What is wrong with price/book value

As so often, Richard Beddard tweeted something interesting that I had to respond to.

The linked PDF argues that price to book value is the most useful valuation measure, more so that earnigs, for two reasons:

  1. The track record of price/book value strategies, and,
  2. the theoretical consideration that a company must be worth more than its resale value.

The article does acknowledge that book value does not accurately reflect the actual value of assets, but merely argues that it is a good enough approximation. This leads to the first of my questions:

  1. Is it still as good an approximation as it was, given changes in accounting standards?
  2. Can it be used to compare the history of companies with different histories?
  3. Does it make sense for services, software and media companies? Does it make sense when the value of a company comes from its brand, or internally developed pate? Given the increased importance of companies of these types in the market, is the measure still useful?
  4. How does buying companies with low price to book value compare with other value strategies which have also historically out-performed?
  5. How strong is the evidence of out-performance?

I will start with the last two of these, because I have not an inkling of the answer. I have not looked at the data, or any papers. I am generally sceptical of any such results as very few people do anything like enough back testing. This is not a problem that is limited to finance: I have previously blogged about this as a potential problem with global warming models. I have looked at earnings based value strategies, and I am convinced that the value effect is probably real.

The change in accounting standards has had an impact on balance sheets that varies from company to company, but it will be particularly great in certain cases: for example, a company that has a history of acquiring others with assets with a book value greatly different from their fair value, may have a far higher book value than it would have done under the old rules, or compared to a peer that has developed equivalent assets internally. Fair value also opens some loopholes that allow inflating book value with optimistic estimates.

The change also reduces the value of a historical correlation of price to book value to out-performance. If book value is not what it was, why should the correlation be what it was?

The advocacy of book value seem partly to stem from a fundamental misunderstanding of what book value is. It is not (in most cases) based on the market value of assets, but on the historical cost of acquiring assets. Over time, between companies, and between sectors, these would diverge hugely. Over-reliance on this metric would lead to many missed opportunities.

I see no reason why price/book should be of anything more than secondary important (except for sectors such as property or mining). It does have a place in stock screening for value investing, but even there it is less important than long term pe or dividend yield.