Hi Professor David,
Indeed Jeremy Grantham is famous for his long-term market segment predictions.
Initially, he formulated his judgments for the upcoming decade; more recently he has shortened the timeframe to seven years. Initially, his projections seemed highly prescient; more recently his projections have proved less prescient. Like most of what happens in the marketplace, a reversion-to-the-mean iron law seems to be exercising its power.
Any prediction worth scoring must be accompanied by a well defined timeframe. Certainly, Grantham’s long-term forecasts meet that standard. A fine organization such as the statistically oriented CXO Advisory Group are well aware of that requirement.
Grantham not only made his more famous longer-term estimates, but he also made shorter-term predictions. The time scale of those predictions ranged from a single month to more than a year. CXO sorted his predictions based on their various time-spans and scored them accordingly. The 40 Grantham forecasts that CXO evaluated were done so consistent with the appropriate timeframes.
I scanned those 40 test items to confirm the timeliness of the prediction/measurement compatibility. They appear to be properly assessed on a time basis. The CXO testing excluded, and therefore did not address, Grantham’s long-term 7 year forecasts.
Dependent on what subject is being addressed, some long-term forecasts are feasible with the likelihood of reasonable accuracy; others are not. The marketplace appears to be in this latter category.
To support my contention, just review the annual checkerboard patterns that have been registered by the various investment categories and are summarized by any Periodic Table of Annual Investment Returns. Here are Links to several examples:
https://investment.prudential.com/util/common/get?file=1D065355D2CC360385257B7D00536F8Ahttps://www.americancentury.com/content/dam/americancentury/ipro/pdfs/flyer/Periodic_Table.pdfThese are just sample tables. The second reference even shows the drastic movements of various bond categories.
Chaos is supreme. I surely do not see any pattern. Category leaders quickly descend to the bottom of the heap. Standard deviations are huge, especially when contrasted against average annual returns. I doubt many folks have the talent and/or the luck to persistently capture this chaotic behavior in any forecasting model.
Super-forecasters do exist, but in very small numbers. Phil Tetlock’s research does establish their existence, although even within this elite group a regression tendency has been observed. Perhaps Jeremy Grantham is a member of this elite group. I hope so; I do like him, but my confidence has been eroded.
Best Wishes.