the capital charge did not create and capital relief cannot reverse.

If capital treatment was not the primary driver of the migration, the specific technical proposal in Bowman's speech — introducing loan-to-value sensitive risk weights — must carry weight on its own merits. The archive has something to say about that as well.

III. The Procyclical Trap

Bowman's most technically sophisticated argument concerns the treatment of residential mortgages held on bank balance sheets:

A low loan-to-value mortgage carries less risk than a high loan-to-value mortgage, and the capital framework should reflect this difference. The logic sounds irrefutable.

The Federal Reserve debated exactly this proposal and produced a specific, documented reason for rejecting it. Thomas Hoenig, President of the Federal Reserve Bank of Kansas City, described the mechanism in January 2011:

Caught below the line. Hoenig was describing a feedback loop that LTV-sensitive risk weights would create by construction. When house prices rise, loan-to-value ratios fall, risk weights decline, and capital requirements ease — releasing lending capacity into a market that is already expanding. When house prices fall, the mechanism reverses: LTV ratios climb, risk weights increase, capital requirements tighten, and banks must curtail lending at precisely the moment when credit contraction is most damaging to the real economy. The capital framework does not passively reflect the cycle. It amplifies it.

Federal Reserve staff quantified the amplification. In April 2012, internal modeling estimated that a one-dollar capital shortfall would produce a $3.73 reduction in bank lending for common equity. The nearly four-to-one multiplier means that LTV-sensitive risk weights would not merely track house price movements. They would accelerate them in both directions — forcing leveraged deleveraging during downturns and fueling leveraged expansion during booms.

Even Daniel Tarullo — then a Governor and one of the Board's most rigorous thinkers on financial regulation — acknowledged the trade-off. He admitted that a uniform approach "would not be useful to deploy in response to asset bubbles" but accepted that limitation because the alternative was worse: a system where the regulatory framework itself becomes a driver of the cycle rather than a buffer against it. Loretta Mester noted that other countries had adopted LTV and debt-to-income limits that vary over the cycle — a middle path between the current uniform weight and the static LTV-sensitivity Bowman proposes. The Fed considered that alternative and chose not to adopt it. Bowman does not mention it.

The archive reveals a further irony. In July 2013 — during the very period the post-crisis framework was being finalized — Mosser advocated for loan-to-value ratios not as a tool for loosening capital requirements, but as a macroprudential instrument for tightening them:

Vice Chair Yellen and Governor Raskin aligned with her position. The institutional consensus in 2013 was that LTV ratios should be used to constrain credit — to prevent the kind of low-equity lending that had amplified the preceding crisis. And yet when the Qualified Residential Mortgage definition was finalized in late 2014, LTV ratios were excluded entirely. Mosser herself noted the gap: the QRM "rests, in part, on the debt service-to-income of the mortgage borrower, but it does not include loan-to-value ratios." The institution failed to codify LTV limits even when its own officials wanted them as a tightening tool. It is now being asked to adopt LTV sensitivity as a loosening tool. The same metric, inverted in purpose — from constraining credit excesses to reducing capital requirements.

Beyond the procyclical dynamics, LTV-sensitive risk weights face a practical implementation problem that the speech does not address. Loan-to-value ratios depend on property valuations, and property valuations are subject to appraisal lag, systematic bias during housing booms, and abrupt regime shifts when markets turn. During the expansion phase of the cycle, appraisals tend to validate rising prices, producing LTV ratios that look comfortable; during contractions, the same properties can lose twenty or thirty percent of appraised value within months, pushing LTV ratios through regulatory thresholds in ways that no bank's capital planning anticipated. Any framework that ties risk weights to LTV buckets also creates cliff effects at bucket boundaries — a loan at 79 percent LTV receives a materially different capital charge than a loan at 81 percent, though the actual difference in credit risk is negligible. These are not theoretical objections. They are the measurement errors and model risks that Mosser spent a decade warning about, now embedded directly in the capital framework rather than confined to bank balance sheets.

The uniform risk weight was not a crude instrument adopted for lack of analytical sophistication. It was a deliberate anti-procyclical design choice, made after the institution identified, modeled, and rejected the amplification mechanism Bowman now proposes to reintroduce. The speech contains no discussion of procyclicality, no mention of the feedback loop, no reference to the extensive internal debate that produced the current framework. The omission is notable — the institution debated this mechanism at length, and the Vice Chair for Supervision does not address it.

Mosser's technical analysis, which undermined the keystone of the speech in the preceding section, connects here as well. LTV-sensitive risk weights are themselves model-dependent — they require real-time house price assessments, appraisal accuracy, and valuation methodologies that carry exactly the kind of model risk Mosser identified. The same technical limitations that made MSR valuations "extremely difficult to figure out" apply to any framework that ties capital requirements to asset valuations that are volatile, subjective, and susceptible to systematic error during stress.

A fair reading of Bowman's position would note that the risk environment has not stood still since 2005. MSR hedging markets have matured. Stress testing provides a supervisory backstop that did not exist when Mosser issued her warnings. Banks that hold MSRs today operate under a more developed risk management infrastructure than their pre-crisis predecessors. These are legitimate developments. But the specific vulnerabilities Mosser identified — the dependence on proprietary models that regulators cannot independently verify, the absence of natural hedges in standard markets, the duration extension that amplifies losses precisely during stress — are structural properties of the asset class, not artifacts of an immature market. Better hedging tools reduce the magnitude of the risk. They do not resolve the fundamental unverifiability that justified the conservative capital treatment. The question the speech does not address is whether familiarity with an unmodelable risk makes it safer, or merely makes the institution more comfortable bearing it.

Both the historical rationale for the current framework and the technical case for the proposed replacement have now been tested against the documentary record. The speech must therefore stand on its remaining arguments: that the migration to nonbank servicing threatens financial stability, that loosening capital requirements will improve outcomes for consumers, and that the capital framework disproportionately harms the most vulnerable borrowers.

IV. The Inverted Remedy

Bowman argues that the concentration of mortgage activity in nonbank institutions creates financial stability risks. Nonbank servicers, she observes, are subject to "far fewer safeguards" than banks and lack the resolution frameworks that ensure continuity of core servicing functions when large institutions fail. The Financial Stability Oversight Council published a report in 2024 confirming these vulnerabilities. The COVID-19 forbearance data is real. The diagnosis is documented. The question the archive can answer is not whether the diagnosis is valid — it is — but whether the prescription follows from it.

In April 2020, as the pandemic exposed exactly the fragilities Bowman describes, Federal Reserve staff warned that a large-scale forbearance take-up could cause nonbank servicers to fail, noting that these firms had originated forty percent of all mortgages in 2019. Staff assessed that the Ginnie Mae liquidity facility would not fully cover the shortfall — it excluded advances for taxes and insurance, which constituted a quarter of required payments. The warning was blunt: failure of nonbank servicers would produce a substantial disruption of the aggregate mortgage market.

The committee's response to this analysis moved in a single direction: enhanced prudential oversight of nonbank institutions. Randal Quarles — then Vice Chair for Supervision, Bowman's predecessor in the role she now holds — acknowledged that the pandemic had "made clear that vulnerabilities in the nonbank sector related to liquidity mismatch, to leverage, and to interconnectedness" persisted. Robert Kaplan, President of the Dallas Fed, pointed to the absence of a regulatory regime or authority for nonbank financial institutions and called for FSOC-led coordination to assess lessons and reforms. Staff analysis concluded that nonbank competitive advantages were driven by structural factors and technology rather than bank regulatory costs, and that capital relief would be insufficient to reverse the migration.

No evidence was found in the committee deliberations of officials advocating for loosening bank capital requirements as the primary response to nonbank mortgage fragility. The institution's answer — reached with the Vice Chair for Supervision's direct participation — was to regulate nonbank institutions up. Not to regulate banks down. The directional reversal — from tightening oversight of the less-regulated sector to loosening requirements on the more-regulated one — follows a pattern documented in The Confession, where distinct institutional justifications converge on the same deregulatory outcome.

The most prescient voice in the documentary record on this precise question belongs to Bowman herself. In November 2020, speaking as a Governor, she warned:

She was right. She went further, arguing that the nonbank mortgage model's apparent success should not be confused with resilience:

The diagnosis she offered in 2020 pointed toward tighter nonbank oversight — toward the FSOC coordination and enhanced prudential framework the committee was building. The prescription she offers in 2026 points in the opposite direction. The institution she now leads on supervisory matters concluded, with her participation, that nonbank mortgage fragility required a new regulatory perimeter around nonbank firms. The speech proposes the remedy the institution already considered and set aside.

The consumer cost argument follows a similar arc. Bowman asserts that fewer banks in the market has "reduced the consumer choice and competition that drives down costs." During normal conditions, the archive shows the opposite: nonbank entry expanded the competitive frontier through technology and streamlined processes. But during periods of stress, the nonbank-dominated market structure imposed severe costs on the borrowers it was supposed to serve. In September 2020, staff found:

A full percentage point. Staff attributed roughly half of that spread widening to capacity constraints and forbearance risks inherent in the nonbank model. The consumer cost was real, but the mechanism was not the absence of banks from the market. It was the liquidity fragility of the nonbanks that replaced them. Capital relief for banks does not address the structural vulnerability that caused the cost spike.

The speech also argues that the capital treatment "may particularly affect mortgage availability and affordability" for low-to-moderate income borrowers, since banks securitize roughly 75 percent of their originations to LMI borrowers. If the capital treatment reduces the value of the MSR, the argument runs, the cost falls disproportionately on the most vulnerable.

The Federal Reserve's own staff built an analytical framework — "frontier analysis" — to determine precisely what was constraining LMI credit access. The answer left no ambiguity:

Low-credit-score borrowers were not priced out by the MSR embedded in their mortgage rate. They were excluded entirely by lender policy — by GSE putback risk, by rigid underwriting overlays, by FICO and debt-to-income thresholds that severed them from the credit market. Governor Duke had identified the shift as early as 2013, observing that "many of these households, especially the younger households, may not qualify for credit in the current underwriting environment." When Fannie Mae finally relaxed its debt-to-income constraints in 2017, it immediately added thirty thousand loans per month to the market. That was the lever. Not risk weights. Not MSR deductions. The underwriting parameters set by the Government Sponsored Enterprises are the bottleneck for LMI credit, and no amount of bank capital relief reaches through that bottleneck to the borrowers Bowman invokes.

The stability, consumer cost, and equity arguments all follow the same pattern: the diagnoses are valid, and the prescriptions do not reach them. One argument remains — the most specific, the most personal to the audience Bowman was addressing, and the one the archive treats with the greatest sympathy.

V. The Genuine Burden

Bowman was speaking at a conference for community bankers, and the most audience-specific claim in her speech concerns the MSR deduction threshold:

Smaller banks cannot build servicing portfolios of sufficient scale without triggering the disproportionately high capital charge. This claim, unlike the others, finds direct support from the archive's most credible witnesses.

George — a career bank supervisor, not a deregulation advocate — identified the MSR capital deduction as a specific and disproportionate burden on community banks a full year before the framework took effect. She was not arguing that capital requirements should be lower for the banking system. She was arguing that the MSR deduction imposed costs on small institutions that bore no relationship to the risks those institutions actually posed.

Elizabeth Duke, herself a former community banker before joining the Board, documented the operational reality in terms that would have been familiar to every person in the Orlando conference room:

Duke was describing a dual burden. Community banks lacked the scale to spread the fixed costs of mortgage servicing across a large enough portfolio to be profitable. They lacked the technology budgets to maintain the compliance systems the regulatory framework demanded. And then the MSR deduction threshold imposed a capital penalty on top of these existing disadvantages, further compressing the margin on an activity that was already marginal for smaller institutions. Duke warned that community banks "simply do not have the resources to appropriately comply with all of the pending regulatory changes" and that the costs fell "disproportionately on them."

Inside the committee, the concern was stated without equivocation. During deliberations in April 2014, the point was made explicitly:

The Federal Reserve received over two thousand comment letters on the Basel III proposals — a volume so significant that the agencies delayed the effective date of the rules to reassess the impact on community banks. Duke and George were not outliers. They represented a documented institutional awareness that the capital framework would disproportionately disadvantage exactly the institutions Bowman was addressing in Orlando.

And yet the archive also shows that community banks were withdrawing from mortgage servicing before the 2013 capital changes took effect. Duke herself identified the cause: the lack of economies of scale, the dependence on purchased software, the inability to absorb the compliance costs that the evolving regulatory environment demanded. Staff analysis recognized by 2016 that technology adoption was a primary driver of nonbank entry — not the absence of bank-friendly capital rules, but the presence of nonbank-friendly operational models that community banks could not match regardless of their capital position. The deduction threshold exacerbated the community bank withdrawal. It did not cause it.

George and Duke were right that the threshold imposed an unnecessary additional burden — a burden of "little public benefit," in George's words. Removing it would address a documented regulatory inequity. On this narrow question — whether the MSR deduction threshold imposes disproportionate costs on community banks for insufficient prudential benefit — the documentary record supports the direction of Bowman's proposal. George made this case fourteen years ago, and the archive validates her judgment.

But the structural forces that drove community banks from mortgage servicing — the scale economics, the technology gap, the compliance infrastructure — predate the Basel III capital framework and will outlast any capital relief. The institutions that left mortgage servicing did not leave because a capital charge made their MSRs less valuable. They left because the business itself became unviable at community bank scale. Removing the deduction threshold corrects the regulatory overlay. It does not rebuild the operational foundation. The institution knew both things simultaneously: the rules were disproportionately costly, and the costs were not primarily regulatory. Capital relief is a defensible component of a broader strategy. The speech presents it as the strategy itself.

VI. The Available Instrument

Every claim in the speech has now been tested against the documentary record. The keystone — that the 2013 capital treatment was an over-calibration — falls against the evidence that it was a deliberate design choice made with full awareness of the costs. The causal claim — that capital treatment drove the nonbank migration — falls against the Fed's own finding that structural factors, not capital charges, were primary. The technical proposal — LTV-sensitive risk weights — falls against the institution's documented analysis of the procyclical amplification mechanism and the implementation problems of valuation-dependent regulation. The financial stability argument — that nonbank fragility requires loosening bank capital — falls against the institution's own conclusion, reached with Bowman's participation, that the appropriate response was to tighten nonbank oversight. The equity argument — that capital treatment constrains LMI credit — falls against the staff's frontier analysis showing that underwriting overlays and GSE putback risk, not capital requirements, are the binding constraints. Only the community bank burden survives, validated by the archive's own witnesses — but even there, the structural forces that drove the withdrawal predate the capital framework and capital relief alone cannot reverse them.

The speech itself reveals why the proposals take the form they do. Bowman acknowledges that capital treatment is "only a small part of the broader mortgage problem" and that comprehensive remedies would require action by the Consumer Financial Protection Bureau, legislative reform, and GSE restructuring — domains that implicate political conflicts the Federal Reserve cannot resolve unilaterally. She then proceeds to propose changes exclusively within the capital framework the Fed controls. This is feasible-policy substitution: adjusting what the institution can change on its own authority while leaving the binding constraints — the GSE underwriting standards, the putback frameworks, the nonbank regulatory gap — to other actors and other timelines. The result is a proposal calibrated to the limits of the Fed's jurisdiction rather than to the structure of the problem. Given everything the archive has shown, the degrees of freedom available to the Vice Chair for Supervision have been exhausted by the constraints she herself identifies. What remains is the instrument she can reach — and the speech is honest enough, in a single paragraph, to say so.

If this reading is correct — if the proposal is a political reweighting rather than a response to diminished risk — the outcomes should be distinguishable from those of genuine recalibration. We should expect the capital relief to proceed without corresponding FSOC designation or enhanced prudential oversight of nonbank servicers. We should expect GSE underwriting constraints — the binding bottleneck for LMI credit — to remain unchanged. We should expect any LTV-sensitive risk weight framework to omit countercyclical adjustment mechanisms. And we should expect that even after implementation, banks will not materially re-enter mortgage servicing at scale, because the structural advantages of nonbank originators — technology, operational cost, legal structure — will persist regardless of how the capital charge is calibrated. If, on the other hand, banks substantially expand mortgage servicing, nonbank oversight tightens in parallel, and LMI credit access measurably improves, the archive-based critique offered here will have been wrong about the binding constraints. These are observable outcomes. The next several years will test them.

The Reweight