A big deal, gone badly wrong

Good morning. Quite the retail sales number yesterday, up 3 per cent in January. Combined with Tuesday’s hotter inflation reading, more are wondering if the US economy is just plain overheating. Manufacturing and housing are still lousy, though. Got this economy all sorted out? Relieve our confusion: robert.armstrong@ft.com and ethan.wu@ft.com.

The FIS mess, part 2

The story of FIS-Worldpay, a $43bn deal that is being reversed after just four years, is important. Big companies do a lot of M&A, and they often make a mess of it, leaving investors worse off. It is worth understanding how things go wrong, and why the problems are so hard to detect before it is too late.

In an earlier piece on the deal, we argued that there was a tension in the way FIS talks about the deal and the proposed spin-off. On the one hand, the company says it had achieved $1.65bn in annual revenue and cost synergies from the deal. The present value of those synergies is in the neighbourhood of $10-$20bn, on plausible assumptions. At the same time, the company had downplayed the impact of losing those synergies by breaking up the deal, calling it “manageable”. But both of those things cannot be true.

The stated justification for the spin-off is that separating the two companies will allow them to pursue two different capital allocation strategies. FIS, which provides banks with payments technology, will be an investment-grade rated, slow-growing, super-steady, cash-generative business. Worldpay, which provides merchants with payments technology, will leverage up to do lots of M&A, as it must in a fast-changing industry. What has changed since 2019, we are told, is that “the pace of disruption” on the merchant side has accelerated, making the combination impractical.

But this explanation is an uneasy fit with the facts. Consider the case of Fiserv, a direct competitor to FIS in the banking software business. Fiserv did its own big 2019 deal to enter merchant services, a $39bn deal to combine with First Data, a Worldpay competitor. There are no controlled experiments in corporate finance, but Fiserve-First Data is as close to being a control group for FIS-Worldpay as one can hope for. The first indication that things have gone differently for the two dealmakers is the stocks, which traded closely until mid-2021, then diverged:

Much of this has to do with a divergence in how the market is valuing the two companies. Here are the companies’ price/earnings ratios over time. After moving in line in the years immediately after the deals, Fiserv’s valuation is 50 per cent higher than FIS’s now:

Finally and most importantly, look at the growth rates of the two companies’ merchant services units. Once again, parallel at first, then diverging:

The growth rates of both businesses have slowed — but while Fiserv’s is still growing 9 per cent in the most recent quarter, FIS’s shrank.

This pattern of facts is not terribly consistent with the idea that the merchant business changed radically, making a spinout of Worldpay necessary. I don’t know exactly what is going on here, but here’s one hypothesis that is more consistent with the facts: the FIS management that bought Worldpay made such a mess of the integration that the easiest thing to do now is separate the companies and fix them individually, perhaps with an eye to selling independent Worldpay to a more competent buyer in the future. This would also make sense of the hasty departure of the FIS management team that did the deal and the arrival of activist investors.

If this hypothesis is correct, how exactly did FIS bungle things? I don’t know, but there is one point worth mentioning. FIS executives were given incentive pay packages, potentially worth millions, for hitting synergy targets (you can read about this program in the latest proxy statement). The problem with this approach is that what counts as a synergy is a bit subjective. How much of the revenue added post-deal is attributable to the deal, and how much is just plain growth? How much of any cost cuts done after the deal were made possible by the deal? Who decides?

I’m not sure, but if you pay people a lot of money to find synergies, chances are they will find them. But trying to get as many sales and efficiencies as possible into the bucket labelled “synergies” does not seem like a good use of management time. That there might have been some pushing of boundaries in this area would be consistent with the fact that FIS claimed to have achieved big synergies, but doesn’t seem worried about losing them in the break-up.

Again, there is a lot I don’t know here. All I can say with certainty is that FIS’s explanation of the spin-off is a poor fit with the facts, and a badly mishandled integration is a much better one. If anyone knows more than I do, I hope they will email me.

In a sense, what the company says doesn’t matter. The stock price is down, the spin-off is happening, the old management team is gone. But I think FIS investors deserve a much clearer explanation of what has gone wrong here.

Berkshire sells TSMC

Sometime in the third quarter last year, Warren Buffett’s Berkshire Hathaway took a $4bn stake in TSMC, the Taiwanese contract chipmaker. A few months on, Buffett seems to have decided, eh, nevermind. From the FT:

[Berkshire] sold more than 85 per cent of its stake in chipmaker Taiwan Semiconductor Manufacturing, a position it had taken just months earlier when it purchased shares worth $4.1bn.

Berkshire cut its position in TSMC by more than 50mn shares, disclosing it held 8.3mn shares worth $618mn. The rapid shift is somewhat unusual for Berkshire, which tends to think of its investments in years or decades as opposed to months.

We’ve no insight into Buffett’s thinking, but it does seem an odd time for a buy-and-hold investor to sell. When Berkshire disclosed the TSMC buy in November, we thought it looked smart. Here was a well-run business with wide competitive moats and high profits. And yet the risk that China might invade Taiwan was keeping the stock at a discount to its peers, nearly 12 times forward earnings.

Today, China appears no closer to invading the island, and the discount to peer firms has closed a bit, to 17 times forward earnings. But TSMC still seems cheap (note that Intel’s p/e ratio below is inflated by its crashing profits):

TSMC’s competitive position within the chip industry is also no shakier. The company did issue cautious guidance last month, saying it might pare back capex as revenue contracts in 2023. But that is because there are too many chips and too little chip demand right now. It’s a sector-wide problem, not a TSMC problem. As investors liking chips’ secular growth story have started to notice, TSMC is outperforming the broader sector this year. Mark Baribeau, a portfolio manager at Jennison Associates, thinks the company is among the best-positioned semiconductor groups for weathering the next year or two.

More competition is coming, however. Intel and Samsung are dumping boatloads of cash into chip manufacturing, coveting TSMC’s process-tech edge. Yet tech leadership in chips is sticky; incumbents have to fumble. Robert Sanders at Deutsche Bank wrote recently that he is “relaxed” and “comfortable” about the threat Samsung and Intel pose:

Intel recently admitted that its tools are operating at only 20-25 per cent of the efficiency of equivalent tools at TSMC, which is likely to be a driver of a massive gap to TSMC on cost and yield learning. In addition, we note that a major advantage that TSMC enjoys is in advanced packaging [ie, bundling multiple chips into one chipset], which is likely to trump competition along with a far superior record of execution compared with Samsung and Intel.

That tech edge lends TSMC pricing power over its customers, even during a semiconductor bust, notes Charlie Chan of Morgan Stanley:

Our recent checks along the foundry supply chain suggest that TSMC will still pass through higher foundry costs to its customers in 2023 via a 3-6 per cent price hike. Coupled with the average ~10 per cent price hike in 2022 . . . 

TSMC also enables the Arm-based CPU, AMD’s share gain, and also China’s [supercomputing] localization . . . We believe these are the customers that will pay for the more expensive technology as the progress of Moore’s Law slows.

As much today as in November, TSMC enjoys durable advantages, pricing power and a history of good management. Its valuation remains cheap, too. Perhaps Buffett no longer thinks so, but it still looks like a vintage Berkshire buy to us. (Ethan Wu)

One good read

The clown show clowns on.