Cómo entender la actual paradoja de los bonos y las acciones

In order to comprehend the current paradox of bonds and stocks, it is essential to examine how the AI boom is being financed. The two most striking movements in the markets this year are the magnitude of the increase in bond yields and the strength and breadth of the stock rally. The paradox lies in why the first has not destabilized the second.

Optimists attribute both to the same reason: the significant increase in corporate profits. According to this interpretation, higher returns are simply a reflection of the greater long-term growth expectations driven by the surge in AI investment. If profits can remain above historical levels, stocks appear cheap. The most optimistic view even goes further, expecting growth to reduce future government deficits.

While there is some foundation to this narrative, as bond movements have occurred in real yields rather than inflation expectations that markets discount, it falls short of three tests. It does not account for the visible strain in credit sectors financing the AI boom, nor does it explain why a simultaneous resurgence in deleveraging operations was observed, betting on assets that would presumably benefit from a weakening dollar.

By overlooking how expenses were paid, it presumes a permanent model that is far from true. Real yields, their slower cousin the natural interest rate, follow growth evolution roughly. But it would be more accurate to say they are driven by the private sector's desire to borrow, and in that sense, the changes over the past three years have been enormous.

There are three ways to finance a data center: using own capital, offering solvency as collateral for others to borrow, or borrowing oneself. This surge has done all three, in that order, and the order and magnitude largely explain the evolution of prices in the markets. Financing a tech boom with one's balance sheet, such as the $200 billion in cash available accumulated by the four biggest spenders (Microsoft, Alphabet, Amazon, and Meta) in 2020, plus the $650 billion annual earnings now generated, has two major advantages.

First, capital investment decisions can be made relatively independently of market capital costs. Second, the economy can benefit from all the advantages of that spending without the upward pressure on interest rates that would normally have occurred if the same spending had been financed through credit. Since this surge directly translates into higher corporate incomes and indirectly into increased wealth for the rest of the economy, investors' natural tendency is to extrapolate.

I estimate that free cash flow of hyper-scalers could have already fallen to zero, but since profitability and balance sheets remain strong, the shift from cash-based financing to credit-based financing aimed at equity investors seems a mere footnote. Even now, despite extraordinary bond issuances, the magnitude of tech issuances still lags behind public debt, leading some to conclude it is irrelevant for interest rates.

However, what matters is not just the scale, but sensitivity to price. Typically, corporate issuances decrease as yields rise, but tech issuances have gone the opposite way. Borrowing on this scale, indifferent to the cost paid, raises yields for everyone else - homeowners and even governments - and is likely to continue doing so as long as the stock market continues to validate it.

Ultimately, what will stop it is a change in current optimism about the profitability of AI: either concerns about excess capacity and margins, or the broader consequences of rising yields. Until then, the debate is likely to continue. However, inferring that the yield increase reflects a stronger economy that will eventually pay off the deficit is investing the causality in reverse.

Deficits worsen when interest rates rise, even when the increase is due to private sector borrowing, which is not very price-sensitive, as well as government spending. This cannot be solved by re-buying long-term debt and financing on even shorter maturities. In recent cycles, a magnitude of real yields increase like this has systematically removed money from risk and ended the cycle.

This time is not, at least so far. This is not a signal that profits have permanently installed at a higher level, but of the extraordinary way in which, until recently, the boom was being financed. Every bubble is explained twice. In its development, emphasis is placed on what investors have been buying. Only later does it become clear that the most important is how everything was financed.