A money market sweep and a Treasury bill are not the same decision, and the difference only shows up on the worst day

For most of last year I told founders that a government money market sweep and a stack of Treasury bills were the same decision with different paperwork. Both hold short government paper, both pay roughly the same rate, both keep the money out of a single bank above the 250 thousand dollar line, and both can be set up in an afternoon by somebody who has never run a treasury function, which is exactly why they get treated as one decision. I was wrong about the first part, and it was bad advice for any company that parks payroll cash in the wrong kind of fund.

The question of treasury bills vs money market funds for startup cash turns on 3 documents that nobody in a seed round reads: an SEC press release from July 2023, the TreasuryDirect page for bills, and the daily rate table the Treasury publishes every business day. I went looking for all 3 after a finance lead asked me why her fund prospectus mentioned a fee on redemptions, and it took me most of an evening to answer her properly. The short answer I gave her is the one I would give you. Keep the next 30 days of payroll cash out of prime funds entirely.

The SEC rule that split money market funds into two kinds

On 12 July 2023 the SEC adopted amendments to the rules governing money market funds under the Investment Company Act of 1940, and the press release announcing them does two things that any treasury policy written after that date has to know about, even though neither of them shows up on a monthly fund statement. The first one is good news.

It removed the gates, and I would count that as the best change in the release. The release says the amendments “remove provisions in the current rule that permit a money market fund to suspend redemptions temporarily through a gate”. A fund can no longer freeze your money for a period because markets are stressed, which was the nightmare scenario people pointed at after 2008.

The second thing is the one that changed my advice. I stopped recommending prime sweeps that week, after 11 months of doing it. The amendments “require institutional prime and institutional tax-exempt money market funds to impose liquidity fees when a fund experiences daily net redemptions that exceed 5 percent of net assets, unless the fund’s liquidity costs are de minimis.” And on top of that, any “non-government money market fund” must impose a discretionary liquidity fee “if the board determines that a fee is in the best interest of the fund.”

Read those two sentences against a bad week. I did, twice, and I did not like the result. A prime fund can charge you to leave on exactly the day everyone else is leaving, because the 5 per cent trigger is a measure of other people panicking. A government fund is outside both the mandatory fee and the discretionary one. The yield difference between the two is usually small. The difference in what happens on the worst day is not small at all, and it is invisible on the statement until that day arrives.

What a Treasury bill actually is, in the Treasury’s own words

Bills are the simplest instrument in the whole comparison, and I still find that oddly reassuring. TreasuryDirect sells them for terms of 4, 6, 8, 13, 17, 26 and 52 weeks. They are “sold at a discount or at par (face value). When the bill matures, you are paid its face value.” The interest is simply the gap between the price you paid and the face value you get back.

The minimum purchase is 100 dollars, in increments of 100 dollars. The 52 week bill is auctioned every four weeks and the shorter terms weekly, so a company can buy into a fresh auction almost any week of the year.

The line I skipped for 2 years is the tax row. I am slightly annoyed with myself about that. It says “Federal tax due on interest earned” and then “No state or local taxes”. For a Delaware corporation operating out of California or New York, that row is worth more than the rate difference between any two funds you are likely to compare, and I have not seen a single sweep product marketed with it.

The curve on 15 September 2026

The Treasury publishes daily bill rates. I prefer them to any bank rate card. On 15 September 2026 the coupon equivalent yields were 3.85 per cent for 4 weeks, 3.92 for 6 weeks, 3.96 for 8, 4.07 for 13, 4.12 for 17, 4.21 for 26 and 4.36 per cent for 52 weeks. The bank discount rates sit lower, from 3.79 at 4 weeks to 4.17 at 52, because the two measures are calculated differently and quote tables mix them freely.

So the curve slopes upward. A year of lock-up pays 51 basis points more than a month on the coupon equivalent measure. On 2 million dollars that is roughly 10,200 dollars a year before tax, which is my arithmetic rather than a Treasury figure, and it is the entire reward for giving up the ability to sell at par whenever you like.

Where each one actually fails

A government money market fund fails on nothing dramatic. I would still hold one for cash needed beyond 6 months. It fails on being a fund: the statement shows shares, the yield floats with every rate decision, and the tax treatment of its distributions depends on what the fund holds, which varies by fund and has no general answer worth printing.

A prime fund fails on the day that matters. I would not hold payroll cash in one. It usually yields a little more because it holds commercial paper and bank obligations rather than only government paper, and since the 2023 rule it is the kind of fund that can charge you a liquidity fee when redemptions spike. For money you might need in a hurry, that trade is backwards.

A Treasury bill fails on flexibility. Hold it to maturity and you get face value on a known date, which is as close to certain as dollar assets come. Sell it early and you take whatever the market pays that morning. So a bill ladder only works if the rungs line up with dates you actually know in advance, like payroll, a quarterly tax payment or the last month of runway, and a ladder built around dates you merely hope for is a bet on rates dressed up as a cash plan.

Where the money sits while it waits

This is where I told founders the wrong thing for longest. Neither bills nor fund shares are bank deposits, so the FDIC line from the insurance brochure does not apply to them at all. What applies instead, when they sit at a broker, is SIPC. Its own page puts the limit of protection at 500,000 dollars, “which includes a $250,000 limit for cash”, and it protects cash only when it is in the account “for the purpose of investing in securities”.

That protection is about a broker failing and losing track of your assets. It is not a guarantee on price, and I had assumed for longer than I like that the two were the same promise. They are 2 separate promises. A bill held in custody at a failed broker comes back to you as a bill; a fund share that has charged a liquidity fee on the way out does not come back at all.

The split I use now

Operating cash stays in the bank. It is sized to one payroll run plus a month of payables, the same rule as for deposit concentration, and I keep it there even when the rate on it is close to zero, because the job of that money is to move on time rather than to earn anything. The next 6 months of known outflows go into bills laddered to the dates of those outflows, bought at auction, held to maturity. Anything beyond that sits in a government money market fund, never a prime one, because the whole point of that layer is being able to leave on a bad day without paying for the privilege.

Here is how I would lay it out on paper. Take a company with 4.8 million dollars and a monthly burn of 300 thousand. Operating cash of 450 thousand stays at the bank. Six rungs of 300 thousand go into bills maturing monthly, 1.8 million in total. The remaining 2.55 million sits in a government fund and gets moved into the ladder one rung at a time as each bill matures. The whole arrangement is dull on purpose, because a treasury setup that needs a meeting every month is one that nobody will keep running by the second quarter, and dull is the design goal.

Questions we get

These come up in almost every conversation about idle cash. I have answered each of them badly at least once, which is why the answers below are shorter than they used to be.

How do you build a t bill ladder for startups without a treasury team? Pick the dates you already know, such as payroll, quarterly taxes and the last months of runway, and buy one bill maturing a few days before each. Weekly auctions for terms up to 26 weeks mean you can add a rung almost any week, and the 100 dollar minimum means no rung is too small.

Is a government money market sweep safe enough for payroll cash? Yes, with one check. It is safer than a prime sweep on the one axis that matters for payroll, because the 2023 rule puts mandatory liquidity fees on institutional prime and tax-exempt funds and discretionary fees on any non-government fund. Check the prospectus for the word government in the fund’s name and its investment policy, not in the marketing name of the sweep.

What yield on idle startup cash is realistic right now? I use the Treasury table, not a bank quote. On 15 September 2026 the Treasury table ran from 3.85 per cent at 4 weeks to 4.36 per cent at 52 weeks on the coupon equivalent measure. A government fund will usually track the short end of that. A bank sweep usually pays less, sometimes much less, and the gap between a sweep rate and the 4 week bill is the most expensive number most seed stage companies never look at, because it arrives as a slightly smaller interest line every month rather than as a bill.

When does a money market fund liquidity fee actually apply? For institutional prime and institutional tax-exempt funds, when daily net redemptions exceed 5 per cent of net assets unless liquidity costs are de minimis. For any non-government fund, whenever its board decides a fee is in the fund’s interest. Government funds are outside both, which is the single fact I would put on the first page of any treasury policy a seed stage company writes.

Does the t bill state tax exemption really matter for a company? The Treasury page says it plainly. There are no state or local taxes on bill interest. Whether that becomes a real saving depends on how your state taxes the company’s income, so it is a question for your accountant before it is a line in the policy. My guess is that for a company taxed in a high tax state it beats most yield comparisons, but that is a guess, not advice.

A short digression about the word sweep

Sweep describes where the money goes automatically and says nothing about what it goes into. I have seen the same bank offer a deposit sweep, a government fund sweep and a prime fund sweep under 3 nearly identical product names, and I still find it hard to read a sweep agreement without wanting a highlighter. Anyway, back to the numbers.

What is not settled here

How many startups hold idle cash in prime funds rather than government ones. Nobody publishes that split, and I do not know of a survey that asks in a form you could act on.

The size of the new minimum liquidity requirements. The SEC release says they went up and does not give the figures, and the full rule text was outside the scope of this piece, so no percentage goes into this article.

Whether the fees have ever been charged in practice since the rule took effect. That would be the real test of the prime versus government distinction, and I have not found a published instance either way.