With Friday’s release of real GDP we can now add another year to this interesting graph. Each dot represents one of the 57 years between 1955 and 2011. The x-axis shows the annual percent change in real GDP for that year and the y-axis is the percent change in new vehicle sales. Naturally, there is strong correlation.
Two outliers are noted in the chart above. At the far right, close to the horizontal axis, we see that in 1966 new vehicle sales declined 2.1% even though real GDP grew by a very strong 6.5%. Why’s that? It’s because new vehicle sales had already increased dramatically in each of the prior four years. (New vehicle sales first exceeded the ten million mark in 1965.)
As an opposite case note that in 1982 new vehicle sales declined by only 1.9% even though real GDP dropped by 1.9%. (Up until our recent recession, that was the largest decline in real GDP in post World War II history.) The disconnect is because new vehicle sale had already fallen significantly in 1979, 1980, and 1981, producing a peak to trough swing of 31%.
By just ignoring those two outliers, the R-square increases from .63 to .71 for the remaining 55 years plotted. If we also ignore 2011, the R-square rises to .74. As noted in the chart, 2011 was only time that an increase in real GDP of less than 2% was associated with a rise in new vehicle sales.
Strong cycles produce outliers, and after an outlier there is a movement closer to the regression line. That will occur again this year, although I fully expect that the plot for 2012 will remain well above the regression line.
Also note that, in this comparison, 2009’s plunge in new vehicle sales (-21%) was not abnormal given the 3.5% decline in real GDP. Indeed, it was almost right on the regression line.
Comments