Distrust Forecasts, Part 2

Photo Credit: D.C.Atty || Scrawled in 2008, AFTER the crash started

=============================

Comments are always appreciated from readers, if they are polite.  Here’s a recent one from the piece Distrust Forecasts.

You made one statement that I don’t really understand. “Most forecasters only think about income statements. Most of the limits stem from balance sheets proving insufficient, or cash flows inverting, and staying that way for a while.”

What is the danger of balance sheets proving insufficient? Does that mean that the company doesn’t have enough cash to cover their ‘burn rate’?

Not having enough cash to cover the burn rate can be an example of this.  Let me back up a bit, and speak generally before focusing.

Whether economists, quantitative analysts, chartists or guys who pull numbers out of the air, most people do not consider balance sheets when making predictions.  (Counterexample: analysts at the ratings agencies.)  It is much easier to assume a world where there are no limits to borrowing.  Practical example #1 would be home owners and buyers during the last financial crisis, together with the banks, shadow banks, and government sponsored enterprises that financed them.

In economies that have significant private debts, growth is limited, because of higher default probabilities/severity, and less capability of borrowing more should defaults tarry.  Most firms don’t like issuing equity, except as a last resort, so restricted ability to borrow limits growth. High debt among consumers limits growth in another way — they have less borrowing capacity and many feel less comfortable borrowing anyway.

Figuring out when there is “too much debt” is a squishy concept at any level — household, company, government, economy, etc.  It’s not as if you get to a magic number and things go haywire.  People have a hard time dealing with the idea that as leverage rises, so does the probability of default and the severity of default should it happen.  You can get to really high amounts of leverage and things still hold together for a while — there may be extenuating circumstances allowing it to work longer — just as in other cases, a failure in one area triggers a lot more failures as lenders stop lending, and those with inadequate liquidity can refinance and then fail.

Three More Reasons to Distrust Predictions

1) Media Effects — the media does not get the best people on the tube — they get those that are the most entertaining.  This encourages extreme predictions.  The same applies to people who make predictions in books — those that make extreme predictions sell more books.  As an example, consider this post from Ben Carlson on Harry Dent.  Harry Dent hasn’t been right in a long time, but it doesn’t stop him from making more extreme predictions.

For more on why you should ignore the media, you can read this ancient article that I wrote for RealMoney in 2005, and updated in 2013.

2) Momentum Effects — this one is two-sided.  There are momentum effects in the market, so it’s not bogus to shade near term estimates based off of what has happened recently.  There are two problems though — the longer and more severe the rise or fall, the more you should start downplaying momentum, and increasingly think mean-reversion.  Don’t argue for a high returning year when valuations are stretched, and vice-versa for large market falls when valuations are compressed.

The second thing is kind of a media effect when you begin seeing articles like “Everyone Ought to be Rich,” etc.  “Dow 36,000”-type predictions come near the end of bull markets, just as “The Death of Equities’ comes at the end of Bear Markets.  The media always shows up late; retail shows up late; the nuttiest books show up late.  Occasionally it will fell like books and pundits are playing “Can you top this?” near the end of a cycle.

3) Spurious Math — Whether it is the geometry of charts or the statistical optimization of regression, it is easy to argue for trends persisting longer than they should.  We should always try to think beyond the math to the human processes that the math is describing.  What levels of valuation or indebtedness are implied?  Setting new records in either is always possible, but it is not the most likely occurrence.

With that, be skeptical of forecasts.