During the passage of the most recent stimulus bill there was much debate over if the government spending even has the possibility of being effective. This debate continues today, largely along partisan lines, so I was curious if there is any empirical evidence to support the effectiveness of government stimulus.
Clearly there are many factors that affect the economy outside of the scope of the of government spending. One would hope a large enough data set and an effective regression analysis should allow you tease out the impact government spending has even with all other factors considered. While I concede that how the government spent the money might significantly change how it impacts the economy, for this post I am focusing on the aggregate spending of the federal government. This post will analyze the impact spending has on the two broadest measures of economic health: GDP and Unemployment.
My analysis consisted of plotting the inflation adjusted %change in spending by year against GDP and Unemployment respectively. This makes for nice charts but is not that meaningful from a statistical standpoint. For that I did a linear regression (thank you stats class and Excel 2007) and determined the R-squared value and produced a histogram. Essentially the R-squared value tells you have often you can expect your model to produce an accurate result and the associated “Significance F” tells you how confident you can be in your R-squared value.
Historically government spending has almost 0 statistical impact on unemployment and a fairly low predictable impact on GDP. The GDP histogram shows that the line does have a positive slope, meaning that if you increase government spending generally you are going to increase GDP. However based on the model above, if you try to predict how changing government spending is going to impact the economy you will only be correct about 1-5 times. Admittedly this is a crude model and I might refine it in the future to determine if how the government spends money has a stronger correlation to the economic performance. However, based on the model above it doesn’t appear that government spending can directly and consistently produce predictable results in the two broadest economic measures we have.
Meaning that government spending most likely won’t do much to stimulate the economy.
Here are links to my much labored over worksheets, please feel free to play with the number and let me know what you find:
Cleaned Raw Data
Thanks and as always comments are welcome, encouraged and appreciated.
If you are interested in learning more about statistics: