Saturday, October 24, 2020

"How reliable is a millennium of macroeconomic data?"

There are insights into the past that can only be realized by immersing oneself into the available information. Looking at a time-series of wool prices you eventually get the fact that "hey wool was pretty important", enough so that people took the time to note valuation and unit volume. It's at that point that you start to get it. Otherwise it's just stats and regressions and presentation.

In stocks the granddaddy of stats is the "The Cowles Commission's Common Stock Indexes 1871-1937."

Cowles only went back to 1871 for a few reasons:

1) The paucity of industrial companies. Among the early publicly traded equities were banks, canals, wooden turnpikes and not much else. Small sample size.

2) Survivorship bias that makes returns look better than they actually were by not counting those companies whose stocks stopped trading.

3) In an era when dividends comprised the bulk of returns available to investors, the corporate records required to quantify that variable are even sketchier than publicly available information on share prices.

Still though, keeping the records is worthwhile and as noted in "Does Stock-Market Data Really Go Back 200 Years?":

Add the current keeper of the Cowles Commission, Yale's Robert Shiller to the laureate list.

And from FT Alphaville, a piece by Claire Jones that elicited a higher level of discourse among the commentariat than we've seen at other venues. Even Tim Worstall was civil, something that is not always the case.

How reliable is a millennium of macroeconomic data? 

Can we really measure what GDP was when the Magna Carta was signed? And do we really need to care?

FT Alphaville is a fan of economic history: not only because it often makes for fascinating reading, but because it also plays an essential role in highlighting how changes in the political, social and technological environment have impacted growth — as well as economic thought and policymaking.

So we’ve noticed with some interest that numerous official and private sector economists are increasingly turning to data sets based on very long economic time series. However, few can match the Bank of England’s, which pulls together almost a millennium’s-worth of economic data going all the way back to 1086. How useful is this information? Well we’d start by making the point that something doesn’t have to be relatable to the present day in order to be worth knowing. Sometimes it’s as simple as things being interesting in and of themselves...


Where Claire and the BofE go macro, we tend to get a bit more micro  If interested here's Mr. Cowles' (and Doc Shiller's) gift:

 Common Stock Indexes...
....My favorite tidbit is the listing, among the pre-1871 industrials, of New York Guano.
Some things never change.
It leads to a big ‘ol hog of a PDF.
And back to the really long series, from a 2010 post,  "The 2012 sunspot sell-off":

...You may want to dip into the big daddy of price series:

"A History Of Agriculture And Prices In England, From The Year After The Oxford Parliament (1259) To The Commencement Of The Continental War (1793)"
by J. E. Thorold‐Rogers, 7 volumes, 1866-1887 which probably influenced Jevons.

Here's another bit o'price series scholarship:

The paper constructs an annual price series for English net agricultural output in
the years 1200-1914 using 26 component series: wheat, barley, oats, rye, peas,
beans, potatoes, hops, straw, mustard seed, saffron, hay, beef, mutton, pork,
bacon, tallow, eggs, milk, cheese, butter, wool, firewood, timber, cider, and
honey. I also construct sub-series for arable, pasture and wood products. The
main innovation is in using a consistent method to form series from existing
published sources. But fresh archival data is also incorporated. The implications
of the movements of these series for agrarian history are explored.