5 Aug 2022
In late March, Berkshire Hathaway announced it was acquiring Alleghany, a US insurance company, for US$848.02 per share. This strange-looking price started at US$850 and deducted Alleghany’s investment banking fees to Goldman Sachs of US$27 million, or almost US$2 per share, which Berkshire refused to pay. It was a visible reminder of an opinion Buffett has voiced for decades about the utility of most investment bankers. We very briefly touched on this point in a recent monthly, when noting that Elon Musk’s weed-joke of an acquisition price for Twitter (US$54.20 per share) just so happened to match Twitter’s bankers’ fair-value calculation too. You’d be forgiven for suspecting that these fair values aren’t so much computed, as reverse-engineered for a desired outcome.
Fair-value opinions go back to the 1980s, when the Marmon Group proposed a leveraged buyout of TransUnion. The management of TransUnion agreed a deal, but shareholders objected to the price being too low, so they sued the board of directors in Delaware and won. The judgment held the board personally responsible for not ensuring the sale price was fair, thereby swiftly sparking a boom in both director liability insurance and advisory investment-banking fees; so much so that this case (Smith vs. Van Gorkom) was thereafter derided as the ‘Investment Banker’s Relief Act of 1985’. After such a ruling, no board was going to sell (or buy) a company without a certificate from an investment bank stating that the price was fair.
Our team has occasionally had the poor fortune of witnessing the warped incentives arising from this practice. One recent example, whose details differ from reality only in that we are not naming names, is fairly representative: acquirer company ‘A’ wants to buy target company ‘T’; some of T’s shareholders (including us) do not like the deal but management do (on which point they might of course be correct, but they might also be succumbing to the lure of their golden parachutes), so they ask their bulge-bracket investment bank ‘BBIB’ for a fairness opinion showing that the company is worth less than the deal price.
The acquirer also wants to show its shareholders that it is making a good deal, so it asks its own investment bank for a fairness opinion showing that the target is worth more than the deal price. It is hard to miss that these two requests are at odds, but no one in investment banking – or the management teams involved - seems to be bothered by this, a reality explained by the fee structure of advisory deals, which in most cases (including for our companies A and T) looks like this:
With most of the fee depending on the transaction closing, it really shouldn’t be a surprise that this fairness opinion has low information value, but it is still disappointing because information value is exactly why such opinions were created in the first place. A recent paper (here ) finds evidence of this bias: the investment banks of acquirers assign higher valuations to targets (via using a lower weighted-average cost of capital) than the investment banks of the targets.
To an investor, the risk of such a biased valuation is the dreaded ‘take-under’, in which she buys a stock thinking it is cheap, only to see an acquirer buy the whole company for far less than the investor thought it was worth. Take-under risk is particularly pernicious in situations where the earnings and share price are depressed in the near term but expected to recover – such as cyclical investments at the bottom of a cycle. Of course, in theory, this shouldn’t be a problem. If one were to model out a company’s cashflows for decades, any near-term cyclical earnings slump shouldn’t impact much on the long-term value of the business. Sadly, however, the long-term plays virtually no role in advisor valuations, which are instead hilariously bereft of common sense (in our view) and display a disarming lack of critical thought. Our example involving companies A and T illustrates this point well. Investment bank BBIB deployed the full smorgasbord of entry-level corporate finance valuation tools to value T: a) discounted cashflow (DCF); b) comparable trading multiple analysis (which uses peer Enterprise Value (EV)/EBITDA multiples to get to a fair value for company T); c) analysis of previous transaction multiples (looking at previous M&A deals in the sector to infer a fair EV/EBITDA multiple); and d) historic takeover premia analysis (taking the current share price and applying the average percentage premium paid in similar deals in the past).
It is worth mentioning, if you’ll excuse the technical interlude, that in theory the DCF valuation is the only one you need. A properly constructed one should give a reasonable bottom-up range of fair values for a company, independent of recent market moods. But in the case of companies A and T, as in most other cases we have come across, the DCF valuation only modelled four years of cashflows and then simply applied an EV/EBITDA terminal value to company T, with this terminal-value multiple inferred from peers. With company T and all its peers in the midst of a cyclical downturn, the peer-derived EV/EBITDA multiple was low: just 4.0-5.5x EBITDA, applied to recovered earnings four years out, which low multiple incidentally was one of the reasons we had bought the stock in the first place. Since most of the value of a company usually lies beyond the next four years, this DCF was really a comparable-multiple valuation in disguise. It is therefore just a duplicate of the second technique.
Comparable multiple analysis applies a peer-average EV/EBITDA multiple to current EBITDA, rather than EBITDA four years out as in the DCF. If current EBITDA is cyclically depressed, this method compounds the mistake in the DCF valuation by not only using a cyclically-depressed EV/EBITDA multiple (in the case of target T, the same 4.0-5.5x as above), but by also applying it to an abnormally low earnings number (in the case of T, the lowest EBITDA the company had earned in over a decade). In doing so, investment bank BBIB not only duplicated the disguised comparable analysis in the DCF valuation, but somehow managed to make it significantly worse.
The same problem appears in the third method, where historical transaction multiples in similar deals are calculated (the average multiple here might well be sensible, and it was in the case of company T) and then applied to current EBITDA, which in the case of Company T was cyclically depressed. The result is once again undervaluation of the target. Finally, historic takeover premia analysis: this one consists of looking at past deals, and instead of using the multiples of such transactions, it uses the share-price premium of those transactions to the price before the deal’s announcement. It is by far the worst of the methods cited here because, for starters, it does not correct for capital structure, which will impact the justified equity premium significantly, but more importantly because it makes no attempt at benchmarking valuation to anything, even to as poor a benchmark as the trading multiples of peers.
The main problem with these metrics is that they all benchmark off current conditions. A full-company take-over represents the ultimate long-term transaction, in which long-term earnings power should be modelled as sensibly as possible. Yet the valuations above make no serious attempt to do so, although they do provide the investment bank with a valuation range so wide as to be able to justify practically anything. We should point out that part of the reason it is considered acceptable to do things this way is that business schools teach it, which presumably goes to explain why, in addition to not having much time for investment bankers, Buffett doesn’t have much time for MBAs either. Still, Corporate Finance 101 now demands that we put all these valuations into the mixer, blend them together, package them in a document that looks suitably legal, and extract the final product, the fair value sausage. That’ll be US$30 million, thank you very much.
The epilogue here, courtesy of the few years that have passed since A took over T, is that T’s technology allowed A to capture a vital growth opportunity, turning A from a potential structural decliner into one of the leading next-gen players of its industry. In this respect, we were probably correct in our original assessment of the undervaluation of T, but such comfort does not compensate for the disappointment of having lost significant upside. Whether we like it or not, this is just one of the risks investors run, and like other risks there are ways of mitigating it. The main one is to try, where possible, to invest in businesses where management are long-term operators, own significant stakes and understand the value of their stock. We have found that they are the most unlikely to sell out too cheap, and certainly the least likely to justify a deal with “Oh but our investment bank said the price is right”.
The value of investments, and any income generated from them, can fall as well as rise.