← Research

Cheap Capital, Expensive Execution

How Treasury’s ECIP gave tiny community banks more cheap capital than their market values—and why management execution determined what reached common shareholders.

On April 29, 2022, Citizens Bancshares published an annual report containing an extraordinary disclosure. The Atlanta community bank had received approval for $90.9 million from the U.S. Treasury. The final investment would reach $95.7 million. Citizens’ entire common equity traded for roughly $23 million.

The terms were even more unusual than the size. Treasury did not receive common stock. It bought perpetual, noncumulative senior preferred shares that carried no dividend for two years and a nominal rate thereafter. The capital counted as Tier 1, had no maturity and caused no common-share dilution.

The disclosure was public. The contract was public. The stock barely traded.

An anonymous investor writing as @dirtcheapstocks found Citizens through the annual report, studied the other recipients and began circulating the idea. M&F Bancorp had disclosed an even larger capital-to-market-value gap seven weeks earlier, but Citizens was the case that first circulated widely. Over the following year, Tim Eriksen of Cedar Creek Partners, Dave Waters of Alluvial Capital, Whit Huguley of River Oaks Capital, Fairlight Alpha and Andrew Porter of Stepping Stone Partners published their own ECIP work.

They were looking at a fact that normal bank valuation screens did not capture. Treasury ultimately invested $8.57 billion in 175 community financial institutions. At least 13 recipients were publicly traded. For several, the new capital matched or exceeded the market value of the common shares.

The historical results require two clocks. Starting from late-September to mid-October 2022 prices, the 13 stocks in Andrew Porter’s later public-company list produced a raw median marked return of 2.17x through July 22, 2026. The SPDR S&P Regional Banking ETF (KRE) returned roughly 1.35x including reinvested distributions. “Marked” means the paper result from quoted prices and dividends, not a realized fund return.

But Porter did not publish the 13-name list until July 2023. Starting every stock and KRE on July 17, 2023 instead, the ECIP median was 1.78x versus 1.87x for KRE.

The later screen therefore does not prove an automatic excess-return basket. ECIP was not free cash: Treasury ranked ahead of common, many banks struggled to deploy the capital and a few spectacular returns existed only at stale OTC quotes.

The evidence for mispricing instead comes from managers who bought before the outcome, the extreme original capital-to-market-value gaps and the company-level conversion routes. ECIP created a rich hunting ground; management quality determined how much reached each common share.

Why the Contract Was Worth More Than It Looked

ECIP was designed to expand lending in communities underserved by the financial system. Treasury invested through community development financial institutions (CDFIs) and minority depository institutions, using preferred stock for banks and subordinated debt for credit unions.

For bank recipients, the standard preferred-stock certificate provided:

  • no mandatory maturity;
  • no dividend for the first 24 months;
  • a 0.5%, 1.25% or 2.0% dividend thereafter, based on qualified-lending growth;
  • noncumulative dividends;
  • Tier 1 regulatory-capital treatment; and
  • priority over common shares for dividends and liquidation.

In plain English, Treasury supplied cash the bank never had to repay on a schedule. A skipped dividend did not accumulate as a future debt. The capital counted toward the regulatory cushion that absorbs losses and supports bank assets. Treasury still stood ahead of common shareholders for dividends and liquidation.

The face value and the economic value were therefore radically different. A $100 million liquidation preference was not $100 million of common book value. Treasury owned the senior claim. Yet a perpetual instrument costing at most $2 million a year was also nothing like $100 million of ordinary bank equity or market-rate debt.

At a 2% dividend, $100 million of ECIP capital cost $2 million annually. At 0.5%, it cost $500,000. As short rates rose through 2022–23, a bank could put the cash into Treasury securities yielding 4%–5% and earn a spread without making a new loan. It could then use the enlarged capital base to grow loans, buy another bank for cash, repurchase cheap common stock or make itself more valuable to an acquirer.

This was the analytical edge: value the instrument by its economics, but do not pretend that common holders owned its face value.

Some early analyses crossed that line. River Oaks’ original “fire sale” framing treated the ECIP cash as if it would flow to common holders in a liquidation. It would not; Treasury ranked first. Stepping Stone’s later comparison included senior preferred equity in some common price-to-book calculations. That made several banks look cheaper than they were.

Those mistakes overstated the downside case but did not erase the opportunity. Common holders did not need to own the ECIP cash directly. They needed the bank to earn more on the capital than it paid Treasury, then retain or distribute the residual on an accretive per-share basis.

Several Managers Found the Same Anomaly—and Bought Different Banks

The contemporary letters matter because they show that ECIP was not a single-stock discovery reconstructed after the fact.

Tim Eriksen’s Cedar Creek Q3 2022 letter called ECIP the fund’s best opportunity in some time. Cedar built large positions in Citizens Bancshares and M&F Bancorp. It bought Citizens around $12 and M&F below $9. Eriksen saw three routes to value: organic loan growth, a cash acquisition or spread income while management waited.

Alluvial took a different approach. Its Q3 2022 letter disclosed a 6.7% basket of BankFirst Capital, United Bancorporation of Alabama and Bay Community Bancorp. Dave Waters rejected a mechanical ranking by ECIP award relative to market value. Cheap capital would not change a bank’s character: disciplined lenders could make more good loans, while weak operators could make twice as many bad ones.

River Oaks made Citizens its largest position and also owned BankFirst and M&F. Fairlight bought Citizens and M&F after crediting the original dirtcheapstocks research. In July 2023, Stepping Stone published the most complete public-company comparison and selected a six-name basket: PCB Bancorp, Bay Community, BankFirst, Broadway Financial, Security Federal and United Bancorporation.

The limited overlap reflected different judgments and, in some cases, the practical difficulty of building positions in thin OTC stocks. Each bank entered ECIP with different credit quality, deposit costs, operating scale, management incentives and acquisition capacity. The contractual anomaly identified the candidates; it did not rank them.

The Easy Earnings Arrived First

ECIP closed at an unusually favorable moment. The preferred initially cost nothing while the Federal Reserve raised short-term rates. Banks could earn a growing spread in Treasury securities before building the staff, deposits and loan pipeline required for full deployment.

Citizens and M&F reported sharp earnings increases. Citizens’ net interest income rose from $16.0 million in 2021 to $34.2 million in 2023, while common EPS increased from $1.97 to $6.27. M&F’s EPS rose from $1.36 to $2.72. The 2023 regional-bank panic also highlighted the solvency value of permanent Tier 1 capital.

The same rate increase then exposed the limit of the easy thesis. Banks could reinvest securities and originate new loans at higher yields only as old assets matured or borrowers appeared. Depositors could demand higher rates immediately. The ECIP dividend also began after two years. Banks that had not built an earning-asset or capital-allocation plan saw the initial spread compress.

By 2024, the outcomes had begun to separate. The capital subsidy remained real; the question was which management teams could convert it for common holders.

Five Routes From Cheap Capital to Common-Share Value

The successful banks used five distinct routes. Most used more than one.

1. Earn the spread and repurchase common shares: Citizens Bancshares

Citizens was the cleanest operating result. The bank received $95.7 million against an April 2022 common market value near $23 million. From the annual-report disclosure at a dividend-adjusted $10.73 per share to July 22, 2026 at $60, the marked total return was 5.59x. From the October 2022 public-write-up date, it was 3.27x. The scorecard later in this article uses 3.27x because October is the common public-thesis anchor.

Citizens did not double its loan book. The capital first increased cash and low-risk securities income, then moved gradually into securities and loans. Net interest income more than doubled by 2023 even though loans rose only modestly. Citizens also avoided major credit losses and retired undervalued common shares. Shares outstanding fell from 1.98 million at the end of 2022 to 1.75 million at the end of 2025. Dividends increased. EPS reached $6.47 in 2024 before falling to $4.72 in 2025.

Cedar sold its Citizens position back to the company in July 2025 after reporting a 2–3x gain. Eriksen still considered the shares cheap, but concluded that management had taken too long to deploy the capital. River Oaks retained its position and placed more value on the continuing earnings, buybacks and residual strategic option.

Both decisions can be rational. Citizens created substantial common value without retiring the preferred. The unresolved option remained, but so did Treasury’s full $95.7 million senior claim.

2. Buy other banks for cash: BankFirst Capital

BankFirst received $175 million when its common equity was worth roughly $167 million. It used its enlarged balance sheet to acquire Tate Financial and Mechanics Banc without issuing common shares. Mechanics arrived with its own $43.6 million Treasury preferred. Because the instrument’s economic cost was far below face value, purchase accounting valued it at $9.2 million. That helped BankFirst book acquired net assets $4.5 million above what it paid. Treasury’s legal face claim remained $43.6 million.

From September 30, 2022 through July 22, 2026, BankFirst’s quoted price rose from $37.25 to $62.75. Including $3.80 of dividends, the marked return was 1.79x.

The result was positive but slower than Alluvial expected. Waters sold BankFirst in early 2025 because another acquisition had not appeared and United Bancorporation offered a better use of capital. River Oaks continued to hold BankFirst.

The disagreement again concerned opportunity cost, not whether ECIP had value. BankFirst proved that cheap permanent capital could fund accretive cash acquisitions. It did not prove that management could repeat the process indefinitely.

3. Sell the bank and carry the ECIP into a larger institution: Bay Community

Bay Community offered the first clean strategic endpoint. It received $119.4 million when its common market value was approximately $65.5 million. CBC Bancorp agreed to pay $14 cash per common share and completed the acquisition on November 1, 2024.

From the October 14, 2022 public-thesis mark of $8.95, the cash consideration plus $0.40 of dividends produced a 1.61x simple total return.

The percentage gain was less dramatic than Citizens’, but the endpoint was executable. CBC’s closing release said it preserved Bay Community’s $119.4 million of ECIP capital and recorded no goodwill because acquired asset value exceeded purchase cost.

Bay Community showed that a strategic buyer could capitalize the subsidy without first retiring Treasury’s preferred.

4. Grow and repurchase shares—but leave capital idle: United Bancorporation

United received $123.75 million against a March 2022 common market value of about $64 million. Loans rose 24% from 2022 through 2025, while common shares fell from 3.77 million in 2021 to 3.06 million in 2025. Reported earnings also included intermittent non-ECIP support: 2024 contained $10.35 million of CDFI award income. From the October 2022 anchor through July 22, 2026, the marked total return was 2.37x including dividends. Alluvial reported a 70% gain by June 2024 and continued to hold the shares.

This was a successful common-stock result, but the bank remained overcapitalized and the preferred continued to pay 2%. In July 2026, Merion Road and Blue Hill called for a $40 million tender, arguing that expenses and idle capital still obscured the value.

United illustrates the difference between creating value and completing the conversion.

5. Preserve strategic optionality after the operating thesis fails: M&F Bancorp

M&F was the most dramatic and most easily misread outcome. On March 8, 2022, it disclosed a $76 million ECIP allocation while its 1.98 million common shares traded at $7.10, implying a market value near $14.1 million. The final investment reached $80 million.

The spread thesis worked first. The operating thesis then failed. Deposit interest expense and the preferred dividend rose, while common EPS fell from $2.72 in 2023 to $1.81 in 2024 and $1.19 in 2025. Cedar exited in 2024. M&F had not turned the balance-sheet advantage into durable standalone earnings.

On July 22, 2026, Optus Financial agreed to acquire M&F. The fixed consideration of $46.57 per share was 6.56x the March 2022 price. An additional $6.73 depends on repurchasing specified preferred shares. If paid in full, total value would reach $53.30, or 7.51x the original March anchor. The 3.05x scorecard result uses the later October 2022 public-write-up price and the announcement-day close; it does not count the contingent payment.

The deal has not closed, and the contingent payment should not be treated as cash. Yet the announcement shows why Cedar’s sale and the original thesis can both be defensible. Cedar correctly diagnosed the failed operating route. A strategic buyer later valued the remaining franchise and capital option.

The Sixth Outcome: A Large Subsidy That Went Nowhere — Broadway Financial

Broadway received $150 million and had one of the largest ECIP-to-market-value ratios in Stepping Stone’s table at 1.87x. It also produced the weakest marked result: 1.10x through July 22, 2026.

The problem was not the contract. Net interest income was almost unchanged at $32.9 million in 2022 and $33.1 million in 2025, while recurring non-interest expense rose from $24.9 million to $31.3 million. The unreduced 2% ECIP dividend cost $3 million in 2025 and exceeded normalized parent earnings. Tangible common book value per share rose only 9.8% through 2025. Broadway paid no common dividend, and issuance offset its small 2023 repurchase.

The original valuation also looked more extreme than it was. Stepping Stone’s 0.23x price-to-book calculation omitted two non-voting common classes from market value while including the $150 million senior preferred in book value. Including all three economic common classes and excluding the senior preferred produces approximately 0.52x common book and 0.67x tangible common book at the price before Broadway’s November 2023 reverse split.

Broadway was still cheap. But a large subsidy cannot rescue a cost base that absorbs the benefit. Across the full screen, ECIP relative to market value had no visible relationship with the subsequent marked return. The ratio found anomalies; it did not rank outcomes.

The Scorecard

Andrew Porter’s 13-name screen is the best contemporary public list of tradeable ECIP recipients. It is not a legal census of all 175 recipients and does not represent any manager’s actual portfolio. The ratio column uses market values from Porter’s July 2023 paper. The return column uses earlier, late-September to mid-October 2022 public-thesis prices and therefore includes hindsight.

BankTickerECIP / Jul-23 mcapMarked outcome
NMB FinancialNMBF0.63x10.90x†
Citizens BancsharesCZBS1.16x3.27x
Harbor BanksharesHRBK1.01x3.19x†
M&F BancorpMFBP1.77x3.05x§
IBW FinancialIBWC5.51x2.72x†
United Bancorporation of AlabamaUBAB0.88x2.37x
Ponce Financial GroupPDLB1.12x2.17x
PCB BancorpPCB0.34x1.81x
BankFirst CapitalBFCC0.88x1.79x
Security FederalSFDL1.00x1.74x†
Merchants & Marine BancorpMNMB0.96x1.62x†
Bay Community BancorpCBOBA1.82x1.61x‡
Broadway FinancialBYFC1.87x1.10x

† Illiquid: no shares traded on more than half of measured days. NMBF had no reported volume on 79% of days and HRBK on 66%. ‡ Completed cash acquisition. § Signed acquisition; not closed.

The median changes materially with reasonable data choices:

  • Raw early-anchor median: 2.17x.
  • Use MFBP’s prior-day stale quote instead of its announcement-day close: 1.91x.
  • Exclude all five stocks with no trades on more than half of measured days: 1.99x.
  • Exclude only the two most extreme illiquid outcomes: 1.81x.

None is an investable basket return.

A uniform July 17, 2023 anchor produces a 1.78x median through July 22, 2026, compared with roughly 1.87x for the SPDR S&P Regional Banking ETF including reinvested distributions. The screen does not establish excess return after it was published. Its value is as a reference set for testing breadth and dispersion. Manager letters disclose selected positions and judgments, not full trade ledgers.

The scorecard also understates the range of mechanisms:

  • PDLB grew net loans 81% from year-end 2022 through March 2026, moved from a $30.0 million common loss to $28.7 million of 2025 common earnings and returned a marked 2.17x.
  • PCB and Security Federal returned 1.81x and 1.74x largely through common-book growth, dividends, repurchases and rerating rather than a special ECIP exit.
  • Bay Community completed a cash sale.
  • M&F signed a sale after its standalone earnings route deteriorated.
  • Broadway barely moved despite one of the largest headline subsidies.

The early-anchor marks were favorable. The causes were not uniform, and the screen did not outperform the bank benchmark from its July 2023 publication date.

Treasury Added a New Option—After the Original Thesis

In November 2024, Treasury published its final ECIP disposition policy. A participant could request a sale or repurchase during the first ten years after meeting one of three thresholds:

  • at least 60% deep-impact lending over any four-year period;
  • at least 85% qualified lending over any six-year period; or
  • a 0.5% ECIP rate for six consecutive years.

The policy introduced another route to common value, but later commentary often generalized the 0.5%-of-par price to issuer repurchases. Treasury may sell an instrument to a qualifying mission-aligned nonprofit affiliate for 0.5% of par. An issuer repurchase or other sale occurs at Treasury’s present-value price. Treasury retains discretion and requires regulatory, program and payment compliance.

Most public recipients had not realized this option by July 2026. Several had signed purchase-option agreements or estimated hypothetical discounts. Those estimates were not common equity. The value crystallizes only when Treasury approves a transaction and it closes.

This option should not be backfilled into the original 2022 thesis. It improved the later payoff tree. It did not explain the early spread earnings, buybacks, acquisitions or rerating.

What the Case Teaches

ECIP rewarded investors who separated three questions that looked like one.

What is the capital actually worth?

The face amount was senior to common and therefore could not be added to common book. The economic cost was nevertheless far below face value. The opportunity lived in the gap between those two facts.

This pattern recurs whenever a government, parent company, litigation settlement or strategic counterparty provides capital on non-market terms. The investor must model the legal claim and the economic burden separately.

Can management use it?

The earliest models assumed recipients could double loans or assets. Most could not. Small banks still needed deposits, loan officers, systems, credit discipline and willing borrowers. Excess capital also reduced urgency.

The best outcomes came from management teams with more than one route: safe spread income while waiting, disciplined organic lending, common repurchases, cash acquisitions or a strategic sale. The weakest cases relied on the capital amount itself.

How does the benefit reach each common share?

Bank solvency and common-share value are related but not identical. Permanent capital made every recipient safer. That did not guarantee higher per-share value.

Citizens reduced its share count. BankFirst and Bay Community used M&A. United grew loans and bought stock but retained too much capital. M&F preserved strategic value after operating execution disappointed. Broadway allowed costs and the preferred dividend to consume the advantage.

The right unit of analysis was never ECIP dollars. It was after-preferred value created per common share.

Multiple specialist managers found the anomaly in real time, bought different recipients and documented their reasoning before the outcome. That contemporary record, not the later 13-name screen alone, makes ECIP a genuine mispricing case.

The headline version—Treasury gave tiny banks more capital than their market values—was directionally right and analytically incomplete. Common holders benefited only when management converted cheap senior capital through earnings, repurchases, acquisitions or a sale.

A contractual discontinuity can define the hunting ground. It does not replace the company work.

Evidence and Source Notes

Return method and limitations

  • The 13-name set comes from Stepping Stone’s July 2023 paper. It is a practical public list, not an exhaustive list of every tradeable ECIP-linked security. Applying 2022 anchors to that later list introduces hindsight; the article reports the uniform July 17, 2023 comparison separately.
  • Returns end July 22, 2026 and use Arc’s Yahoo-derived adjusted daily data, issuer disclosures and two secondary price proxies. They are marked security outcomes, not audited manager returns.
  • KRE comparisons use Nasdaq closes and Stock Analysis/S&P Global distributions. Reinvesting each distribution at the contemporaneous close produces approximately 1.35x from October 5, 2022 and 1.87x from July 17, 2023.
  • BFCC uses its September 30, 2022 issuer-disclosed $37.25 close, $3.80 of subsequent dividends and the July 22, 2026 $62.75 quote.
  • CBOBA uses the October 14, 2022 public-thesis price of $8.95, the completed $14 cash acquisition and $0.40 of intervening dividends.
  • The July 2023 uniform comparison uses secondary adjusted monthly prices of $35.51 for BFCC and $7.97 for CBOBA because daily histories were unavailable.
  • MFBP’s scorecard result starts after the original March 2022 disclosure. The earlier $7.10 price to the July 22 announcement close was 6.59x; the signed fixed consideration is 6.56x that anchor.
  • SFDL, HRBK and MNMB had no reported volume on 53%–66% of measured days. NMBF and IBWC had no volume on 79.4% and 89.0%.
  • CBOBA’s acquisition is complete. MFBP’s is signed but not closed; 12.6% of maximum consideration remains contingent.

Program and legal sources

Contemporary investment research

Company-level primary sources