Recessions that began from high valuations brought a median bear market of 29%, against 17% for the rest, even though unemployment rose by about the same amount. 6/

Valuations provided no additional context on recession risk beyond profit margins. When a recession does occur, valuations matter. With margins falling and valuations high, the S&P 500 fell 15% or more within a year 25% of the time, against 14% with falling margins alone. 5/

Expensive markets can stay expensive as long as there’s no recession. Profits and profit margins tell you about business cycle risk. With profit margins declining, a recession began within the next year about 27% of the time, compared to 5% when margins were rising. 4/

When equity valuations were in the top third of history, a recession began within the next year 11% of the time. When valuations were not in the top third, it was 17%. So valuations provided no signal at all to recession risk over the next 12 months. 3/

The stock market is valued at 34.2x corporate profits. This is the market value of all corporate equities, not just the S&P 500. 2/

When money supply growth was between 4% and 8% and cyclical payroll growth was between 0% and 2% (today's conditions).... A 0.5% increase in core inflation was 2x as common as a 0.5% decrease over the next 12 months.

AI capex is clearly driving economic activity, but multiplier effects are seen everywhere. Manufacturing production and utility production are both rising strongly.

Over the last 10 years, AI hardware production is up 135% while the rest of goods production is down 1.2%.

Since 2023, AI Hardware has accounted for about 60% of growth in durable goods manufacturing.

I think the early data clearly shows a measurable impact of AI on productivity. Adoption is low, so it's not measurable in the aggregate.