James Montier: Who Is Cooking The Books – The C-Score via Value Investing Insight
In good times, few focus on such ‘mundane’ issues as earnings quality and footnotes. However, this lack of attention to ‘detail’ tends to come back and bite investors in the arse during bad times. There are notable exceptions to this generalization. The short sellers tend to be amongst the most fundamentally driven investors. Indeed, far from being rumor mongers, most short sellers are closer to being the accounting police. To aid investors in assessing the likelihood of accounting shenanigans, I have designed the C-score. When combined with measures of overvaluation, this score is particularly useful at identifying short candidates.
- Contrary to the silly populist backlash which sees short sellers as rumour mongers and conspirators, they are actually amongst the most fundamentally driven of all the investors I interact with. Rather than being some malignant force within the markets, in my experience short sellers are closer to the accounting police (something the SEC once purported to do!).
- Whilst companies often accuse short sellers of lying and conspiracy, it turns out that the accusers are often the guilty party. Owen Lamont from Chicago University has examined the battles between corporates and short sellers in the US between 1977-2002. He found that ultimately it was the shorts that were right; the stocks underperformed the market by a cumulative 42% over three years after the start of the battle.
- Inspired by the shorts’ focus on fundamentals, I have created a C (for cooking the books or cheating) score which seeks to capture how many of six common earnings manipulations a firm is engaging in. Of course, the C-score is only a first step in analyzing whether a company is or isn’t cooking its books.
- That said, the C-score does seem to have some power to separate out stocks which go on to underperform. Those stocks with high C-scores underperform the market by around 8% p.a. and 5% p.a. in the US and Europe respectively (over the period 1993-2007).
- Of course, the C-score is likely to be more effective when it is used in tandem with some measure of valuation. After all, it is often the case that high flying stocks will be more tempted to try to ‘cheat’ to maintain their status. This is borne out by the data. Stocks with a C-score of 5 and a price to sales ratio of greater than 2 tend to generate a negative absolute return of 4% (in both the US and Europe). Around 50-60% of the stocks identified see negative returns.
- A list of US and European stocks that currently have a high C-score and high price to sales ratio can be found inside. Highlights include such names as Apple, and Amazon in the US and Rio Tinto and Lonmin in the UK!
James Montier: Cooking the books, or, More sailing under the black flag
In my recent note on required reading for investors (see Mind Matters, 16 June 2008 http://sgresearch.socgen.com/publication/strategy_update(20080616)_2c3.pdf), I suggested that after a five-year bull run in earnings, investors were at risk of paying too little attention to earnings manipulation. In the good times, sadly few care about such mundane issues as earnings quality or footnotes (as foolish as that is). In the bad times, these characteristics tend to come back into widespread fashion.
However, in a world where most analysts are more concerned with forecasting quarterly earnings to two decimal places over the next five years, and writing up company press releases, their ability to actually analyze a company seems to be in danger of becoming a lost art. The most fundamental orientated analysts I have come across are without a doubt the short sellers. These guys, by and large, really take their analysis seriously (and so they should since their downside is effectively unlimited). So the continued backlash against short sellers as rumor mongers and conspirators simply leaves me shaking my head in bewilderment. I can only assume that those making these claims are either policy makers pandering to shorted companies, or shorted companies themselves. Rather than being some malignant force within the markets, in my experience short sellers are closer to accounting police.
The post was originally published here. Highlights: Resolving gas supply issues ensures longevity A pioneer in renewable energy should be future proof Undemanding valuation could lead to re-rating Q1 2022 hedge fund letters, conferences and more
James Montier: Companies lie, short sellers police: the evidence
This viewpoint was confirmed by a insightful study by Owen Lamont1 (then at Chicago University). He wrote a paper in 2003 examining the battles between short sellers and the companies they shorted. He examined such battles between 1977 and 2002 in the US. He focused on situations where the company that was being shorted protested innocence by suggesting it was the subject of a bear raid, or a conspiracy, or alleged that the short sellers were lying. He also explored firms that requested investigation by the authorities into the shorts, urged the stock holders not to lend shares out, or even set up repurchase plans (presumably to create a short squeeze).
The results Lamont uncovered show the useful role played by short sellers. The chart below shows the average cumulative return to the shorted stock. In the 12 months after the battle started, the average stock underperformed the market by 24%. In the 3 years after the battle commenced these stocks underperformed the market by 42% cumulatively! The shorts were right, it was often the companies that were lying and conspiring to defraud investors, not the reverse!
James Montier: Who is cooking the books – The C-score
I recently explored one method of screening for short candidates (see Mind Matters, 13 May 2008 2. However, it occurred to me that a more accounting based screen might also be useful in identifying potential shorts by looking at those who might well be cooking their books, or trying every last trick to ensure that they can beat the analysts’ quarterly forecasts. To this end I have created the C (for cooking the books or cheating) score to help measure the likelihood that a firm may be trying to pull the wool over investors’ eyes. The score has six inputs, each designed to capture an element of common earnings manipulation:
I. A growing difference between net income and cash flow from operations. In general, managements have less flexibility in shaping cash flows than earnings. Earnings contain a large number of highly subjective estimates such as bad debts, pension returns etc. A growing divergence between net income and cash flows may also indicate more aggressive capitalisation of costs.
II. Days sales outstanding (DSO) is increasing. This, of course, signifies that accounts receivable are growing faster than sales. This measure is really aimed at picking up channel stuffing (sending inventory to customers).
III. Growing days sales of inventory (DSI). Growing inventory is likely to indicate slowing sales, never a good sign.
IV. Increasing other current assets to revenues. Canny CFOs may know that investors often look at DSO and/or DSI, thus they may use this catch-all line item to help hide things they don’t want investors to focus upon.
V. Declines in depreciation relative to gross property plant and equipment. To beat the quarterly earnings target, firms can easily alter the estimate of useful asset life.
VI. High total asset growth. Some firms become serial acquirers and use their acquisitions to distort their earnings. High asset growth firms receive a flag in this score.
These elements are scored in a simple binary fashion, so if a company has increasing DSI it will receive a score of one. These are then summed across the elements to give a final C-score bounded from 0 (no evidence of earnings manipulation) to 6 (all the flags are present).
James Montier: Does the C-score work?
The C-score is only a first step in analyzing whether a company is or isn’t cooking its books. However, it does seem to work relatively well. The charts below show the performance of stocks by C-score across Europe and the US over the period 1993-2007 (portfolios are formed in June and held for one year).
See full PDF below.