How effective is Government Spending?
Basically I got this idea from another thread where for some reason or other the New Deal came up and I started looking at the actual numbers and came to a few conclusions.
In that thread I had gone to the fed and grabbed GDP data going back annually to 1929, right when the stock market collapsed. I collected data on investment, federal expenditures and consumption. I threw in the CPI index as a safe control variable to tie down the possiblity of other variables playing a role and came up with some interesting points. First and foremost federal expenditures were nearly a third as effective as private. And given that, federal expenditures tended to "crowd out" investment. These are not revelations by any means, (atleast to the noncommies on the board). What was interesting, again not a revelation, but the analysis of the effect of federal expenditures on consumption and then on investment. The ineffectiveness on consumption is not just consumption growth at a third the rate of investment, It was the double whammy. Those dollars the federal government spends not only produce less given opportunity lost BUT the more effective investment is penalized and lost on top of that.
Alas there were some issues I had with my models. Number one was the problem that my sample size was a bit too small. Typically a good sample has about 200+ observations. Usually you can get away with less and still follow what's called a bell curve to maximize the accuracy, but it's just not a safe rule of thumb. So I had to find more data, which I did. Secondly, Princep of all people, pointed out the fact that my model did not include state expenditures, it may still be missing an important peice of the analysis. My reasoning for that was, the little amounts of spending that the states actually did as compared to the federal government, the overall trend or effect government expenditures had would still be maintained. But for the sake of argument I went ahead and added the values as well. Finally, the biggest issue that I had and I stated at the time was that a more sophisticated analysis could be done by importing time into the equation. Is it possible that government spending get's better over time? That government's quit being suchups? Well let's see... (If you don't want to read all the reasoning and look at all the pretty pictures, just feel free to jump to the conclusion, i'll outline it.
First order of business is to explain the GDP and why consumption matters. We all know GDP means how much an economy makes. It's like Uncle Sam's income. The US GDP is/ was 14 trillion, meaning our annual salary is producing 14 trillion USD. Now in stats we have many statistical problems that can arise. For one the variable may really be wrong and absolutely and utterly uncorrelated even though it displays the value and sign you were looking for. In other cases, a section of variables in and of themselves may work great but together they are useless. And most important of all is what's called collinearity. Variables are supposed to be correlated, that's the whole point of the analysis is to measure that correlation, but sometimes variables can not only be correlated, but directly tied into another variable. For example the equation for GDP is GDP= investment + consumption + government expenditures + net exports. You CAN NOT measure the correlation between GDP and consumption because consumption is a component of GDP. However, you can measure the correlation between Government expenditures and consumption. Consumption by itself is the most important variable in the GDP equation. See exhibit A...
The fed's goal when it comes to "fixing" an economy is to get consumption up. Number one the higher consumption is, expecially in a diversified economy the better off the people are. And number two, it's so fricken large. A 2% increase dwarfs a 2% increase in net exports, government expenditures and investment, or atleast it did dwarf government.
So What I have gone on to do is really measure the effect of government spending/ investment versus private investment. Federal means federal level investment and expenditures, investment means private investment expenditures, and state's means state and local expenditures and investment. Government expenditures represents the sum of state and local AND Federal. All values have been controlled for inflation, seasonally adjusted and finally are in billions. All data comes from the St. Louis Fed Reserve Data Base and stretches quarterly from 1949 to 2008. First up, the true effect of investment and government expenditures on consumption...
So what does this tell us? Much the sam as my first study did in another thread, private investment is more effective than government expenditures and investment, about 1.5 times as effective. Going further I broke down the expenditures, here's for you government fans, and actually found that State and local expenditures are MORE effective than private. They are virtually identical, but the State tends to just a bit edge out the private competition. Impressive I must say, but then I began thinking. What about it that makes the state as effective if not more than investment and much more effective than at the federal level? Part of it probably has to do with economies of scale. This notion that their is an optimal size for an entity to be effective. If you're too small you aint worth
The most interesting point that I first stumbled across, well not really stumbled per se... It's that government spending has a two fold effect. If more dollars are being spent by the federal government, those are dollars that the more effective private investment can't spend. At the same time, those dollars that the federal government are actively diminshing the total spending amount of private investment. Look at the graphs above, for every billion dollars the federal government spends, that's diminshing the ability of the private sector to spend by 485 million. That's a lost 1.11 billion dollars in what consumption could have done. Add to the lost 111 million when comparing private investment and federal expenditures, we're talking federal government losing us 1.2 billion for every billion they spend. And of course we just happened to run up a 7 trillion or so deficit. Ouch!
So to really give you guys an illustrated picture of how much more effective private investment is versus government investmetn and expenditures, I've gone ahead and generated a graph. I've taken the percentage change in consumpton generated solely by federal, state and local expenditures and investment versus pirvate expenditures and investment.
Like I mentioned, I've gone ahead and run the models above, taken the values and created two more sets of data, one incorporating the pure growth rates of consumption with respect to investment and the other with respect to government. The results are pretty convincing. The values while extremely variable (for the math geeks I ran a moving average over 8 intervals, in this case two years) still show that the average return on investment to consumption growth rates is definitely higher than government. For any of the economist, the negative slope of the trend lines actually illustrate the production function and it's diminihing return.
And now to the final part of the analysis, incorporating time. Is what I did above "allowed" given the time nature of the data? Of course it is. It's actually called a right censored OLS. OLS is probably the most effective way to analyze period. It's simple quick and pretty accurate and gives more than adequate answers. However, there are other much more advanced and sophisticated means of analysis, but they all tend to carry with them a higher chance of error. Plus their level of difficulty and "art" required to truly analyze a given problem requires much more talent from the individual. Luckily I have that talent. This model incorporates time. For instance it tells you whether federal expenditures were effective yesterday or today. It's not always the case that the world is static, sometimes elements take a bit of time to react and cause change.
Just a quick warning, these values are not understood the way the other models were because it's time series and I've differenced the values for stochastic reasons. It's more of predictive power indicators versus how much. Think of it ordinal versus cardinal. What we see here is that the return on investment for consumption is not immediate. It's lagged one year according to the yellow graph. But... federal spending is actually immediate, it's spot on. What's more, and I've bolded the important parts, the P value (P>T), displays the significance of the federal expenditures, and it's high enough to be irrelevent. Over time, federal expenditures is not only weak, it probably plays no role at all.
With the other graph, maroon, we see the effects of federal expenditures on investment. And you can see it's negative. Increasing federal expenditures will result in a falling level of investment, and it's VERY significant.
What we can take from the final model is that federal expenditures over the course of 70 some odd years really doesn;t effect consumption, but it does effect investment, the real driver of consumption, negatively.
CONCLUSION:
If you made it this far, congratulations. I'm impressed. When it's all said and done what we can take from this, from data supplied by the federal reserve is that there is plenty of evidence that private investment, the free market if you will, is more effective than the public. We as individuals know where to put our money versus any buearcracy. The states may spend their dollars better than the federal government, but the states like the private investment are also subjected to competitive forces. And it just may be, that this is a way to increase the effectiveness of federal spending. In terms of world trade, ease of travel and low barriers internationally, if people can move freely and effect the choices by something as big and bumbling as the government, maybe the government will make better choices. Inject it with a dosage of freemarket principles, competition. But until then, what we know is that the federal governemnt, and government expenditures are less effective.
edit: I probably should have posted this in the academy
Reply With Quote
Originally Posted by Eastern Prince
