Calculating Performance

[Eluna]

How do you all go about calculating the % return on your retirement portfolio? So far I have been using this formula: end of year total/ (start of year+contributions) - 1 to get my overall performance across the accounts.

I have just been ignoring dividends currently as they are not very much yet, but the plan is to just include dividends in the contributions as the plan right now is to just reinvest them. Is this how you all do it or am I missing something?

[bovie]

You are missing something: cashflows.

Use XIRR.

[Eluna]

You are missing something: cashflows.

Use XIRR.

Sorry for the possibly naive question, but what do you mean by cashflows?

[FrancisBayes]

I submitted a column (since approved) about how to calculate annualized real returns, so I will post it in this thread later when it's published. But until then...

1. Calculate money-weighted return (MWR) using the XIRR function: https://www.whitecoatinvestor.com/ho...xirr-function/
- Cashflows = deposits and withdrawals into your accounts
- Ignore dividends unless you withdraw them
- Calculate MWR for accounts and portfolios (eg, retirement, college), not each asset class

2. Calculate after-inflation (real) return
= (1 + MWR ) / (1 + inflation rate) - 1

[bovie]

Sorry for the possibly naive question, but what do you mean by cashflows?

The money you put into and take out of the account.

The timing of the cashflows, and associated returns of the assets you hold with respect to that timing, affects your overall return.

[Tim]

I submitted a column (since approved) about how to calculate annualized real returns, so I will post it in this thread later when it's published. But until then...

1. Calculate money-weighted return (MWR) using the XIRR function: https://www.whitecoatinvestor.com/ho...xirr-function/
- Cashflows = deposits and withdrawals into your accounts
- Ignore dividends unless you withdraw them
- Calculate MWR for accounts and portfolios (eg, retirement, college), not each asset class

2. Calculate after-inflation (real) return
= (1 + MWR ) / (1 + inflation rate) - 1

- Ignore dividends unless you withdraw them
By this are you simply saying to ignore the cashflow of the dividends?
since 1945, reinvested dividends have contributed 33% of the total return in the S&P 500. Essentially, dividends can improve your performance by a third without doing anything. “
Reinvesting to rebalance is different than automatic reinvesting. Cumulatively dividends are significant.
Thoughts? Why ignore?

[Eluna]

You are missing something: cashflows.

Use XIRR.

So I guess I am considering cashflow in that I am adding in my contributions, but I am not doing anything with the date added which it looks like XIRR factors in. Certainly seems like it would give me more accurate data vs what I’m doing so will need to play around with that function some.

[Eluna]

I submitted a column (since approved) about how to calculate annualized real returns, so I will post it in this thread later when it's published. But until then...

1. Calculate money-weighted return (MWR) using the XIRR function: https://www.whitecoatinvestor.com/ho...xirr-function/
- Cashflows = deposits and withdrawals into your accounts
- Ignore dividends unless you withdraw them
- Calculate MWR for accounts and portfolios (eg, retirement, college), not each asset class

2. Calculate after-inflation (real) return
= (1 + MWR ) / (1 + inflation rate) - 1

Thanks for the link! I will definitely need to play around with that function. Not entirely sure why dividends aren’t considered contributions, but I won’t claim to have even the slightest idea how that function works so guess it just factors them in when?

[bovie]

Thanks for the link! I will definitely need to play around with that function. Not entirely sure why dividends aren’t considered contributions, but I won’t claim to have even the slightest idea how that function works so guess it just factors them in when?

To slightly correct what FrancisBayes wrote, a dividend can be ignored if it is not automatically reinvested. Whether it’s actually withdrawn is irrelevant.

If it is not automatically reinvested, it is treated as a cash outflow from the security paying the dividend (which is what it actually is).

Then it is consequently treated as a cash inflow to whichever other security you choose to reinvest the dividend proceeds in—or even as a separate inflow to the same security that originally paid the dividend, if it happens to be reinvested on a date other than that on which the dividend was paid.

And if the dividend is never reinvested, then it simply remains an outflow with no corresponding inflow.

Again, it is the timing of the cashflows that is the key here. The XIRR function simply solves for the rate of return given a starting value and date, an ending value and date, and non-periodic cashflows between them with corresponding dates. That’s it.

So if you take money out on Monday (dividend paid) and then automatically put the same amount of money back in on that same Monday (dividend automatically reinvested), the result is the same as not recording either transaction (zero sum).

See?

[Eluna]

To slightly correct what FrancisBayes wrote, a dividend can be ignored if it is not automatically reinvested. Whether it’s actually withdrawn is irrelevant.

If it is not automatically reinvested, it is treated as a cash outflow from the security paying the dividend (which is what it actually is).

Then it is consequently treated as a cash inflow to whichever other security you choose to reinvest the dividend proceeds in—or even as a separate inflow to the same security that originally paid the dividend, if it happens to be reinvested on a date other than that on which the dividend was paid.

And if the dividend is never reinvested, then it simply remains an outflow with no corresponding inflow.

Again, it is the timing of the cashflows that is the key here. The XIRR function simply solves for the rate of return given a starting value and date, an ending value and date, and non-periodic cashflows between them with corresponding dates. That’s it.

So if you take money out on Monday (dividend paid) and then automatically put the same amount of money back in on that same Monday (dividend automatically reinvested), the result is the same as not recording either transaction (zero sum).

See?

Thanks for the fantastic explanation!

[FrancisBayes]

- Ignore dividends unless you withdraw them
By this are you simply saying to ignore the cashflow of the dividends?
since 1945, reinvested dividends have contributed 33% of the total return in the S&P 500. Essentially, dividends can improve your performance by a third without doing anything. “
Reinvesting to rebalance is different than automatic reinvesting. Cumulatively dividends are significant.
Thoughts? Why ignore?

I think bovie explains my thought process better than I could’ve. “Ignore” dividends not in terms of performance but as entries for XIRR…IF automatically reinvested.

But even if not reinvested and not withdrawn, I’d think dividends should be ignored because XIRR should be used for portfolios and account types, not asset classes. Unless you subtract cash in your brokerage account when you determine your portfolio’s value, the performance of cash sitting in your account is still part of XIRR/MWR.

[bovie]

I think bovie explains my thought process better than I could’ve. “Ignore” dividends not in terms of performance but as entries for XIRR…IF automatically reinvested.

But even if not reinvested and not withdrawn, I’d think dividends should be ignored because XIRR should be used for portfolios and account types, not asset classes. Unless you subtract cash in your brokerage account when you determine your portfolio’s value, the performance of cash sitting in your account is still part of XIRR/MWR.

Perhaps a little more clarification is warranted here.

I’d argue that XIRR is indeed more for asset classes than account types—or more specifically, for individual holdings.

I don’t really care what the return of my brokerage account was for the year. Or my Roth IRA. I do care what each of my funds returned for me—regardless of where they are held—which will not be the same as the time-weighted returns published by the brokerages.

And, of course, I care about the return of the portfolio as a whole. This is done easily enough by having an XIRR calculated for each holding, as well as an XIRR calculated for the column with the daily cash flow sums, of all holdings in all accounts—and if for some reason you want to know the returns at the account level as well, this is easily calculated with just another column of cash flow sums.

Regarding the dividend that is paid, not reinvested, and not withdrawn—this does still need to be treated as an outflow. Because that is exactly what it is. It will affect the XIRR calculated for that specific asset which paid the dividend—over long periods of time, the effect will be enormous.

However, the proper way to then deal with this is to consequently create an inflow for your money market fund on the same date as that on which the dividend was paid.

This way you will have a net zero flow for the portfolio as a whole (no money into or out of overall, just shuffled around from within), but the return calculated for your dividend-paying asset will be correct—and on top of that you will be able to correctly calculate the return on your cash in the form of your money market account return.

This is the most accurate and granular way to utilize XIRR.

[FrancisBayes]

I’d argue that XIRR is indeed more for asset classes than account types—or more specifically, for individual holdings.

I think it depends on whether one is a "buy-and-hold" investor or actively managing their assets (eg, stock-picking).

If I'm buying-and-holding, I can't compare the performance of S&P 500 index fund that I buy bimonthly in my 401(k) with that of small-cap value ETF that I bought lum-sum in my Roth IRA in January. Even the timing of rebalancing is arbitrary for comparison of my asset's performance. It's not the asset's fault if I rebalance before it bottoms or never rebalance. You have to use CAGR to compare asset performance over a period of time.

In contrast, you might want to track the performance of your "fun-money" brokerage account or 529 account. In that case, you can track cashflows and compare performance of each account (but ignore dividends as cashflows). Or even stock A vs stock B (but count dividends if not reinvested).

[bovie]

I think it depends on whether one is a "buy-and-hold" investor or actively managing their assets (eg, stock-picking).

If I'm buying-and-holding, I can't compare the performance of S&P 500 index fund that I buy bimonthly in my 401(k) with that of small-cap value ETF that I bought lum-sum in my Roth IRA in January. Even the timing of rebalancing is arbitrary for comparison of my asset's performance. It's not the asset's fault if I rebalance before it bottoms or never rebalance. You have to use CAGR to compare asset performance over a period of time.

In contrast, you might want to track the performance of your "fun-money" brokerage account or 529 account. In that case, you can track cashflows and compare performance of each account (but ignore dividends as cashflows). Or even stock A vs stock B (but count dividends if not reinvested).

Going to have to disagree here. It does not depend on whether one is a buy-and-hold investor or a stock picker.

Both of these people are clearly interested in the returns of their individual assets, as well as their portfolio overall.

Are you saying that buy-and-hold investors should not calculate the returns on their investments? That's what it sounds like, which of course is ridiculous.

I'm not sure what you mean when you say you "can't compare the performance of S&P 500 index fund that I buy bimonthly in my 401(k) with that of small-cap value ETF that I bought lum-sum in my Roth IRA in January." Sure you can; comparison (and general tracking) is the whole point of XIRR. And CAGR.

Now, it is more appropriate to compare those returns individually to their respective benchmarks. And what anyone does with all of this information in general is, of course, up to them. But any comparison can be made.

That said, most people will be adding to positions over time anyway so as to maintain their desired allocations, so that "lump-sum" of yours is going to have to get XIRR applied to it eventually if you want an accurate return. And even if you don't add to the position, XIRR will exactly equal CAGR if there are no subsequent cash flows. That's literally what it calculates.

But the point here is that you can calculate XIRR for individual assets, and then additionally from those same inputs you can calculate XIRR for the portfolio as a whole. But you can't do it the other way.

You can't get granular data (i.e., asset-level) from coarse data (i.e., portfolio-level). But you can get coarse data from granular data. So why wouldn't you?

Regarding "having to use CAGR to compare asset performance over a period of time," this is not true. This is exactly what XIRR does. XIRR is just a weighted average of individual CAGRs. Nothing more. They are directly comparable in terms of returns.

One is for the annualized return on a single investment made with a lump sum and with no subsequent cash flows, and the other is for the annualized return on a series of cashflows. They are both smoothed annualized rates of return, and can be directly compared assuming that they were utilized appropriately, and will exactly equal each other if both used for a one-time lump sum investment.

[FrancisBayes]

Sure, you can track individual assets, but for what purpose? It depends on the question that you're hoping to answer with XIRR.

In most years, stocks increase in value, so lump-sum will outperform DCA. For one year, let's assume that SCV and S&P500 have identical annual returns. If I buy SCV lump-sum in January and S&P500 bimonthly over 12 months, SCV might seem to have outperformed with XIRR for one year. But SCV IRL did not. If you tilt towards SCV thinking it would outperform large-cap stocks, then XIRR is going to mislead you unless the parameters are the same.

IMO, XIRR's most useful for ensuring that your portfolio's annualized return doesn't lag behind your expected annual return in your financial plan assumptions. It doesn't matter how your individual assets perform as long as your portfolio returns fine. If it doesn't then you have to reevaluate your asset allocation, asset selection, and investing strategy. For the asset allocation and selection, you would not use XIRR to reevaluate. For investing strategy, you would (e.g., do I tend to let cash sit idle for too long?).