There are hundreds of ways a DCF can be constructed - many of which are inaccurate, or don't capture issues like dividends and share repurchases correctly. The most accurate model that I have been able to come up so far, is as follows (the difference is in step c):
a) Calculate Free Cash Flow for each of the forecast years = Cash flow from operations - capex. As this would include cash flow from equity affiliates and cash outflow to minority interests, we do not need to account for them separately.
b) Calculate Unlevered Free Cash Flow = Free Cash Flow + Interest.
c) Calculate Unlevered Free Cash Flow/dil. share for each of the forecast years.
d) Calculate the Present Value, discounting using WACC. This is the Enterprise Value per share. If calculating one year price target, it is the two year out Unlevered FCF from which the forecast series to be discounted should start.
e) Subtract Net Debt/share. If calculating a one year price target, use next year forecasted net debt. Net debt = Long term debt - cash - net debt attributable to minority interests.
f) Add in value of investments accounted to by fair method and cost method/share.
That's it - we have the share price.. Simple!!
The crucial step is step (c) - I haven't seen any analyst do that. But if we don't, then share repurchases in outer years get modeled incorrectly. What is done by other analysts (and hitherto by me) is to go directly from step b to step d. And then subtract net debt, add fair value investments, and divide the resulting number by next year estimated dil. share count to come up with share price.
However, this is incorrect. For if share repurchases are being modeled in outer years, and we use next yr share count to calculate share price in the last step, it doesnt capture the decrease in number of shares that happens in outer years. While unlevered cash flow does go up (because as cash is used to buy back stock, net debt goes up, and so deos the tax shield provided by debt), the full impact of reduced share count doesn't get captured..
Maybe that is why DDM (Dividend discount model) is a better model to value companies.. But then in DDM, you have to make companies that dont pay any dividends to also pay dividends... Which has its own set of problems (what is the payout ratio etc..)