There's no taste for accounting
By Scott Sumner
Modern economics tends to use sophisticated statistical techniques to try to infer causal relationships. Boring old accounting is out of style. I think this is unfortunate.
The most important book on macroeconomic history was Friedman and Schwartz’s Monetary History of the US. The book relies heavily on simply documenting lots of statistics, such as the monetary aggregates, prices and output. And yet it was hugely influential. More recently, Thomas Piketty’s Capital in the 21st Century also relied heavily on documenting accounting-type data.
Matt Rognlie and Kevin Erdmann have done a lot of good work by simply shining a light on overlooked accounting data, such as the fact that much of the recent rise in the share of income going to capital is the implied rent earned on houses.
In my recent book entitled The Midas Paradox, I try to do for the gold/currency ratio what Friedman and Schwartz did for the monetary aggregates—show how a key indicator of the stance of monetary policy behaved during the 1930s. In retrospect, it’s surprising that people had not already done this. I wasn’t the first to recognize the importance of gold demand (Eichengreen, Barro, Glasner, and others had already done so by the late 1980s), but I couldn’t find any data that would be the equivalent for the gold view of the Depression of what Friedman and Schwartz did for the monetary view. (BTW, the gold view is essentially a monetary explanation, if you view money as the medium of account.)
Obviously you need more than accounting to establish causality. I use 3 techniques in my book:
1. Linking the accounting data to well established economic theories, like the supply and demand model (for gold, or for labor).
2. Linking economic shocks with financial market responses (stock prices, commodity prices, bond yields, bond risk spreads, gold prices, spot exchange rates, forward exchange premia, etc.)
3. Linking one macro times series with another, mostly using basic regression analysis to establish correlation. I’m not a fan of Granger causality; I prefer to look for evidence of time varying correlation that is consistent with the underlying theoretical explanation.
The narrative is there to help disentangle the endogenous changes from those that are in some sense exogenous.
PS. The global gold/currency ratio rose by about 9.6% between October 1929 and October 1930. That’s a highly contractionary policy, ceteris paribus. Does that piece of accounting data lead you to re-evaluate your view as to what triggered the Great Depression?
PPS. I have a new piece on kidney markets over at the Washington Post. I am on vacation now, but will resume my usual schedule in 2016.