John --
Yes, it takes into account all costs -- you can input whatever transaction costs you believe it will cost as a percentage of the sale price and it takes that into account, as well as the taxes you will pay on profit, depreciation recapture, etc. It takes all that into account and spits out an IRR for each year if you were to sell that year.
You are generally right about IRR, but the most important aspect of IRR is the timing of the cash flows. If I buy a 50K house for 30K, I have "made" 20K in equity. IRR takes into account when you actually get paid that equity.
Take this scenario of the 50K house bought for 30K. For simplicity of argument let's put aside transaction costs and taxes, to examinte the concept of IRR. In this case, I've made a 66% ROI on my 30K investment. If I sell it now, it's a 66% ROI. If I hold it for 15 years, I delay getting that 20K benefit for all that time, and all the while that 20K is just getting around 3% a year, the assumed rate of appreciation of the property. The simple math is that it's a 45% return over those years. BUT if I'm good at finding these deals, I'm way better off cashing in that 20K, even if I have to pay taxes and transaction costs, because I know I can get another 66% return with it, maybe even twice in a year, as opposed to 3%.
As far as a property that's purchased closer to retail, the returns are more on a bell curve, and the IRR will tell you to sell at a more distant date -- they seem to pretty consistently fall at around the 8-9 year mark. The IRR rates are lower at first, continually rise until you get to around 8-10 years, then they start to fall. Thus the optimum time would be to sell when the IRR is highest, in one of those years. I do understand the theory of the IRR calculations, not too many people know how to actually calculate it (it's a strange trial and error calculation best left to a financial calculator). However, to be frank and honest with you, I'm not exactly sure why the IRRs tend to peak in years 8-10, I just know that they do, and I trust that all money factors have been taken into account in getting those figures.
Each case is different, obviously, and are based on projections and not the actual real world numbers, but it gives you an idea going in of what your ideal exit time would be to maximize your return.
MIRR addresses the concerns you have about false positives -- it is "Modified" IRR that takes into account the fact that the money kicked out by an investment won't necessarily earn the same return as the money that's still tied up in the investment. I.e., if I'm making 14% IRR in year 5, and part of that return is rental income in the form of cash, that cash won't make the same 14% -- I have to put it into another investment that may make more or less than the 14% return I'm getting on the investment in the property. Thus you punch in an ROI of an alternative investment that you assume you can earn with that cash that is kicked out to the sidelines. Let's say it's a CD, so MIRR will calculate the IRR while taking into account the cash that was made in rentals at only a 4% return going forward. Thus it gives you a more accurate picture of your overall IRR.
As for Buffett, having grown up in Omaha myself, where there will surely be several churches named after him when he dies, I would never question his wisdom. But, he deals in stocks, we deal in real estate, and I think it can be safely argued that in real estate we can make a quicker ROI than in stocks, given the ability to purchase property at such great discounts that can be sold at retail only a few months later. It allows for greater ROIs on an annualized basis, if you are aggressive in cycling through investments for the maximum gain. I think that's why the IRR calculations in real estate are so critical.
