Data dependence as "constrained discretion"

Ben Bernanke and Rick Mishkin called the use of data dependence in monetary policy "constrained discretion". We are in another of those periods of constrained discretion regarding inflation and monetary policy. 

Robert Hetzel, a former Fed economist, said that policymaking has a flavor of "guess and correct." There are forecasts and guesses of what should be the key target variables, inflation and growth, and then policy is corrected to move toward the target. 

The data dependence is an important part of the guess-and-correct view. There is no set rule, but a set of adjustments toward the expected target. 

Going broke and not taking profits

Good investors focus on risk. Risk is the downside. Bernard Baruch said it well when he said nobody ever went broke taking a profit. - Sam Zell

This is a classic rule in trading, yet it has little meaning. Anytime you have a positive profit, you can take it, yet you are likely leaving money on the table. Simply put, do you take profits on what could be noise or price variation due to current volatility? 

Perhaps better to take profits against valuation. This requires some valuation statement that may be a mistake, but it is a better requirement for profit-taking. 

Nobody ever went broke selling at or above fair value. 

Dispersion - what does it mean?

 From the newsletter, Owenomics, we see that stock dispersion is at levels not achieved since 2008 and 2000. Now, this may not be an indicator of a market top, but it does tell us something about market behavior. 

Dispersion is not the same as volatility. Volatility measures deviations from the mean over a given time period. Dispersion measures the deviation of returns across a set of assets. It is a cross-sectional measure. Higher dispersion means there is a greater variation in the winners and losers relative to the mean return. This could mean there is a disruption in the current market regime or a rotation between industries and firms. It could mean there are a few very strong winners or losers.

Past periods of strong dispersion include the bursting of the tech bubble in 2000, the bursting of the housing bubble in 2008, and the Great Financial Crisis. Disruption leads to dispersion, but greater dispersion does not necessarily imply a general market decline.

The end of being drunk on AI?

 

There is an interesting index on token expenditures called the Silicon Data LLM Expenditure Index that measures the price of tokenization for AI. It is showing a decline as users move to cheaper models. If the price of tokens increases, demand will respond. 

Companies are now monitoring token usage and prices and ensuring that employees don’t treat tokens as a free good. This is natural behavior on the part of firms, but it also means that revenue growth for AI providers will likely slow and fall short of market expectations.