Tuesday, November 12, 2013

Shifts in Relative Prices vs Changes in the Price Level - A Preliminary Analysis Employing the Sauerbeck-Statist Index

I’ve recently grown interested in testing the predictive value of changes in relative prices as opposed to changes in the price level. So when I found myself with a few extra hours the other day, I ran some preliminary tests using the Saurbeck-Statist Index.

The Sauerbeck-Statist Index was published yearly from 1846 to 1948. It employs several baskets of commodities that are merged to form the final index. These are divided into two groups. The first group, “food”, includes baskets “vegetable”, “animal”, “sugar, tea, and coffee.” The second group, “raw materials”, includes “minerals”, “textile fibres”, and “sundrie”. In a set of regressions with changes in Great Britain’s GDP between 1878 and 1938, 1) I test the explanatory power of the sum of the absolute values of year-to-year percent changes in the price of these baskets, 2) of the sum of the absolute values of year-to-year percent changes less the percent change in the overall index, and 3) the absolute value of the percent change in the overall index. I anticipated that either 1) or 2) would provide more insight into changes in GDP, but this was incorrect. Regression 1) predicted about as well as 3), and 2) was far less accurate than both. (I also ran regression with squared variables, none of which proved more accurate and none of which are shown here.) The results are pictured below:

dlngbgdp = Percent Change in Great Britain GDP
APV = Absolute Price Volatility
RPV = Relative Price Volatility
absdSSI = Absolute Value of Year-to-Year Percent Changes in the Sauerbeck-Statist Index
wwi = Dummy Vairable 1915-1917

The signs on each measure of changes in price are negative as expected, so nothing appears unreasonable above. Also, the coefficient for changes in the Sauerbeck-Statist Index is about six times as large as the coefficient for the measure of Absolute Price Volatility. This is not surprising as there are 6 sectors considered in the Sauerbeck-Statist Index. The results appear consistent.

This is in no way a final say on the matter. There are certainly problems with this method of testing the impact of relative price changes.Summing changes in separate indices is not the same as summing changes in all prices as weighted according to their value relative to GDP. This also does not attempt in any way to separate output according to their location in the structure of production. Still this is worth considering because it does show the sum of changes in the value of certain baskets in different sectors - a proxy for price volatility - changes that surely would encourage economic discoordination.

In the near future, I would like to attempt a much more thorough, and probably more modern for the sake of convenience, analysis of changes in relative prices. For now, these preliminary results do not suggest that the sum of such changes may not be are a better predictor than changes in the overall price level. This is no surprise as rational producers will substitute away from inputs that become more expensive and consumers will substitute away from goods that do the same.

No comments:

Post a Comment