Thought starter
Half of buyouts miss the plan they were bought on
Private equity beats the stock market on average, and one buyout in four loses money. Both statements are true, and they describe the same distribution. The one deal in four that returns more than 50% a year carries the average. Behind it sit the one in ten that goes broke and the half that never reach the plan they were bought on. So the question is why so many individual decisions, made by capable people with full data rooms, go wrong in the same direction.
The record, deal by deal
Fund-level returns are the number everyone quotes. Deal-level records are less flattering, and they have been studied long enough to be stable.
Roughly one deal in sixteen ends in bankruptcy or reorganisation: 6% of 17,171 buyouts worldwide between 1970 and 2007. A separate study of 7,500 investments found one in ten went bankrupt. It found one in four returned more than 50% a year. That pairing tells the whole story of the asset class in one line.
Losing money is more common than going broke. One database holds buyout deals collected from fund managers during due diligence. In it, the median realised deal returned about 2x the capital invested, before fees. A quarter of deals returned less than 60 cents in the dollar. The funds in that data lost money on roughly three deals in ten, and about 15% of their capital.
Falling short of the plan is more common still. A buyout is typically bought on a case of 2.5x to 3x. If the median deal delivers about 2x before fees, more than half of deals fall short of the case that justified the price.
We infer this from the two figures. The inference is also generous. The same study found funds manage exits to avoid reporting a multiple just below 1.0x, especially while raising their next fund. The recorded losses are a floor.
Every one of those deals had an investment committee, a model and a diligence report. The failure is usually a decision, and rarely a missing fact.
Why it happens
The research points to a small number of biases, and they push the same way. They push towards paying more and expecting more than the record supports.
The price follows the credit, not the asset. Across 1,157 buyouts, the cost of borrowing was the main driver of how much debt a deal carried. The same held for the price paid, after allowing for what comparable public companies cost.
More debt went with lower fund returns. Cheap debt adds nothing to what a business is worth. It lets a fund pay more, and funds do.
Money chases deals. When capital flows into funds, the prices paid for new investments rise. The rise bears no relation to how the companies later perform. A fund with money to deploy and a clock running is a partial judge of value.
The winner's curse. In an auction, the winner is the bidder who most overestimated the asset. Across some 22,000 acquisitions of private companies, nearly half of acquirers were worth less the day the deal was announced. The effect grew with more competing bidders, less information and more confident management. A competitive process is designed to find the most optimistic model in the room and give it the company.
The plan starts from a year the business never had. The planning fallacy takes this form in the deal room. The model anchors on the seller's plan. Levers explain the growth. Nothing the business has done explains it.
Competitors are assumed to stand still. The team that has spent three months on the deal is, in the researchers' phrase, in love with its recommendation. The result is a case with sized levers and a credible exit. It sits on a plan that starts from a base year the business never had, or grows at a rate it never has.
Known risks get explained away. Diligence is commissioned to confirm the thesis. By the final round, the room has stopped asking whether to do the deal. It asks how to get it done.
That is deal fever. Little is known about the business beyond what the seller has said. The risks that are known are being explained away.
The loss is never booked. The same incentives that inflate the entry case delay the admission afterwards. A manager may hold an underwater asset until it can be exited at 1.0x, or shade the reported mark. Holding on is rational for a manager raising the next fund, and expensive for everyone else.
The common thread is rarely the analysis. Most deal models are competent.
A study of 1,048 investment decisions weighed the decision process against the analysis. The quality of the process explained six times more of the variance in returns than the quality of the analysis. It also found that companies which worked at removing bias earned about seven points more on their investments. Good analysis cannot save a process that will not give it a fair hearing.
How to take the bias out
Better decisions before the bid matter more than better monitoring after it. The price is the one number that never recovers. In our experience across engagements, four practices separate the funds whose deals follow the plan from the funds whose plans follow the deal.
Bridge the plan to the record, driver by driver, before anyone argues about levers. Volume, price, mix, openings, share. Reconcile each to the last three years. Test the base year for one-offs. Name what changed for every step above the run-rate.
The relevant row on our grid runs between two cells. At one end: "the plan outruns the run-rate, and the only explanation offered is the levers". At the other: "the plan has been re-based to the drivers we believe, the base year tested for one-offs, and the price rests on that case, not management's". Most deals sit in the middle: bridged at the top line only.
Run the kill discipline first. Before the model is built, name the three things that would end the deal. Test them in order. Agree the walk-away triggers in advance. The top of that row is a fund with walk-away triggers agreed before the auction, and the discipline to use them.
Set the price before you hear the seller's. Anchoring is the cheapest bias to defeat and the one most often left alone. Write down the walk-away price at the start, with the case that supports it. Require a written reason for every move above it. Pay someone in the room to argue against the deal, and let them go last.
Use your own record as the reference class. The most useful benchmark for a plan is what the last ten plans you bought delivered, driver by driver. It beats the sector's growth rate. Few funds can produce that table. The ones that can buy differently.
After the close, follow-up is the same bridge run forward. Plan against actuals, driver by driver, every quarter. Each variance triggers a decision. Re-running the entry assessment at twelve months, honestly, shows the gap between the deal that was bought and the business that was acquired. It shows it while there is still time to act.
None of this needs a bigger data room. It needs the questions asked in the right order. And it needs people who write down the answer they would need before they hear the one they are given.
A deal check exists for this. Five dimensions, twenty-odd practices, a sentence per stage. It takes ten minutes, and you do it before the committee meets. The result is private to the person who answers.
Its verdict is one of four lines. Keep going. Slow down and finish the homework. Slow down, there are real doubts. Or stop.
Whichever it is, that is the conversation the committee should be having.