Acerca do custo do dinheiro (parte I)

Uma das coisas que me dá gosto fazer é pegar num texto escrito e tentar traduzi-lo para ciclos/loops de dinâmica de sistemas para tentar perceber as implicações de coisas que não vemos nesta jogada, mas ficam implícitas como consequências naturais prováveis do que está a acontecer agora. Recordo o que fiz com um texto de Edward Hugh em 2009.

Ontem, o FT publicou um artigo que me despertou para desenhar mais uns ciclos, "Why a US debt binge is starting to matter", de Ruchir Sharma.

2 ciclos que se reforçam, são autocatalíticos: 

• R1 - a bola de neve da dívida; e
• R2 - a euforia da IA

2 ciclos que balanceiam:

• B1 - o crowding out; e
• B2 - o travão do custo do dinheiro

"Runaway debt is starting to matter, triggering a global sell-off in government bonds last month. And the first material impact could be that higher interest rates on US bonds short-circuit the AI boom.

...

In recent decades, the government role shifted to providing constant stimulus, even in good times. Long after 2008, it continued to rapidly run up debts. This decade, the US has consistently run budget deficits of around 6 per cent of GDP - more than twice the average of earlier decades.

For much of this period, households and corporations avoided taking on too much new debt. It was only in the last year that tech hyperscalers began racking it up to finance the massive AI infrastructure build-out as their cash surpluses dwindled. Their debt levels, though, are still manageable for such large companies.

The substantial borrowing excesses in this cycle have so far piled up on government books. And problems begin where excesses run deepest.

A critical warning comes from interest payments on public debt, which in the last five years have more than doubled to over 3 per cent of GDP. 

...

My research now suggests the marker to watch is the yield on 10-year US Treasury bonds, the global benchmark for long-term borrowing costs, which is now at 4.8 per cent. When it decisively breaches 5 per cent, the upper end of its range since the dotcom period, the Al bubble could pop. This breach would signal the start of a new era of tighter money, in which Al mega projects will be harder to fund. When Big Tech must compete for capital with a government paying a yield of more than 5 per cent on bonds - which matches an inflation expectations-adjusted return of over 2.5 per cent many will find themselves crowded out of the debt markets."

Se um investidor consegue obter mais de 5% numa obrigação americana sem risco, muitos projectos empresariais têm de proporcionar retornos bastante superiores para justificar o risco. O financiamento torna-se mais difícil precisamente quando as empresas de IA necessitam de centenas de milhares de milhões para centros de dados, chips, energia e redes.

Vou tentar desenhar um segundo conjunto de ciclos, relacionado com  com o impacte disto em Portugal.