13 April 2024

oops

One of the base themes of these missives: thou shalt not estimate beyond the range of the data set. Of course, the corollary is also mentioned in stride: time series analists do so as a matter of course. Getting the future right based on past experience is fraught with danger. If you're in the investing venue, you're told, as a matter of course, "past experience is not predictive of future returns" in various words. And, of course once again, the monetary authorities (esp. the Big Economies) use just such 'modeling' to set policy in such a way as to extrapolate the past data experience into the future data experience in the way that implements the desired policy objectives. Keeping employment 'high' and inflation 'low' at one and the same time is the authorities' walk down the razor's edge. It should come as no surprise that the authorities reflexively adopt policies their constituents (aka, Big Bidnezz) prefer irregardless of what the data says.

Some times it just goes sideways. Well, here is the tale of the Bank of England. It's sordid. And, highlights not only 'past is not necessarily future' theme, but (not explicitly, alas) that macro data is almost wholly sample data, and of unknown quality. Now, governments could, with simple legislation, enforce data collection of interest in signficantly larger, more stratified, samples. Or go Full Monty with census data. Doesn't happen, for the most part (there are, in the USofA, 'census' of some sectors). The reason that doesn't happen isn't cost, but rather that The Powers That Be at any time (whether left or right wingnut) want the 'flexibility' to implement some policy with the excuse that more granular data just isn't conclusive or up-to-date. The prime fig leaf for liars.
[The Bank] added that it would need to put in "substantial investment" to develop the data, modeling and staff to support the forecasts. The changes will take a while to put in place, and the bank will provide an update on its progress before the end of the year.
This is the report of the Review. Further, if you feel up to it, I don't right now, here's the Review.

No comments: