Forecasting only works in a booming economy - CNBC Africa

Forecasting only works in a booming economy


by Sid Wahi 0

“But professor, how do you know that sales are going to grow by 26.15 per cent over the six years?” The short answer? She doesn’t know.

After four years at business school, I don’t claim to be an expert on economic or financial forecasting. My experience is just a drop in the ocean when compared to what experts like Fama & French or every investment banker out there has learnt through the years. But still, I have a fundamental problem with the assumption that value today, depends on the theoretical value tomorrow.

There are a few concepts we are all pretty much familiar with and now, I feel that it is my moral obligation to expose them for what they really are. 

Operating Cash Flow Forecasts

  • Sales forecasting: While it may be possible to forecast demand and sales targets for the current year in numerous industries, without a problem, most valuation methodologies use 5 to 15 year forecasts. Whether it be regression or multi-variant moving average analysis, the drivers of these forecasts are indeed based on past data. Don’t get me wrong, they may be a decent predictor for performance in the short run, however, they fail to account for unseen events (black swan is too graceful for a clusterf@#k event, so let’s go with flying dodo – no offence to Taleb).

    What happens if a meteor shoots out of the sky and wrecks your production facility? Daily volume? ZERO. Sure you can assign a probability of occurrence to a similar event, or better yet, spend millions on mitigating every possible risk out there, but that is again based on either past experiences, or your perception of the future (bias?) or the latest study carried out by Mckinsey and Co. Historical Data and manager biased probabilities don’t seem to have too much of a practical basis to me.
  • Operating Cash flow: If your topline is wrong, that just leaves the middle and end to screw up, so let’s start with the basis for most OCF projections. There are a number of methods, percentage growth of sales, historical percentage growth of expenses and common size income statements. One way or the other, they are a function of sales. And like the above, if you don’t know what is happening tomorrow, how can you project an income statement 10 years into the future? The final years of any projection are essentially shots in the dark. Beyond the first year of the model, the ability to accurately forecast earnings diminish significantly. And like with sales forecasts, the numbers of tomorrow depend largely on the numbers of yesterday. I’m sure in 2006, Kodak were expecting 45 per cent gross margins from their film processing business in 2015.

The Capital Asset Pricing Model: The basis for this model was set by Harry Markowitz as a tool for reduced risk through stock diversification. It is used commonly by investment professionals (investment bankers, asset managers, etc) to determine the cost of equity and ultimately is an input for intrinsic discounted cash flow based valuation. My issue with this measure is with two of its components: the beta and the market risk premium.

  • Beta: Simply put, a firm’s beta coefficient tracks its performance versus the overall market. This is calculated using historical stock market returns of the firm versus an index as a benchmark. The problem here, is again, HISTORICAL RETURNS. The same issue pertains to Market Risk Premium, which again, is the difference between the return of the market versus a risk free alternative (usually long term government bonds). Essentially, the underlying principle is that we can use yesterday to predict tomorrow and all of that information is captured by one variable.

    Sure there are adjustments. My favorite is the adjusted beta which is a weighted average of the raw beta and a beta of one. The principle here is that a firm’s beta over time will move towards one as it becomes more efficient. But in how long? These assumptions will only work when the economy is doing well, and I can prove that to you. It’s much easier to determine how much of the next iToy Apple will sell when the economy is booming - it’s as much as they can produce! But during a downturn, they know their sales will decline. And by how much is another story. It still works if you exceed the expectations you predicted with your “adjusted beta”, because now, you just call excess returns “alpha”.

    Oracle wasn’t joking when it named its predictive modeling software package “Crystal Ball”. I guess what I am really driving at over here is that, if I were to value your company today at 100 dollars based on a stream of payments I forecast using historical data, what it should receive over the next 10 years, I am taking a pretty risky bet. It’s a nothing more than a gamble window dressed to look like there may be some theory behind it.
  • Growth Rates and Terminal Value: These factors are perhaps the most sensitive inputs in a discounted cash flow analysis (the most common type of valuation). Let’s not go into how these variables are calculated, or the rationale behind constant growth rates, or else we may be here a while. The key takeaway here is that a small change in either variable can significantly impact valuation. For instance a drop of 100 basis points on your WACC could move your valuation up by as much as 200 per cent. Using a growth rate of just 4 per cent could mean that your terminal value component (the present value of the final years cash flow grown at the growth rate indefinitely) could be as high as 80 per cent of your total value, of value of the forecasted years. Small errors here (assuming that we know what a company really is worth) could mean significant gains or losses for either the buyer or seller.

In Closing

The only true way to assess value is to find out what someone else is going to pay you for your company or capital asset. I may buy your car for 50 bucks, because that’s how much its worth to me, and if you haven’t eaten for a week, 50 bucks is exactly what you need.

You can’t predict stock prices tomorrow, but are pro forma statements modeled 5, 10, 15 years into the future a great way of measuring intrinsic value? As I mentioned earlier, I’m no expert on the subject, just a confused analyst trying to make sense of a complicated world. Maybe investor expectations and forecasts are just fundamentally flawed and we have grown too accustomed to them. Maybe we need to forget the idea of looking at future profitability and create a system focused on profitability today. Whatever the answer is, make sure your business is inherently futuristic and overly reliant on historical data to make assumptions on the future.