Keynesian theory tells us that, when economic activity falls, government expenditures should rise. That is, government expenditures should be countercyclical and lean against the business cycle. Has this happened in the U.S.? In the graph, the red line shows growth of government expenditures and the blue line shows growth of private (nongovernmental) economic activity. And it looks like when one line is high the other is low. Does this mean government expenditures are countercyclical?

Actually, this is partly an optical illusion: On average, government expenses have grown more slowly than the rest of the economy, and thus the red line is more often low and the blue line is more often high.

A better way to examine this question is with a scatter plot, shown below. Each axis represents one indicator, and each dot corresponds to a quarterly data point. If government expenses were countercyclical, the cloud of dots would have a somewhat negative slope, with more dots in the top left and bottom right quadrants than the top right and bottom left. The scatter plot actually has a congregation in the middle, which shows there’s little correlation between the public and private sectors of the economy, at least in terms of expenditures.

How these graphs were created: For the top graph: From the Domestic Income and Product release, select “Real Gross Domestic Product, Chained Dollars” and then “Quarterly”; select the GDP and government expenses series; click “Add to Graph.” In the “Edit Graph” panel, add to line 1 (GDP) the government expenditure series again and apply the formula a-b. Use “Percentage Change from Year Ago” as units for the lines and use the slider to start the data in 1953:Q2 (which is after the explosion of government expenses for the Korean War) to avoid distorting the view. For the bottom graph: Use the top graph, but use the format tab to select graph type “Scatter.”

Suggested by Christian Zimmermann.