Who's minding the store?

What a small dividend bump and a huge capex increase might mean for investors in the Mag7

In the aftermath of the Internet bubble bursting some twenty years ago, Michael Jensen penned an essay on the high “agency” costs of overvalued companies.* Decades earlier, Jensen had outlined his agency view of investing in large corporations: Management is your agent; having them rather than you run the company is efficient and allows scale, but it comes at a cash, difference-of-opinion, and potential conflict-of-interest cost…+ In that context, a dividend payment is a small lever that minority shareholders have to constrain the potential excesses of their agents inside corporations.

Surveying the market ruins in the early 2000s, Jensen applied his agency framework to the specific problem of numerous overvalued equities, a scenario that did not really exist when his original argument was made. Two decades is the equivalent to a century or more in Wall Street years, as the investment profession is bereft of any sense of historical sensibility. So I had a chuckle when I saw a Mag7 company announce a miniscule dividend increment against the backdrop of a huge jump in outlays on AI infrastructure. What would Jensen think about this current crop of market leaders and their spending plans?

First, the details. In this particular instance, the company spent about $5.1 billion in dividends last year, its first full year of shareholder remuneration. It proposes taking the sum up ~$250 million to just over $5.3 billion or so in 2025. Those payments translate into a trailing dividend yield of 0.27% and a forward yield of 0.285%. That is, nothing. Free cashflow (FCF: CFO-Capex) in 2024 was roughly $54 billion. The company trades at 35 times that trailing FCF and more than 350 times the trailing dividend. Although the company has already been spending aggressively on Capex—$37 billion in 2024, up from around $15 billion per year prior to the AI/data center explosion, it plans to spend another $60-$65 billion in 2025. (The company says the spending will be on “core” operations, as well as “generative AI efforts,” but it’s no longer clear that there is a major difference between the two.)

Some of the Mag7 have de minimis dividends; others have none at all. All are spending huge sums on AI, at least $300 billion in 2025. We’re now clearly in Everett Dirksen territory.^ (Group member Nvidia is in the enviable position of being the recipient of that spend.)

Against that backdrop, it might be useful to review some of Jensen’s concerns. What he does not do—and I propose also not doing—is wrestle with the definition of overvalued. In his 2005 article, he dismisses the issue as self-evident. And in the aftermath of the Internet bubble, it was. But we’ve had high valuations now for many years. Investors are inured. So it’s not as clear as it might have been then what overvalued means now. It is fair to say, however, that the Mag7 are richly valued. Even their proponents would admit as much, pointing out that the growth opportunities for these companies are so abundant that their valuations simply and fairly reflect that rosy future of all things AI. And to judge by the business and investment outcomes of the past decade, it’s hard (but not impossible) to argue against that stance.

Instead, Jensen focused on the behavioral risks of managers operating inside what he considered at the time overvalued companies. His points are worth rehashing here. You can decide whether in the current context of all things AI and enormous infrastructure investment, the management of the Mag7 have moved beyond standard agency costs to eye-popping agency risks.

The first is the pressure on managers to meet expectations, whether they be in regard to revenues, margins, earnings or cashflow. A little pressure is good; a lot of pressure is not: “Generally, the only way for managers to meet those expectations year in and year out is to cook their numbers to mask the inherent uncertainty in their businesses… Once we as managers start lying in the earnings management game, it is nearly impossible to stop because the game cascades forward. If we are having trouble meeting the earnings targets for this year, we push expenses forward and pull revenues from next period into this period. Revenues borrowed from the future and today’s expenses pushed to tomorrow require even more manipulation in the future to forestall the day of reckoning.”

In the twenty years since those lines were written, you could argue the opposite has come to past: managers are incentivized to maximize up-front investments—keeping up with the investment Jones—and rewarded by the stock market for doing so. But the distortionary effect is just the same.

Next, Jensen delves into what he calls managerial heroin: “Like an addictive drug, manning the helm of an overvalued company feels great at first. If you are the CEO or CFO, you are on TV, and covered by the press, investors love you, your options are increasing in value and the capital markets are wide open to the firm. But as drug users learn, massive pain lies ahead. The source of the problems … lies in the following fact: by definition if your stock price is overvalued[,] we know that you cannot, except by pure luck, produce the performance required to justify the stock price. If you could, it would not be overvalued. So as time goes by it begins to dawn on managers of such overvalued firms that times are getting tough. You realize the markets will hammer you unless your company’s performance justifies the stock price… To appear to be satisfying growth expectations, you use your overvalued equity to make long run value-destroying acquisitions; you use your access to cheap debt and equity capital to engage in excessive internal spending and risky negative net present investments that the market thinks will generate value; and eventually you turn to further accounting manipulation and even fraudulent practices to continue the appearance of growth and value creation.”

That is very strongly stated, and here too a lot has changed since 2005, but the risks associated with being the market darling have not.

Jensen finally considers what steps might at least reduce the risk of bad investment decisions by managers in these companies. He dismisses many of the standard guardrails as being ineffective, but circles back to perhaps the original guardrail: corporate governance, the boardroom, the answer to the question: “Who’s minding the store?”

According to Jensen, the Internet bubble was not a failure of all the dark fiber or the busted .coms. It wasn’t even a failure of the managers who oversaw the value destruction. Instead it was a failure of the boards that greenlighted the spend, and approved the budgeting and compensation systems that led managers to make those bad decisions: “What we witnessed was massive failure in which the boards of directors of company after company failed to stop the corruption and the associated destruction of organizational value.”

His proposed remedies for board practice are peculiar, likely reflective of the characteristics of the Internet bubble, and not clearly widely adaptable today. But his location of the problem—the boardroom—is and remains spot on.

Now, about that $300 billion in data center spend. Yes to investing in new technology. Yes to moving fast and breaking things. Yes to creative destruction. But what boardroom analysis occurred? What were the dissenting opinions, if any? What are the repercussions to the board and management if they get it wrong? Did they ask the shareholders, whose interests they are supposed to represent?

We know the answers to these questions: Very little, not likely, none, and no. Well, perhaps it will work out this time. For the sake of numerous retail speculators and the much larger population of passive investors, I hope so.

And on Jensen’s original point that a dividend can serve as a tool for investors to manage agency costs, one would have to add, only if it is large enough to matter. De minimis dividends don’t count, either to investors or to the corporation.

Comments welcome.

---

*Michael C. Jensen, “Agency Costs of Overvalued Equity,” Financial Management, Vol. 34, No. 1 (Spring, 2005), 5-19.

+Michael C. Jensen & William H. Meckling, “Theory of the firm: Managerial behavior, agency costs and ownership structure,” Journal of Financial Economics, Vol. 3, no. 4 (October, 1976), 305-360; Michael C. Jensen, "Agency Costs of Free Cash Flow: Corporate Finance and Takeovers," American Economic Review Vol. 76, no. 2 (May, 1986), 323-329.

^https://www.senate.gov/artandhistory/history/minute/Senator_Everett_Mckinley_Dirksen_Dies.htm

Microsoft Co-Pilot generated image with a prompt of “No one minding the store.” I could not get it to correct the mispelling of merchandise. Oh, well.