Joining the National Bankers Association MDI network to the SBA 7(a) register on FDIC certificate number — an exact join, not a name match. What it shows about historical mission-lender origination is the strongest argument for EMBI and the hardest problem in underwriting it, at the same time.
How much of this lending do mission-driven banks do today?
Everything below is the evidence for those three sentences. You can read only this box and still follow the argument. Underlined terms open a plain definition — the full glossary has them all.
Name matching is the obvious approach to flagging mission lenders across roughly 6,000 distinct lender strings, and it is the wrong one. For depository institutions the join key already exists in the file, and name matching does not merely cost time — it produces confidently wrong answers.
The 7(a) register carries bankfdicnumber on 88.9% of loans and bankncuanumber on a further 1.2%. The NBA network file carries FDIC certificate numbers for all 43 institutions. That is an exact, reviewable, zero-ambiguity join — and it runs in seconds.
Only the remaining 100,375 loans — non-depository SBLCs and lenders coded to closed or missing institutions — require name matching. That is roughly a thousand names, not six thousand, and they are the ones a human should look at anyway.
Normalising the 4,494 distinct 7(a) lender names and matching against the 43-bank NBA list returned 69 apparent hits. Checked against the certificate numbers, only 28 were real — a precision of 40.6%.
The failure is not random. Mission banks are named after civic virtues, and so are thousands of ordinary community banks: Liberty, Citizens, Unity, Commonwealth, First State, Legacy. The normaliser cannot tell them apart, and the errors land on the largest institutions.
Normalised name matching maps Citizens Bank, National Association — FDIC certificate 57957, 43,692 loans, $2.07B of capital — onto Citizens Savings Bank and Trust Company, certificate 10319, a Black-owned MDI that made nine 7(a) loans in the entire period. Accepting that single match would have overstated mission-lender origination by roughly eightfold and put a fabricated number in front of a rating committee.
Every apparent match between an SBA lender name and an NBA network bank. Coral rows are false positives.
| SBA lender name | SBA cert | Loans | Matched to | MDI cert | Result |
|---|---|---|---|---|---|
| Citizens Bank, National Association | 57957 | 43,692 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| Citizens Bank | 22601 | 772 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| American National Bank | 19300 | 735 | American Bank, National Association | 21567 | false positive |
| American Bank National Association | 23886 | 645 | American Bank, National Association | 21567 | false positive |
| First State Bank | 57910 | 405 | First State Bank | 18301 | false positive |
| American Savings Bank FSB | 32526 | 338 | American Bank, National Association | 21567 | false positive |
| Liberty Bank | 34087 | 292 | Liberty Bank and Trust Company | 20856 | false positive |
| Texas Bank | 21074 | 264 | Texas National Bank | 3337 | false positive |
| The Citizens Bank | 17211 | 122 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| First Security Bank | 17120 | 122 | First Security Bank and Trust Company | 17001 | false positive |
| Liberty National Bank | 57505 | 109 | Liberty Bank and Trust Company | 20856 | false positive |
| American Federal Bank | 29140 | 107 | American Bank, National Association | 21567 | false positive |
| Legacy Bank | 17676 | 102 | Legacy Bank | 59204 | false positive |
| Liberty Bank, FSB | — | 88 | Liberty Bank and Trust Company | 20856 | false positive |
| The First State Bank | 17588 | 68 | First State Bank | 18301 | false positive |
| Citizens Trust Company | — | 63 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| Liberty Savings Bank FSB | 32242 | 50 | Liberty Bank and Trust Company | 20856 | false positive |
| Texas Bank and Trust Company | 17842 | 47 | Texas National Bank | 3337 | false positive |
| American Bank and Trust Company | 20537 | 41 | American Bank, National Association | 21567 | false positive |
| First State Bank and Trust Company Inc | 14185 | 41 | First State Bank | 18301 | false positive |
| American Bank | 34422 | 40 | American Bank, National Association | 21567 | false positive |
| Texas National Bank | 34014 | 40 | Texas National Bank | 3337 | false positive |
| Citizens National Bank | 24085 | 34 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| Citizens Bank and Trust Company | 1972 | 33 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| The Citizens National Bank | 4678 | 32 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| American Savings, Inc. | — | 17 | American Bank, National Association | 21567 | false positive |
| Citizens Savings Bank | 30270 | 12 | Citizens Savings Bank and Trust Company | 10319 | false positive |
| Anchor Bank | — | 9 | Anchor Bank | 57931 | false positive |
43 institutions, joined on certificate number to 1,008,922 7(a) approvals across FY1992–FY2009. Thirty-one appear in the register at all; twelve made no 7(a) loan in eighteen years.
A line for each year from FY1992 to FY2009 showing what share of all dollars came from the 43 network banks, with and without East West Bank.
Look for the peak around FY1999–FY2002 and the steep drop after — that fall shows the network's role in the programme shrinking sharply.
Share of annual approved dollars originated by the 43 NBA-network institutions, with and without East West Bank. The share peaks in FY1999–FY2002 and falls by three-quarters thereafter.
| Institution | Type | Cert | Loans | Capital | Avg | Default rate | States |
|---|---|---|---|---|---|---|---|
| East West Bank | Asian | 31628 | 3,966 | $1.30B | $328,773 | 10.0% | 15 |
| Liberty Bank and Trust Company | Black | 20856 | 249 | $38M | $150,842 | 28.2% | 10 |
| OneUnited Bank | Black | 23966 | 185 | $29M | $157,686 | 26.6% | 3 |
| Citizens Trust Bank | Black | 8033 | 161 | $29M | $178,329 | 21.2% | 2 |
| Industrial Bank | Black | 14679 | 156 | $28M | $177,639 | 21.6% | 6 |
| Carver Federal Savings Bank | Black | 30394 | 95 | $28M | $292,254 | 42.2% | 1 |
| United Bank of Philadelphia | Black | 33568 | 88 | $9.4M | $106,453 | 6.8% | 3 |
| The Harbor Bank of Maryland | Black | 24015 | 87 | $13M | $152,079 | 5.8% | 3 |
| The Citizens Bank of Edmond | Women | 4084 | 63 | $7.0M | $110,571 | 17.5% | 1 |
| Redemption Bank | Black | 21448 | 54 | $9.6M | $178,444 | 16.0% | 1 |
| Unity National Bank of Houston | Black | 26351 | 52 | $2.4M | $45,253 | 50.0% | 1 |
| Falcon International Bank | Hispanic | 26856 | 43 | $7.0M | $162,954 | 26.8% | 1 |
| City First Bank, National Association | Black | 34352 | 33 | $7.3M | $219,697 | 12.0% | 3 |
| Mechanics & Farmers Bank | Black | 12266 | 29 | $4.2M | $144,222 | 8.0% | 1 |
| Sovereign Bank | Native | 25738 | 26 | $7.2M | $277,823 | 23.8% | 1 |
| Texas National Bank | Hispanic | 3337 | 26 | $3.4M | $130,235 | 21.7% | 1 |
| First Independence Bank | Black | 20179 | 19 | $3.0M | $157,474 | 33.3% | 1 |
| OPTUS Bank | Black | 35241 | 17 | $2.6M | $152,274 | 28.6% | 1 |
| Chickasaw Community Bank | Native | 11521 | 16 | $5.8M | $363,250 | — | 2 |
| Citizens Savings Bank and Trust Company | Black | 10319 | 9 | $1.3M | $141,290 | 28.6% | 1 |
Institutions in the network with no 7(a) loan in FY1992–FY2009: Anchor Bank · AllNations Bank · Adelphi Bank · GN Bank · Legacy Bank · Icon Business Bank · Grand Bank for Savings, FSB · Generations Community Bank · Ponce Bank, National Association · Paramount Bank · The Hopeton State Bank · Warsaw Federal Savings and Loan Association. Several are recent charters, so absence here is not evidence of inactivity today — it is evidence that the FY2010-onward file is required to say anything current.
The headline finding — that MDI-originated 7(a) loans outperformed the market — is real, and it is almost entirely one institution. Decomposed, the picture inverts, and the inverted picture is the one EMBI has to underwrite.
| Measure | East West Bank | Other 30 network MDIs | All other lenders |
|---|---|---|---|
| Funded loans | 3,426 | 1,208 | 880,583 |
| Capital originated | $1.12B | $202M | $163B |
| Average loan | $325,707 | $167,381 | $184,576 |
| Median loan | $200,000 | $95,945 | $79,000 |
| Default rate (resolved) | 10.0% | 23.8% | 20.3% |
| Gross loss (% of balance) | 4.9% | 14.0% | 10.7% |
| Jobs per loan | 7.1 | 2.3 | 5.7 |
| Express-family channel share | 20.0% | 47.4% | 57.1% |
East West Bank is a $70-billion institution today. Within this window it wrote loans averaging $325,707 at a 10.0% default rate, using streamlined Express-family channels for only 20.0% of its volume. It is an Asian-American MDI by designation and a large commercial bank by behaviour.
The other thirty network institutions wrote 1,208 loans averaging $167,381, at a 23.8% default rate against 20.3% for all other lenders in the same period. They supported 2.3 jobs per loan against 5.7.
That is what community mission lending looks like in the data: smaller businesses, smaller tickets, higher losses. It is not a failure of the model. It is the reason the federal guarantee exists, and the reason credit enhancement is in EMBI's structure at all.
Two sets of horizontal bars comparing East West Bank, the other 30 network , and all other lenders, on and on use of streamlined loan channels.
Longer bars mean worse outcomes or heavier use of streamlined channels; compare the community MDIs' bars to East West Bank's to see where the risk really sits.
Funded 7(a) loans, FY1992–FY2009, resolved-loan denominator.
Share of each group's loans written through Express, Community Express or LowDoc — the channels that carry the highest losses.
The community MDIs put 47.4% of their volume through Express-family channels against East West's 20.0%. Programme-wide those channels lost roughly twice what conventionally underwritten 7(a) lost. A material share of the community-MDI default gap is therefore a channel choice rather than a borrower-quality difference — and channel is contractible in an origination agreement.
With 1,208 loans across thirty institutions the sample is also thin. This is a directional baseline, not a credit model, and it should not be extrapolated to a $4.7 billion origination programme without the post-2009 data.
Of 28,741 funded 7(a) loans made to Kansas and Missouri borrowers between FY1992 and FY2009, the number originated by an NBA-network MDI is 2.
Additionality is the hardest thing to evidence in impact finance. It usually has to be argued. Here it can simply be counted: the mission-lender origination channel in this region has a measured historical baseline of approximately zero, against a market that moved $4.99B of 7(a) capital over the same period.
There is no displacement story to rebut and no attribution dispute to have. Any capital EMBI originates through mission lenders in Kansas and Missouri is, by construction, new to that channel.
A rating committee will read the same figure and see no in-region track record. There is no regional loss history for mission-lender originations because there are essentially no regional mission-lender originations.
The NBA network's physical presence in the two states is four branches across three institutions — CBW Bank in Weir, Kansas; Liberty Bank and Trust in Kansas City on both sides of the line; Paramount Bank in Hazelwood, Missouri. Origination capacity has to be built or imported, and the plan should say which.
The same fact — a baseline of two loans — is the additionality case to the foundations and the track-record gap to the rating agency. Both readings are correct, and the materials should make both arguments rather than only the first.
Kauffman formation data narrows the surrounding claim. Kansas and Missouri convert entrepreneurial activity into SBA credit at or above the national rate on three independent denominators — 15th and 19th of 51 on new employer businesses, 17th and 27th on new entrepreneurs. The regional constraint sits upstream of credit, at business formation: Kansas generates roughly a third fewer business applications per capita than the nation.
That does not weaken the additionality finding here, which concerns mission-lender origination specifically and stands at 40 loans of 28,741. It does mean the surrounding argument cannot be "capital is being withheld from this region." The defensible version is narrower: a specific origination channel is absent here, and the product it would deliver reaches a narrow slice of the people actually starting businesses.
Before the CDFI list is joined, it is worth stating what it will show, so the result is not misread when it arrives.
Non-depository CDFIs — loan funds, the largest category by count — could not originate 7(a) loans at any scale during the period this file covers. The Community Advantage pilot opened the programme to mission-driven non-depositories in 2011, and SBA lifted the moratorium on new Small Business Lending Company licences in 2023.
A historical CDFI 7(a) share computed on FY1992–FY2009 data therefore measures a rule, not a market. It is close to zero because the programme was closed, and it says nothing about present-day CDFI capacity.
CDFI banks and CDFI credit unions are in this data and are identifiable — by FDIC certificate and NCUA charter respectively. Self-Help Credit Union, a certified CDFI, appears with 509 loans. Those are the joins worth running on this file.
The 100,375 loans carrying neither an FDIC certificate nor an NCUA number. These are the only names requiring hand verification, and this is the list to work down.
| Lender | Loans | Capital | Default |
|---|---|---|---|
| Readycap Lending, LLC | 18,427 | $8.95B | 32.0% |
| Business Loan Center, LLC | 8,948 | $3.99B | 37.4% |
| COFSB, National Association | 8,065 | $347M | 10.3% |
| VelocitySBA, LLC | 7,070 | $69M | 38.5% |
| MISSINGMAINBANKID | 6,684 | $1.19B | 3.1% |
| GE Capital Small Business Finance Corporation | 5,950 | $2.71B | 21.0% |
| Federal Deposit Insurance Corporation | 5,473 | $1.00B | 11.5% |
| Loans from Old Closed Lenders | 3,908 | $1.88B | 5.6% |
| Wachovia SBA Lending, Inc. | 3,052 | $1.84B | 1.0% |
| Newtek Small Business Finance, Inc. | 2,369 | $846M | 29.0% |
| Business Lenders, LLC | 2,083 | $788M | 26.4% |
| Citibank (West), FSB | 1,235 | $211M | 1.0% |
| AmSouth Bank | 1,005 | $197M | 7.3% |
| First Western SBLC, Inc | 961 | $447M | 12.3% |
| New York Business Development Corporation | 894 | $285M | 19.5% |
All 4,494 distinct 7(a) lender names with FDIC certificate, NCUA number, loan volume, capital, charge-off rate, states served, Kansas–Missouri volume and years active — plus empty, highlighted columns for CDFI certification, CDFI type, FDIC MDI status and verification method. The 269 certified development companies and 4,606 504 third-party lenders are on their own sheets in the same shape.
Joining the FDIC MDI list is a single operation on the certificate column. Joining the CDFI list splits: depositories on cert, loan funds by hand against the non-depository sheet. The name-match evidence sheet is included so the method decision is documented rather than repeated.
FY2020–FY2026 records close the structural argument in Section 04. Community Advantage opened 7(a) to mission-driven non-depositories in 2011 — after the legacy file ends. It is measurable for the first time, and what it measures is not encouraging.
| Channel | Loans | Capital | Avg | In KS/MO | Resolved | C/O |
|---|---|---|---|---|---|---|
| Community Advantage | 2,240 | $355M | $158,357 | 52 | 36% | 28.3% |
| NBA-network MDI banks | 1,568 | $1.33B | $851,095 | 11 | — | — |
Largest Community Advantage originator: CDC Small Business Finance Corp., 747 loans. 86 lenders participated nationally. Charge-off rates here sit on a book only 36% resolved and will rise.
The baseline moved, barely. Across FY1992–FY2009 the region recorded 40 mission-lender 7(a) loans in eighteen years. Across FY2020–FY2026 it recorded 63 in seven — 52 Community Advantage and 11 from NBA-network MDI banks, against 8,056 regional loans in total.
Nationally the NBA network's share is essentially unmoved: 0.78% of capital now against 0.82% in the legacy era. Twenty-seven of the 43 institutions appear, up from 31 over a much longer window — the network is more active per year, not larger in share.
The additionality argument survives intact. Mission lenders remain a rounding error in Kansas and Missouri under both regimes. There is still no incumbent to displace.
The credit argument got harder. Community Advantage charges off at 28.3% on the 36% of its book that has resolved — the worst of any modern channel, on data too young to have surfaced most of its losses.
Community Advantage is the closest existing analogue to EMBI's stated model: mission-driven lenders, smaller tickets, underserved borrowers, federal guarantee. Its average loan is $158,357 and its early loss rate is roughly four to five times the Preferred Lenders channel.
That is not an argument against the model. It is an argument that the credit enhancement stack has to be sized against Community Advantage experience rather than against programme-wide averages — and that the origination agreements should specify documentation and underwriting standards rather than inherit the channel's defaults.
Five consequences for the research period.
For banks and credit unions it is a join on an identifier that already exists in both files — executed, and in the workbook. What remains is hand verification of roughly a thousand non-depository lender names covering 100,375 loans, and that work should be done by a person reading names, not by a fuzzy matcher.
The name-match evidence sheet is included so the method decision is documented. A 40.6% precision rate is not a tuning problem.
With a regional baseline of 2 loans out of 28,741, EMBI is not improving a mission-lending channel in Kansas and Missouri — it is creating one. That is a stronger and more defensible claim, and it is countable rather than modelled.
It also means the initiative should not promise to shift a percentage of existing mission-lender volume. There is no meaningful denominator to shift.
Aggregate network default of 13.6% against 20.3% for everyone else is arithmetically true and substantively misleading. Remove one institution and it inverts to 23.8%. Anyone who re-runs the analysis will find it.
The defensible version: mission lenders using conventional underwriting channels performed in line with or better than the market, and the loss differential in the community-MDI segment tracks channel mix and borrower size — which is what the guarantee and the credit enhancement are for.
The community MDIs ran 47.4% of volume through Express-family channels against East West's 20.0%, and those channels carry roughly double the programme-wide loss. This is the one driver in the whole dataset that is both large and directly contractible.
A covenant capping streamlined-channel share, or requiring Certified Lender Program-equivalent documentation, is a cheap and evidence-backed credit enhancement that costs nothing in mission terms.
Non-depository CDFIs were largely excluded from 7(a) until Community Advantage in 2011 and the 2023 SBLC licence expansion. Any CDFI share computed on this file will be near zero for that reason, and reporting it without the explanation would understate present capacity badly.
Run the join on CDFI banks and credit unions, where the certificate and charter keys make it exact — and treat everything else as a question the FY2010-onward file has to answer.
This is document 4 of 5 in the reading path. Back to the Research index.