Fortieth in a series on jobs whose pay system is stranger than the salary. It sits beside two earlier tolls on bonds: the credit rating agency that the issuer pays to grade its own bond, and the mortgage bond seller who skims a spread each time a pooled loan changes hands. The municipal bond dealer works the last crossing, between a town's bond and the household that buys it, and it collects inside the price, where most buyers never look. Every figure below is cited to a Municipal Securities Rulemaking Board rule or research paper, a Federal Reserve bank, FINRA, the SEC, or a fund prospectus filed with the SEC. The MSRB's own site would not open for me, so I read its documents from dated Internet Archive copies and link those copies. Where a number comes from trade press rather than a primary source, I say so.
Buy a $10,000 bond from a town's water system through a brokerage account and the confirmation may show no commission at all. That does not mean the trade was free. The MSRB's investor guide says the mark-up "is not an additional itemized fee... but instead is included in the price." Its worked example is as plain as it gets: pay $10,000 for a bond whose prevailing market price was $9,900, and you paid a $100 mark-up. One percent, folded into a number that looks like a price.
How big that slice is depends on something most people never think of as part of a price: how much you buy. In 2025, by the MSRB economists' measure, trades of $100,000 or less, which the MSRB calls odd lots, cost an average of 50.4 basis points, against 21.1 basis points for trades of $1 million or more. In the paper's dealer-type tables, where all 2025 odd lots averaged 59.5 basis points, a small odd-lot trade done the classic retail brokerage way cost about 1.7 times the average for all odd lots (my arithmetic from the paper's tables). Households are the largest holders of municipal bonds, as the hospital captives piece laid out, and they pay the highest rate to get in.
Principal, agent, and where the money sits
A dealer can handle your trade two ways. As an agent, it finds the other side and charges a commission, and the MSRB guide says commissions must be disclosed on the trade confirmation. As a principal, it sells you a bond from its own inventory or buys yours into it, and it is paid by the gap between the price you got and the market price at the time. On a purchase that gap is a mark-up. On a sale it is a mark-down.
The guide adds a detail worth keeping. If one dealer buys a bond from one customer and sells it to another a few minutes later, the trade record can show a mark-down to the seller and a mark-up to the buyer. Neither customer pays both. Each pays one side.
The rule over all of it is MSRB Rule G-30: the price, mark-up included, must be "fair and reasonable." The rule also says that keeping the dealer's profit reasonable does not excuse an unfair price.
What the gap costs, by size and by kind of dealer
The MSRB's research staff measure trading cost as an effective spread: the gap between the same-day volume-weighted price customers paid to buy a bond and the price customers got to sell it, as a percent of the midpoint. That is a round trip. A buyer who holds to maturity pays only the buy side, which is roughly half. Each paper measures a little differently, so I am not stitching their numbers into one series. I give each one on its own terms.
The newest and best is the MSRB's April 2026 paper on odd-lot trades (Wayback copy, May 10, 2026). Its history of the average effective spread runs from 173 basis points in October 2009 down to a low of 42 in late 2021, up to a peak of 63 in May 2022, and back to about 50 in late 2025. Its Table 2 splits the trades by size, in basis points:
Scroll to compare columns and rows when the table extends beyond the box.
| Year | Trades of $100,000 or less | Trades of $1 million or more | Ratio |
|---|---|---|---|
| 2019 | 72.3 | 17.4 | 4.2 |
| 2020 | 78.0 | 26.8 | 2.9 |
| 2021 | 50.0 | 22.9 | 2.2 |
| 2022 | 61.2 | 17.9 | 3.4 |
| 2023 | 58.4 | 17.8 | 3.3 |
| 2024 | 51.9 | 17.1 | 3.0 |
| 2025 | 50.4 | 21.1 | 2.4 |
In 2025 the small trade cost 2.4 times the block. The ratio has narrowed since 2019, and in 2025 part of the narrowing came from the block side getting more expensive, not only the small side getting cheaper. The small buyer still pays more than twice the rate.
The same paper then splits the 2025 odd lots by the kind of dealer on the other side, and the gap gets wider. At dealers that trade almost entirely with institutions, odd lots averaged 32.7 basis points, with a median of 5.6. At all other dealers they averaged 88.1, with a median of 41.7. The paper's Table 5 isolates the classic retail brokerage trade: an odd lot at one of those other dealers, outside a fee-based account. It averaged 101.9 basis points, with a median of 52.9. Odd lots in fee-based accounts at institutional dealers averaged 17.1. NTBC stands for non-transaction-based compensation: a trade with no mark-up, mark-down or commission, which the paper says is much more likely to come from managed accounts that charge a fee on assets instead. The paper says sales credit paid to advisers explains part of the gap. In other words, some of the toll on a retail muni trade is the salesperson's pay, and it rides in the price.
The other papers point the same way:
- MSRB, August 2023 (Wayback copy, January 16, 2026). As of March 2023, trades under $100,000 carried an effective spread three times that of trades over $1 million. The paper points to falling prices plus the IRS de minimis rule as the likely cause. A footnote describes how dealers price: they "typically quote a fixed amount of spread (e.g., 25 basis points when expressed in yield)" and use grids for premium, discount and zero-coupon bonds. Trades flagged NTBC ran about 30 basis points cheaper.
- MSRB, March 2025 (Brookings-hosted copy; used because msrb.org refused direct reads and no archived MSRB copy was found). From January 2023 to June 2024, munis averaged 53 basis points, corporate bonds 36 and agency bonds 40. The gap between small and large trades was 38 basis points for munis, 25 for corporates and 32 for agencies. Munis cost the most of the three.
- Federal Reserve Bank of Chicago, Chicago Fed Letter No. 530, October 2026 (chicagofed.org). The bank's economists traced chains of trades through the regulatory record and measured round-trip markups directly. Those fell from about 200 basis points in the Great Recession to about 60 before COVID, and the smallest, retail-sized trades fell from about 230 to 70 to 80. Markups rose in early 2022 and have fallen since 2024. Trades that came out of a dealer's inventory cost 75 to 100 basis points more than prearranged trades. The letter also says effective spreads may understate what investors pay.
Why is this market so expensive to trade in? Part of the answer is thinness. The odd-lot paper counts about 26,700 securities trading on an average day, out of roughly 1 million outstanding, and only about 6,900 of them have both a customer buy and a customer sell that day. Most bonds do not have a fresh price you can look up. Meanwhile, the market is busy in total: the MSRB's 2025 Year in Review counts 17.6 million trades in 2025, up 22 percent and the fourth record year in a row. A search summary of a separate MSRB paper, which I did not read at the source, says 87 percent of 2025 trades were $100,000 or less. Treat that share as secondary. The direction it points is the same as everything above: this is a market of small trades, and small trades pay the most.
What the confirmation has to show, and what it can leave off
Since May 14, 2018, MSRB Rule G-15 has required the dealer to print the mark-up or mark-down on your confirmation, "as a total dollar amount and as a percentage of the prevailing market price." The SEC approved the rule on November 17, 2016 (81 FR 84637). FINRA Rule 2232 applies the same test to corporate and agency bonds (finra.org).
The catch is in the trigger. The disclosure applies only when two things are both true:
- The dealer traded as principal with a customer who is not an institution.
- The dealer made one or more offsetting trades totaling at least your size on the same trading day. It bought the bond the day it sold it to you, or sold it the day it bought it from you. Trades with an affiliate that are not at arm's length must be looked through.
So the printed number covers the riskless-principal case, where the dealer was in and out the same day. A bond that sat in inventory for a week and then went to you does not trigger it. The rule also exempts a functionally separate trading desk, new-issue sales at the list offering price, and municipal fund securities such as 529 plans. The MSRB guide on mark-ups says it in one line: a missing disclosure does not mean no mark-up was paid.
Two items appear on confirmations to non-institutional customers for most municipal securities (municipal fund securities are excluded): the time of execution, and a reference to the bond's page on EMMA, the MSRB's public price site, which must be a hyperlink on electronic confirmations. Those two items are what let you check the trade yourself, which I walk through below.
Did the rule lower costs? The MSRB's July 2019 study (Wayback copy, September 12, 2025) compared the year before and the year after. Trades of $100,000 par or less went from 86.3 to 77.5 basis points, down 8.8 basis points, or 10.2 percent. Trades over $100,000 went from 38.4 to 29.3. The paper found "no discernible impact" from the rule beyond the downward trend already under way. FINRA's 2019 exam findings, which I read only as a search summary, said some firms left firm compensation out of the disclosed mark-up or listed sales credits as separate line items. I found no FINRA retrospective review of the rule.
How the dealer decides what the market price was
The disclosed mark-up is only as good as the "prevailing market price" it is measured from, and the dealer computes that number. Rule G-30, Supplementary Material .06 sets the method:
- The starting presumption is the dealer's own contemporaneous cost, or its contemporaneous proceeds for a mark-down.
- The dealer can move off that presumption in instances such as a change in interest rates, a change in the bond's credit quality, or news that came out after its trade, and it may use other evidence when it had no same-day trade.
- If it does, it must work down a list in order: contemporaneous inter-dealer trades, then the dealer's trades with institutional accounts, then inter-dealer quotations for actively traded bonds. After those come trades in "similar" securities, and last, economic models.
- Isolated trades get "little or no weight."
The MSRB's fair-price guide gives the reader's version: subtract the disclosed mark-up from the price you paid, and what is left is the prevailing market price the dealer used.
The new-issue toll: the underwriting spread
A new bond pays a toll before any household sees it. The underwriter buys the issue from the town or school district and resells it, and its pay is the gross spread. The MSRB and the Government Finance Officers Association do not publish a series on it that I could find. The figures I have come from LSEG data reported by The Bond Buyer (article), which I did not read in full, so treat them as secondary:
- The average gross spread was $4.30 per $1,000 of bonds in 2025, up from $4.23 in 2024. That is 0.43 percent of the issue (my arithmetic).
- Negotiated deals averaged $4.64, up from $4.47. Competitive deals averaged $1.78, down from $2.69.
- About 80 percent of deals in the first half of 2025 were negotiated.
The negotiated spread is about 2.6 times the competitive one (my arithmetic). Measured by spread alone, most deals went the more expensive route.
That spread is not shown to the retail buyer as a mark-up. New-issue sales at the list offering price are exempt from G-15 disclosure, and the takedown, or concession, the selling dealer earns is not printed as a mark-up. Rule G-11 lets an issuer run a retail order period, in which orders that meet the issuer's retail rules are taken first or exclusively. A dealer submitting retail orders must state in writing, by the time of formal award, whether each order meets the issuer's criteria and is conditionally committed, and whether it holds more than one order from the same customer for the same bond.
The issuer has one protection the household does not. Under the Dodd-Frank Act, a municipal advisor owes a statutory fiduciary duty to the government it advises; the SEC's rule implementing the registration regime took effect January 13, 2014 (Release 34-70462). An underwriter acting as underwriter is excluded. The dealer across the table from a household owes "fair and reasonable." The advisor across the table from a town owes a fiduciary duty. The river this toll sits on is large: the MSRB's Municipal Market Facts, January 2026 edition, puts 2025 issuance at a record $582 billion from about 50,000 issuers.
How an ordinary person gets into this business
Selling municipal bonds to the public takes a FINRA-administered exam, and running the desk takes another. The fees are small. The gate is the firm, not the test.
Scroll to compare columns and rows when the table extends beyond the box.
| Exam | Role | Format | Passing score | Corequisite | Fee |
|---|---|---|---|---|---|
| Series 52 | Municipal Securities Representative | 75 scored questions, 2 hours 30 minutes | 70% | SIE | $260 |
| Series 53 | Municipal Securities Principal | 100 questions, 3 hours 15 minutes | 70% | SIE, plus what FINRA lists as the Series 52 top-off (S52TO) | $265 |
| Series 50 | Municipal Advisor Representative | 100 questions, 3 hours | 71% | none | $265 |
| Series 54 | Municipal Advisor Principal | 100 questions, 3 hours | 70% | Series 50 | $265 |
To be the firm rather than work for one, the path under MSRB Rule A-12 runs in this order:
- Register with and be approved by the SEC.
- For a dealer, notify its registered securities association, FINRA, or for a bank dealer, its bank regulator.
- File Form A-12 with the MSRB and pay the $1,000 initial registration fee, then a $1,000 annual fee due by October 31.
The advisor side has its own path. A municipal advisor registers with the SEC on Form MA for the firm and Form MA-I for each individual. Under MSRB Rule A-11, the firm then pays a fee for each covered professional, meaning each qualified representative with an active Form MA-I: $1,130 in 2026, $1,200 in 2027, $1,270 in 2028 and $1,340 in 2029.
I did not find the FINRA new-member application fees, the dealer-side MSRB Rule A-13 rate amounts or an SEC filing fee for Form MA, so I give none. The exam and MSRB fees above run from a few hundred dollars to the low thousands. The larger cost of entry is building a firm the SEC and FINRA will approve, and I did not find a price for that.
How a retail buyer checks a price on EMMA
The MSRB guide "Did I Get a Fair Price?" (Wayback copy, February 25, 2024) lays out the check:
- Find the nine-digit CUSIP on your confirmation or statement.
- Type it into the EMMA Quick Search box to reach the bond's Security Details page, and open the trade activity.
- Read the trade table: trade date and time, settlement date, price, yield, trade amount, and trade type, which is customer bought, customer sold or inter-dealer.
- Compare your price and time with the trades around it, especially inter-dealer trades, which are the closest thing to a wholesale price.
Dealers must report trades within 15 minutes of the time of trade under Rule G-14, with an end-of-day exception for list-offering-price and takedown trades, so the trades on either side of yours are usually there.
The guide's own example is a New York State general obligation refunding bond, CUSIP 649791AM2, on June 30, 2016. At 11:34 a customer sold $50,000 at a price of 112.703, and an inter-dealer trade printed at the same price. About 15 minutes later a customer bought $50,000 of the same bond at 113.372. The difference is 0.669 points of price, which on $50,000 of par is about $334.50 (my arithmetic). The guide is careful here, and so should you be: the data are anonymous, so the second trade may have been a different dealer selling from its own inventory. It shows what the record can suggest, not who was paid what.
A fund instead of a bond
The other route a household has is a fund. The MSRB's "Ways to Buy Municipal Bonds" (Wayback copy, July 22, 2026) says individual bonds usually come in $5,000 minimums, fund minimums run $500 to $5,000, and exchange-traded funds trade at a market price around their net asset value. Two national muni ETFs publish their costs in prospectuses filed with the SEC:
Scroll to compare columns and rows when the table extends beyond the box.
| Fund | Total annual operating expenses | Turnover | Source |
|---|---|---|---|
| iShares National Muni Bond ETF (MUB) | 0.05% | 11% | 497K, June 29, 2026 |
| Vanguard Tax-Exempt Bond ETF (VTEB) | 0.03% (0.02% management fee, 0.01% other) | 23% | 497K, February 27, 2026 |
An expense ratio is charged every year. A dealer spread is charged once per trade. To put them in the same unit, my arithmetic, not a claim about any one trade: one round trip at the 2025 odd-lot average of 50.4 basis points equals about 17 years of VTEB's 0.03 percent, and the 101.9 basis point retail brokerage average equals about 34 years. A buyer who holds a bond to maturity pays roughly half the round trip, so halve those figures for a one-way purchase. The funds also trade bonds themselves, at the turnover rates above. This is a comparison of units, not a recommendation of any fund or of funds over bonds.
What a salaried reader should take from this
A fee you cannot see is still a fee. The muni mark-up is built into the price, and the rule that makes a dealer print it covers only the same-day case. A confirmation with no mark-up on it may simply be a trade the rule does not reach. The MSRB says so in its own guide.
In this market, size is price. The smallest trades paid 2.4 times what million-dollar blocks paid in 2025, and the Chicago Fed's chain-by-chain measure has told the same story since the Great Recession. The bonds are the same. The buyer is smaller.
The kind of firm on the other side matters as much as the bond. An odd lot cost 32.7 basis points on average at dealers that trade almost entirely with institutions, 88.1 at the rest, and 101.9 for the classic retail brokerage trade. The MSRB's economists say adviser sales credit explains part of that. When a salesperson is paid out of the price, the price carries the salesperson.
A public price record is only useful if someone reads it. Dealers report trades within 15 minutes, and EMMA shows them with time, price and size. The confirmation links to it. The whole check takes a CUSIP and a few minutes, and it is the one tool in this market built for the household rather than the dealer.
Related reading
- The Hospitals' Own Insurers in Cayman, and the Cycle That Feeds Them: who holds municipal bonds, why the interest is tax-free, and the tax-equivalent-yield math, from the issuer's side.
- How credit rating agencies are paid: the issuer pays the grader, and the grade on a municipal bond is one more fee in the deal.
- How mortgage bond sellers are paid: the dealer spread on a different kind of bond, one layer above the mortgage.
- Who gets paid on your mortgage: the companion piece, every toll on one $400,000 home loan.
- Hospital charities and their tax-free portfolios: tax-exempt hospital bonds as a subsidy, seen from the public side.
- How financial planners are paid: the percentage fee that often sits on top of the account holding these bonds.
- The Toll Booth: the full index: where this piece sits among the other tolls.
Fact-check notes and sources
The disclosure and pricing rules come from MSRB rules, the SEC's approval in the Federal Register, and FINRA's rulebook. The cost figures come from MSRB research papers and a Federal Reserve Bank of Chicago letter. The fund costs come from prospectuses filed with the SEC. The exam facts come from FINRA. The MSRB's site would not open for me on November 22, 2026, so every MSRB document was read from an Internet Archive copy, and each link below names that copy's date.
- The mark-up is "included in the price," a missing disclosure does not mean no mark-up, and the $10,000 price minus $9,900 prevailing market price equals $100 example: MSRB investor guide "What Is Mark-up?", 2018.1 edition, Wayback copy of December 12, 2024.
- Agency commissions disclosed on the confirmation, one side per customer in a chain, subtracting the disclosed mark-up to find the prevailing market price, the EMMA steps and the CUSIP 649791AM2 example (June 30, 2016; customer sale and inter-dealer trade at 112.703 at 11:34; customer purchase of $50,000 at 113.372 about 15 minutes later; the anonymity caveat): MSRB, "Did I Get a Fair Price?", 2017.1 edition, Wayback copy of February 25, 2024. The 0.669-point difference and the $334.50 figure are my arithmetic; they are not a dealer's disclosed mark-up.
- Rule G-15(a)(i)(F) (dollar and percentage disclosure, the principal, non-institutional and same-day offsetting-trade conditions, the look-through for affiliates, the functionally separate desk, list-offering-price and municipal fund securities exceptions, and the time-of-execution and EMMA link on every non-institutional confirmation): Wayback copy of August 12, 2026. Effective May 14, 2018. SEC approval November 17, 2016, Release No. 34-79347, 81 FR 84637. FINRA Rule 2232(c) and (d) for corporate and agency debt: finra.org, approved as Release 34-79346, 81 FR 84659.
- Rule G-30, Supplementary Material .06 (the contemporaneous-cost presumption, the three grounds to overcome it, the ordered list of evidence, "similar" securities and economic models, "little or no weight" for isolated trades) and G-30's "fair and reasonable" standard: Wayback copy of July 13, 2026.
- 2025 costs by trade size and dealer type (173 bp in October 2009, 42 bp low in late 2021, 63 bp peak in May 2022, about 50 bp in late 2025; Table 2; Table 4's 32.7 bp and 88.1 bp averages with medians of 5.6 and 41.7; Table 5's 101.9 bp average and 52.9 median for non-NTBC odd lots at other dealers and 17.1 bp for NTBC odd lots at institutional dealers; the sales-credit explanation; the 26,700, 1 million and 6,900 counts): MSRB, "Transaction Costs for Municipal Securities Odd-Lot Trades", April 2026, Wayback copy of May 10, 2026. The 2.4x ratio and the full Table 2 are the paper's. The paper's text says both the average and the median at other dealers are "nearly 2.7 times" those at institutional dealers; the averages bear that out (88.1 against 32.7), but the table medians (41.7 against 5.6) work out to about 7.4x by my arithmetic, so I give the medians and no median ratio. The definition of NTBC and the managed-account explanation are from the paper's footnotes 23 and 31.
- The 2019 before-and-after figures (86.3 to 77.5 bp and 38.4 to 29.3 bp; "no discernible impact"): MSRB, "Mark-up Disclosure and Trading in the Municipal Bond Market", July 2019, Wayback copy of September 12, 2025. FINRA's 2019 exam findings on mark-up disclosure are from a search summary of the FINRA 2019 report and are secondary.
- The March 2023 three-times gap, the de minimis explanation, the "25 basis points" quoting footnote, and the NTBC discount of about 30 bp: MSRB, August 2023, Wayback copy of January 16, 2026.
- Munis 53 bp, corporates 36 bp, agencies 40 bp, and the size gaps of 38, 25 and 32 bp (January 2023 to June 2024): MSRB research paper, March 2025, Brookings-hosted copy.
- Round-trip markups on trade chains (about 200 bp to about 60 bp; retail-sized trades about 230 bp to 70 to 80 bp; inventory trades 75 to 100 bp more than prearranged trades; effective spreads understate costs): Federal Reserve Bank of Chicago, Chicago Fed Letter No. 530, October 2026, written by bank economists.
- 17.6 million trades in 2025, up 22 percent, the fourth record year: MSRB 2025 Year in Review, Wayback copy of September 25, 2026. The 87 percent share of trades at $100,000 or less is from a search summary of a separate MSRB paper and is secondary.
- $582 billion of 2025 issuance and about 50,000 issuers: MSRB Municipal Market Facts, January 2026 edition, Wayback copy of October 9, 2026 (the file path says 2022-09; the content is the January 2026 edition).
- 15-minute trade reporting and the end-of-day exception: Rule G-14 RTRS Procedures, Wayback copy of August 12, 2026.
- Underwriting spreads ($4.30 per $1,000 in 2025 against $4.23 in 2024; negotiated $4.64 from $4.47; competitive $1.78 from $2.69; about 80 percent negotiated in the first half of 2025): LSEG data as reported by The Bond Buyer. Secondary; I did not read the full article or the LSEG data. The 0.43 percent and 2.6x figures are my arithmetic. No MSRB or GFOA series on spreads was found.
- Retail order periods and the G-11(k) written statements: Rule G-11, Wayback copy of May 21, 2026.
- Municipal advisor fiduciary duty, the underwriter exclusion, the January 13, 2014 effective date, and Forms MA and MA-I: SEC Release 34-70462.
- Exam formats, passing scores, corequisites and fees: FINRA pages for Series 52, Series 53, Series 50 and Series 54, read November 22, 2026.
- Dealer and advisor registration order and the $1,000 initial and annual fees: Rule A-12, Wayback copy of August 12, 2026. Per-professional advisor fees of $1,130 to $1,340 for 2026 to 2029: Rule A-11, Wayback copy of August 13, 2026.
- Fund minimums and how ETFs trade: MSRB, "Ways to Buy Municipal Bonds", February 2026, Wayback copy of July 22, 2026. MUB expenses and turnover: 497K of June 29, 2026. VTEB expenses and turnover: 497K of February 27, 2026. The 17-year and 34-year comparisons are my arithmetic and describe units, not any trade.
- Not asserted, and why: a federal wage figure for municipal securities representatives (I did not pull one, so this piece has no wage section); the dollar total of municipal bonds outstanding (the Federal Reserve table I tried was not available); any FINRA retrospective review of the 2018 rule (none found); the FINRA new-member fees, MSRB Rule A-13 dealer fee amounts and any SEC Form MA fee (not found or not verified); MSRB Rule G-42 on advisor duties (not read); and the status of a proposed one-minute trade-reporting standard, which I saw only in a search summary.
This post is informational and journalistic, not investment, tax, legal, or career advice, and nothing here is a recommendation to buy or sell any bond or fund. It describes MSRB rules and research, a Federal Reserve bank's research letter, FINRA materials, SEC releases and fund prospectuses. Costs, fees and rules change, several MSRB documents were read from archived copies, and some figures are secondary as noted, so verify current data before relying on any of them. Mentions of specific companies, funds, agencies and publications are nominative fair use, and no affiliation is implied.