Over at his blog, Eric Rauchway takes note of my attempt to summarize the Great Depression debate in one picture. He asked me what years I used to make the trend. The answer is I used the entire 20th century to estimate the fitted linear trend. The figure below shows log real GDP and trend log real GDP for this time. This figure also make clear why constructing the trend this way is a reasonable approach: it shows a persistent pattern of growth for the entire period, except for the 1930s. (Click on figure to enlarge.)
Eric also notes that my original graph only speaks to one of the three Rs--relief, recovery, and reform-- of of the the New Deal, the recovery. I am not sure there is a good way to summarize the other Rs, but the figure below which similarly graphs per capita log real GDP and its trend may provide some perspective on the relief front. (Click on figure to enlarge.)
All data comes from the historical GDP database at EH.Net
On the what data set is best, don't know much about the consistency of these data over time, but generally convergence depends upon the span of the data, so the longer the time series the better.
ReplyDeleteYou probably use the same story, or something like it - it's like measuring the trend temperature for global warming - do you want to try to do that with just a few decades of data, measured, say, by the minute so there is a ton of data, or would you rather have annual measurements covering thousands and thousands of years (less data points, but almost infinitely greater span)?