The relevant wiki https://www.bogleheads.org/wiki/Paying_ ... itch_funds starts with the following:

TBH, I've always been confused by this page when considering switching from a specific "fund" to an "alternative fund". Are these funds with identical performances? The ones of which I'm aware (e.g., VTI / ITOT, VXUS / IXUS) have identical costs also, and this probably has to be the case. Are these funds with dissimilar performance? If so, why would the logic not apply to nonsensical things, such as considering switching from bonds to stock ETFs (where the considerations clearly must take into account the estimated mean and variance).Suppose that you have a fund in a taxable account, and you have an alternative fund with lower costs or lower taxes.

I'd like to ask if the following calculations make sense. As an

*example*, let's consider switching from MCHI https://www.ishares.com/us/products/239 ... -china-etf (Ishares MSCI China, which I unfortunately have due to a past horrible managed account at Merril) to VXUS. Let's say all gains are long term, and that taxation on both those as well as dividends is 25% (which is the case for me, but that's a separate story).

MCHI has a TER Of 0.59%, dividends of 3.26%, and over 5 years made -22.08%.

VXUS has a TER of 0.08%, dividends of 3.24%, and over 5 years made 18.61%.

1. One alternative is to just ignore the past value growth of each of the assets, but extrapolate the dividends. In this case, the dividend

is 3.26% * 0.75 for MCHI, vs. 3.24% * 0.75 = 0.73% for VXUS. For 5 years, the effective dividends are approximately 5 times that. This is additive - for each dollar, I need to count this as the excess net revenue.

2. Another alternative is to extrapolate the dividends and and value. In this case, the value of 1 dollar in MCHI 5 years from now is approximately 1 * (1 - 22.08 / 100) * (1 + 5 * 0.225 * 0.75 / 100) (using the approximation that (1 + \alpha)^n is about (1 + n \alpha)). This is because post taxation, the reinvestment of dividends increases the amount by a multiplicative factor of (1 + 0.225 * 0.75 / 100), but the growth of the value per year is around (1 - 22.08 / 5 / 100). Similarly, for VXUS, the value of 1 dollar in MCHI 5 years from now is approximately 1 * (1 + 18.61 / 100) * (1 + 5 * 3.24 * 0.75 / 100).

3. A more sophisticated alternative is to take into account the variance-covariance matrix between assets with some penalty, but I don't know what it should be, so I'm considering in the preceding only things roughly in the same "asset class", e.g., ex-US stock assets.

Regardless of the alternative chosen, the cost for switching out of MCHI is 0.25 times the capital gains in there. In both cases, I don't see where the TERs really come into play.

Do any of these two first alternatives make sense? Neither? Both? If both, how would you choose? Happy to learn from your views. Thank you very much.