Effectiveness of swapping between muni & treasury money markets

[indexfundfan]

Rounding was what I suspected too, but those synthetic 1-day yields that went negative cannot be explained away because of rounding.

Would you be willing to share the data source so we can look at the calculations?

I took the data from Fidelity using the method you provided earlier. Copied and pasted the data into a text file which I loaded into Octave (a free clone of Matlab) to process.

Here are the files for anyone interested

https://drive.google.com/file/d/17LOxQg ... sp=sharing

[hkcj]

When I need to make the swap every 10 days or whatever it is, If I am picking up 20 extra dollars every day by doing this, I don't get why it is so silly. Large money managers chase around basis points for things. Just because some people here don't want to pick up a nickel doesn't mean other people don't build machines to Hoover up thousands of nickels

I doubt people would argue it isn't worth it if you have large amounts of cash.

Even then, though, some people with large amounts of cash have large incomes. That time might be better spent earning more money, etc.

Anyway I completely agree it's worth understanding - if it's likely to be of benefit for your balances.

For me it's probably not worth it with my current balances and alternatives (What would I pay for 10 minutes more free time? What do I expect to be paid for that?). (But I appreciated all the analyses helping me decide that.)

[Electron]

Maybe in the future the old quarterly cycles caused by normal flow of Muni coupon payments may even return if they not exploitable by the new insights gained by managers this year.

Quite a few years have seen a cyclical rise in June with a peak near the end of the month or in early July. According to the Schwab article that has been posted a lot of municipal bonds are issued with a July 1 coupon date and a July 1 maturity date. The increased demand for VRDOs as a short term holding in early July had a noticeable impact and yields then declined.

Earlier in the week I mentioned VMSXX hitting a peak of 4.05% in March 2020 with VUSXX at 1.01%. That was during the Covid crash in the markets.

There was a very similar yield spike in 2008 following the bankruptcy of Lehman Brothers.

In this case, the yield was even higher and it took longer for the normal relationship to return. I wonder how many of us would have been comfortable buying VMSXX at that time.

[Hacksawdave]

Maybe in the future the old quarterly cycles caused by normal flow of Muni coupon payments may even return if they not exploitable by the new insights gained by managers this year.

Quite a few years have seen a cyclical rise in June with a peak near the end of the month or in early July. According to the Schwab article that has been posted a lot of municipal bonds are issued with a July 1 coupon date and a July 1 maturity date. The increased demand for VRDOs as a short term holding in early July had a noticeable impact and yields then declined.

Earlier in the week I mentioned VMSXX hitting a peak of 4.05% in March 2020 with VUSXX at 1.01%. That was during the Covid crash in the markets.

There was a very similar yield spike in 2008 following the bankruptcy of Lehman Brothers.

In this case, the yield was even higher and it took longer for the normal relationship to return. I wonder how many of us would have been comfortable buying VMSXX at that time.

These two examples illustrate what happened during the credit seizures of 2008 and 2020. The swap index spiked well above the Federal Funds Rate by 400-600 bps. September 24th of 2008 the FFR was at 2% while the SIFMA spiked to 7.96. On March 18th of 2020 the SIFMA went to 5.20 after the Fed had two prior emergency meetings to cut rates to zero.

It took three rate cuts spread out in October-December of 2008 down to zero to free up the credit markets. In 2020 the Fed realized it early on and put the hammer down. This is what a real credit crunch/crisis looks like.

[Hogan773]

When I need to make the swap every 10 days or whatever it is, If I am picking up 20 extra dollars every day by doing this, I don't get why it is so silly. Large money managers chase around basis points for things. Just because some people here don't want to pick up a nickel doesn't mean other people don't build machines to Hoover up thousands of nickels

I doubt people would argue it isn't worth it if you have large amounts of cash.

Even then, though, some people with large amounts of cash have large incomes. That time might be better spent earning more money, etc.

Anyway I completely agree it's worth understanding - if it's likely to be of benefit for your balances.

For me it's probably not worth it with my current balances and alternatives (What would I pay for 10 minutes more free time? What do I expect to be paid for that?). (But I appreciated all the analyses helping me decide that.)

Sure

The various posts on "it isn't worth my time" assume that this takes some big investment of time. I literally spend 10 seconds once a day looking at the FZEXX yield on the Fidelity page. It is just one of many internet tabs on my browser on my phone. That's it. And then like every couple weeks I need to spend a couple mins logging onto Fidelity and ordering the exchange from one to the other. We are literally talking about minutes per month of my time. I don't think any person is THAT busy that they literally have no free minutes in a week or a month. And yes I guess people with lots of money make money but it's not like I have some magical ability to commit those extra few minutes per month to my job and goose my income further. But yes I get it that if one doesn't have much in the MM funds and we are talking about picking up a few cents per day, not worth it

[afan]

I think it is not so much the time and the perception that it is fruitless. The relative yields change daily. For those who are right on the borderline between treasury, taxable and munis, the difference at any one time may be minimal. Switching funds to hope that a future 0.05% difference in after tax yield will persist for a few days seems a fool's errand. One could just as likely find that they are constantly investing in the fund whose relative performance was higher last week but worse for the week after buying it.
In that case, even zero seconds invested in monitoring and switching may be counterproductive.

If the differences were large and sustained, then switching could be valuable. Of course, if the differences were sustained, then there would be no reason to switch frequently.

[retiringwhen]

If the differences were large and sustained, then switching could be valuable. Of course, if the differences were sustained, then there would be no reason to switch frequently.

I am in the middle rates so it is nearly equivalent at VUSXX, and VMFXX/VMRXX with those moments where VMSXX looks great. I instead just jump between Treasury and Federal once or twice a year based on current trends. That seems to work pretty well. 5 years ago, VMSXX was in the play, but that stopped being true after the 2017 tax changes. If I were in the 40% Federal range, I'd be pretty tempted to swap VUSXX and VMSXX as each one is better by a lot almost 50% of the time, but it does require diligence every 30 days or so.

[Hogan773]

I think it is not so much the time and the perception that it is fruitless. The relative yields change daily. For those who are right on the borderline between treasury, taxable and munis, the difference at any one time may be minimal. Switching funds to hope that a future 0.05% difference in after tax yield will persist for a few days seems a fool's errand. One could just as likely find that they are constantly investing in the fund whose relative performance was higher last week but worse for the week after buying it.
In that case, even zero seconds invested in monitoring and switching may be counterproductive.

If the differences were large and sustained, then switching could be valuable. Of course, if the differences were sustained, then there would be no reason to switch frequently.

Okay but that is the whole point here...it is not just random on any given day. VMSXX has been following a very specific "sine wave" type curve with its yields. That is what people have shown above. So as you watch that unfold then you just try to generally switch from one to the other around the time the sine wave is crossing to be on the lower side of VUSXX adjusted yield, and then again later once it has bottomed out and heading back to cross and be above.

It's not like today the adjusted yield is higher and tomorrow it will be lower for 2 days and then the next day it will be higher again. And since VUSXX doesn't jump around that much then it is easy to just watch the VMSXX sine wave and do some comparison.

[Electron]

These two examples illustrate what happened during the credit seizures of 2008 and 2020. The swap index spiked well above the Federal Funds Rate by 400-600 bps. September 24th of 2008 the FFR was at 2% while the SIFMA spiked to 7.96. On March 18th of 2020 the SIFMA went to 5.20 after the Fed had two prior emergency meetings to cut rates to zero.

It took three rate cuts spread out in October-December of 2008 down to zero to free up the credit markets. In 2020 the Fed realized it early on and put the hammer down. This is what a real credit crunch/crisis looks like.

Here is a chart showing the March 2020 yield spike in VMSXX at the start of the pandemic. These spikes are not surprising given the lower liquidity in the tax-exempt money markets along with the higher risks. The VRDOs held in these funds are derivative securities based on long term Municipal bonds that are backed by a bank letter of credit.

The chart also shows the period from 4-06-16 through 5-12-17 where the SEC yield on VMSXX exceeded the SEC yield on VUSXX. That period would have made things easy for anyone seeking the highest after-tax yield.

[Hogan773]

Curious if FSXXX would be a closer proxy to the yield on VMSXX? I was just seeing that the stated 1 day yield on FZEXX seems to run around 10-15bp lower than what I calc out as the actual day to day yield on VMSXX (using the "daily balances" function on Vanguard site to show the difference in accrued interest every day). Right now VMSXX is showing right around 2.9% which happens to be about the same as what FSXXX is showing for its one day yield

If anyone has a sense of whether FSXXX would be a better proxy or not pls advise. FSXXX is just one of several share classes (Class 1, Class 2, Select etc) of the Fidelity Money Market Tax Exempt fund so it could just be a coincidence that the yield on that FSXXX share class happens to match today. The yield on some other share classes is more in the 2.8 area which is in line with FZEXX

[retiringwhen]

Curious if FSXXX would be a closer proxy to the yield on VMSXX? I was just seeing that the stated 1 day yield on FZEXX seems to run around 10-15bp lower than what I calc out as the actual day to day yield on VMSXX (using the "daily balances" function on Vanguard site to show the difference in accrued interest every day). Right now VMSXX is showing right around 2.9% which happens to be about the same as what FSXXX is showing for its one day yield

If anyone has a sense of whether FSXXX would be a better proxy or not pls advise. FSXXX is just one of several share classes (Class 1, Class 2, Select etc) of the Fidelity Money Market Tax Exempt fund so it could just be a coincidence that the yield on that FSXXX share class happens to match today. The yield on some other share classes is more in the 2.8 area which is in line with FZEXX

FTCXX may be even better. I will need to get some history in place to figure out which is better.

[D Train]

I'm very late to this discussion, but caught up by re-reading the thread. And thanks again to retiringwhen for putting the tool together in the first place. Had a few random thoughts.

The main reason I started down this rabbit hole was to add robustness to my thinking about treasury function. We run large cash balances, and are in top fed/state rates, and near-top local rate (NYC).

I had been manually tracking MMF yields on a weekly basis since mid-2022, but in late 2022 the TEY delta got meaningful compared to bank demand deposits. But the muni MMF yields seemed unstable, so rather than switching, I decided to watch them more closely. I started at 30 day, then 7 day, then finally daily observations. And that's when the oscillation became obvious, and then my anecdata became demonstrable when I found retiringwhen's backtest chart. (Some of the TEYs were nuts, in one case briefly at nearly 8%) For us, assuming we remain in top marginal rates (can be uncertain since comp is very back-end loaded), the whole crossover between Muni & Treasury never makes sense.

But the oscillation is an interesting phenomena to see, but I havent yet read an explanation of the underlying mechanics. I've seen suggestions in the thread that it could relate to the Wed SIFMA swap rate reset. But this doesnt make sense to me because the peak-trough cycle seems to run about 1 month, so a weekly reset is really just reflecting underlying changes to the SIFMA index over the course of a month. If SIFMA = VRDO rates, what's the underlying mechanism to VRDO periodicity?

The other explanation I read is the semi coupons of munibonds, but that doesnt seem explanatory either. Bonds are generally going to pay semi, the periodicity doesnt only exist around coupon payments.

Finally, some people have talked about the munispreads during GFC and Covid, but that seems to obviously be a risk premia question because of systemic risk (or idiosyncratic covid/Lehman type risk, if you prefer), which has nothing to do with the monthly/40day periodicity that is observed.

One thought for retiringwhen to mull over -- is there a subset of people who would find it helpful to consider AMT risk for calculating TEY? This may be impossible to do, since I think a lot of taxpayers find AMT impossible to predict.

Anyhow, great amazing tool and discussion.

[retiringwhen]

One thought for retiringwhen to mull over -- is there a subset of people who would find it helpful to consider AMT risk for calculating TEY? This may be impossible to do, since I think a lot of taxpayers find AMT impossible to predict.

AMT is really a bridge too far.

I am not aware of any electronically retrievable calculation in realtime of the % AMT for a muni fund. The source I am using for the back-test data does not include that level of detail AFAIK. The best one could do is search previous year 1099s, Annual reports and/or ICI files to retrieve back looking data.

I do have the data for the last 5 years for VMSXX, but only that fund since Vanguard publishes their ICI files. BTW, the ICI files are an amazing level of detail for back-test and tax analysis! They do not seem to be published publicly widely by the rest of the industry though. Vanguard and State Street seem to be the only Fund Complexes that share this level of data to the general public. I would love additional links or data sources, especially for Blackrock and Fidelity!

I also think the AMT marginal tax calculation could get pretty complex and hard to predict, thus making even the marginal rate calculation sketchy.

My best advice is to look at one of the Fidelty AMT-free funds if AMT is really an issue and see if they are competitive with some hand calculations.

As an aside, here are the AMT% for VMSXX for the last 5 years. It is not even a very steady number:
2018 13.38%
2019 18.50%
2020 11.70%
2021 13.71%
2022 13.88%