On the five-year anniversary of the collapse of Lehman Bros., we look back to see whether one of the more interesting StarMine models was a useful canary in the coal mine.
The model is the SCR (Structural Credit Rating), which forecasts the probability of default within 12 months, and provides an implied credit rating. As seen below, the SCR chart highlights the performance of the model on Lehman. By late 2005, the rating had peaked and was declining well in advance of the share price peak. By early 2007, the rating had downgraded to BBB and was flirting with junk status.
The model analyzes the volatility of the equity, asset drift and leverage. It has two significant advantages over the traditional ratings agencies â€“ one, it recalculates every night; meaning there’s no breaks for vacation, no gaps in coverage â€“ it simply reviews all available data every evening. To that same point â€“ because the calculation is driven by a machine, it’s not going to pause because something appears too big to fail. Second, there’s no conflict of interest. You don’t have to go far to find people talking about problems in the traditional rating model where the company/issuer being evaluated pays for its rating.
By late 2007, the StarMine rating for LEH was BB, and the stock was still above $50. The large spread between that and the agency rating of A would have given pause for thought, especially as the rating had barely fluctuated over a 10-year period and the stock price was well and truly off its highs.
Why use a credit model?
Most of the larger issuers tend to be listed â€“ so the credit market gives another way to assess stockholdings. This is common practice at hedge funds that look at all parts of the capital structure, not simply the equity.
What’s bad for a bond investor is normally bad for a shareholder. If a company is being downgraded and is having greater difficulties meeting its debt obligations, it’s hardly going to be raising its dividend or organising a share buyback. These signals tend to be very valuable for stockholders.
One of the more unusual credit models created by StarMine is a text analytics signal that looks for certain phrases and words in news, filings and transcripts that may highlight problems early on. It’s a helpful way to ensure phrases are not missed such as “may have a problem with covenants” in filings or footnotes.
Last, there’s a very traditional ratio approach that looks at the same ratios the traditional ratings agencies evaluate â€“ but uses 23 metrics and recalculates on a nightly basis. When you combine these together you get some interesting analytics that really reduce the risk of getting blindsided by events that start in other asset classes that affect the share price.
Here’s a quick screen â€“ U.S. companies with both the structural and smart ratios ranking in the top quartile (i.e. scores greater than 75) and an intrinsic value of less than 1 (to remove those overpriced crowded trades that the press loves covering).
So what have we got here? Stocks with strong fundamentals that are less likely to suffer a credit downgrade and that are trading below the StarMine Intrinsic Value model’s calculation of fair value. When evaluating a cheap stock, one of the key questions is whether the stock is really cheap (mispriced) or whether you’re missing something â€“high leverage, a market peak, volatile earnings, etc. A screen like this should increase the probability that a stock is genuinely mispriced. There’s also a market cap filter just to keep the list manageable. These credit models are an interesting way to evaluate stocks from another angle, or another means of idea generation. From a risk perspective, they’re also a very cost effective alternative to traditional ratings.
Receive stories like this to your inbox as they are published. Subscribe here and follow @Alpha_Now on Twitter or check out the Thomson Reuters Eikon blog. If you are looking to access Thomson Reuters data or analytics, register for a free trial.