Or perhaps - can you live with them!
Many investors cannot live without forecasts.Either they want a plethora of detail for each holding, with sales, earnings, dividends, target prices etc etc - or if they are more aloof they may just want a "market return ' - (what the equity universe will do in the next year or similar).
So how practical is this? And how important?
Take a simple, "easy to model" company.
To forecast its earnings per share, you just need to predict its:-
- sales (so you need both price and volume)
- costs - fixed and variable by category with both external and internal rises and falls
- Interest charge (net), with allowance for interest on interest, and rate changes
- Working capital needs and gains
- Tax, both rate and timing
- Capital management in terms of cpapex, deals, dividends/buybacks
Say realistically a minimum of 12 parameters, but more likely 30 or more.
And say now on each one your ability to guesstimate it is high - you can always get within 10% of what actually eventuates. A big problem is that while you may be saved by compensating errors, you can also get your errors compounding. There is a sophisticated techinique called Monte Carlo simulation, whereby you could (if you knew the likely range on all the inputs), calculate the possible range on the output (earnings per share), but common sense tells you rather swiftly that the range of outcomes is wide. Very wide.
Then for the target price you just need to gauge the price-earnings ratio which the wider market will apply to that earnings figure on the date of interest, and BINGO - you can talk to a target price calculated to several decimal places. Wider. Much wider.
But this can be seriously comforting if you like figures to base your market positions on.
So how good are forecasts?- lets start with those from professionals.
Exhibit 1 (from James Bianco on Barry Ritholz' blog the Big Picture - http://www.ritholtz.com/blog/author/james-bianco/ ) show the US equity market analysts' efforts at predicting the aggregate earnings at the operating level on the SP500. As you can see, in the last 20 years, with all the fancy computers etc etc, the analysts, while right on several occasions, have in truth been consistently wrong on the same side.

And in exhibit 2 from Franco Busetti in his seminal South African book the Effective Investor, one can see at a glance the constancy, but invalidity, of paid economists' forecasts of the rand/dollar exchange rate

As Franco writes in his book and in his blog post here on tickertalk:-
"Our largest investing mistakes often come not from wrong decisions but from right decisions based on wrong forecasts."
"The graph shows the R/$ exchange rate together with the Reuters Econometer consensus forecasts. The thick line shows the actual path of the R/$ exchange rate and the thin lines show the "consensus" forecast’s expected path for the rand at different times....The picture has many amusing aspects. Rand forecasts are always for depreciation. When the rand is depreciating, the consensus forecasts further depreciation; when the rand is appreciating, it forecasts a change to depreciation. Only the last five of the 126 forecasts shown here were for appreciation. Effectively, over a third of the forecasts shown (those from 2002 through 2004) were therefore forecasting a turning point in the currency (usually a very brave thing to do) and they were all wrong."
"Remember the Harvard Economic Society’s Weekly Letter of 16 Nov 1929: “A severe depression like that of 1920-21 is outside the range of possibility.”"
"..there is a feeling that one has to have forecasts, no matter how bad, in order to do valuations. Second, wrong forecasts are always subsequently justified, leading to the perennial hope that they should do better in future. Classic excuses along the lines of “It would have been correct except for … (some factor) which happened and was unpredictable”, “It was almost right, except for...” or “It's still going to happen!”"
"Forecasting is intrinsically very difficult and the results are therefore usually poor. Consensus forecasts tend to anchor, use blind extrapolation, ignore fundamentals and mean reversion and are frequently irrational. The time we spend trying to make very accurate earnings and dividend forecasts is disproportionally large relative to their influence on valuations. We should spend more time analysing instead of attempting to forecast."
Read more at http://www.tickertalk.co.za/blog.php?user=francob&blogentry_id=2480
So?
1. If you must have a forecast, make your own. If you didn't know what a faceless forecaster, whose call is any way masked by "consensus", was thinking (or smoking?) as the forecast was made, how can you possibly use it? And if its wrong, who will you blame?
2. Think through the likelihood of your forecast being right, and then of that rightness making your decision to buy or sell the share be right.
3. Maybe think differently. Accept that you know little about the present and less about the future.
4. Try the "Jupiter Portfolio" approach. If a little green man came on a space ship, could see everything, but was humble and wise enough to know forecasts wouldn't help him - what would he put into his portfolio. And why?
Cheers
Stuart