It's All About the Economy, Stupid

That sentiment, coined by James Carville for Bill Clinton’s successful 1992 presidential campaign, is just as true today as it was then and likely has always been.  Go figure, people care about the economy because it affects our everyday lives.  From filling our gas tanks to putting food on our table to receiving a paycheck from our jobs, nearly every aspect of our daily lives are affected in one way or another by the state of the economy.

Having said that, the effect the economy has is nebulous and somewhat rarely are we actually able to see the smoking gun.  The perfect example of said “smoking gun” are oil and gas prices.  Given the reserves that we have within our borders as well as some of the governments we have to deal with to obtain crude oil as well as its crucial role in our economy, the price of crude oil is very commonly reported and often politicized.

A more difficult effect to wrap your mind around is the salary you earn at your job.  If you earned $50,000 last year and $51,000 this year, it’s very simple to conclude that your salary went up 2% and that despite issues elsewhere in the economy, you are doing pretty well.  Well, let’s say that we are in the midst of mild inflation and the Consumer Price Index (CPI) rises from 220 to 230.  That would mean that, adjusting for inflation, your income fell from $50,000 to $48,782.61, a drop of 2.4%.

Often times, when looking at the economy, it’s easy to get sucked in by the allure of numbers that have no context.  If our government has a $1.4 trillion dollar deficit and a presidential candidate says that he will reduce the deficit by $1 trillion, the obvious assumption is that if elected, that candidate will reduce the deficit from $1.4 trillion to $400 billion (which, by the way, would likely qualify as a miracle).  However, what if that reduction is spread out over 10 years?  Now, all else being equal, that candidate has pledged to put us on a track where the deficit will be $1.3 trillion for the next ten years instead of $1.4 trillion.  While saving a trillion dollars (especially when a government frequently runs deficits) is a very good thing, the context is critical, especially when listening to politicians talk about the economy.

Which brings us to the crux of this particular project; which party is better for the economy?  Is there a relationship between which party controls the three major parts of our government (for this discussion, the House of Representatives, the Senate, and the Presidency) and how well the economy does over that time period?

HYPOTHESIS

Since this is my first quasi-research project, I’ll make a hypothesis; there will be a relationship, it won’t be what you expect, and I don’t believe it will end up being all that significant (although I do predict it will be statistically significant) because of the litany of asterisks that will be required.

ECONOMIC THEORY

In principle, the economy of the United States is a free market capitalist model.  Two main things promote efficiency in free markets; number of firms/buyers and their ability to get to the market.  Anything that reduces the number of buyers or sellers in a market (below a certain point) and anything that hinder the two of them meeting in the marketplace has the potential to reduce the efficiency of the market.

Historically speaking, the single largest hindrance to American economic efficiency is… the American government.  Just about the only thing that the government does to increase economic efficiency is levying taxes (or some other instrument) to account for negative externalities, or negative side effects of a process that are not accounted for in the price of a good.  The perfect example of a negative externality is pollution; it’s a byproduct of the production of a good but because the good is distributed to people not affected by the pollution that does not come into the price of the good.  The government can step in and correct this imbalance.

However, many other things the government does (such as other taxes) lower economic efficiency and this makes far more sense when we consider that that is not their goal.  You could easily argue that given our current distribution of government funds and redistribution of wealth, our nation is more concerned with social equality than economic efficiency.  Whether that is right or wrong is a discussion for another time.

This is where party identification comes into it; Republicans typically favor smaller governments that spend less and interfere in the markets less.  All of these (in a vacuum) promote economic efficiency.  Therefore, logic would dictate that during congressional sessions where Republicans have more power than Democrats, the economy should perform better.

THE DATA

The next question that comes up is how do we measure the performance of the economy?  Without that, we can’t even begin to determine whether or not a party’s position in congress has anything to do with a successful economy or not.  There are two main measures that have been used (right or wrong) as measures of the health of our economy that have data available for several decades; GDP and the Dow Jones Industrial Average.

GDP is one of the solid go-to’s when it comes to economic analysis.  Logic dictates that when it comes to an economy, the more you’re producing and the more you’re consuming, the healthier the economy is.  Intuitively, this makes sense.  On a micro scale, if you have less money, you buy less and if you have more money, you buy more.  The more financially healthy you are, the more you consume (for the average consumer).  The same logic applies here, only we’re applying it to a country instead of to your wallet.

 

A lot is going on in this chart so allow me to explain it.  The blue line is GDP (in billions of dollars).  The red line is GDP that has been adjusted for inflation using the Consumer Price Index (CPI).  The green line is one that I will be using in my regression that comes up later; real GDP per capita.  One of the issues when using GDP to gauge the production of an economy over time is that population is a very important factor.  If a country has 100 people and 200 units of output (GDP) and then ten years later has 500 people and 300 units of output, can we say that the economy has done well?  The GDP per capita has gone from 2.0 to 0.6, a huge reduction in the standard of living for that country (unless there has been serious deflation issues with their currency… either way, it doesn’t look good).

The early results, just by looking at this graph is that the United States has done exceedingly well in the 20^th and 21^st centuries.  In 1929 the GDP was $104.6 billion and after adjusting for inflation and normalizing by the population, that figure was $9.75.  Over the course of the past 85 years, those figures have jumped to $17.5 trillion and $51.50, respectively.

The other measure I mentioned was the Dow Jones Industrial Average (DJIA).  This is very tricky because of the way people have begun to use the stock market to get rich instead of looking at it as a vehicle for long-term investments.  Nevertheless, I believe it warrants inclusion because of a self-fulfilling prophecy.  If the stock market goes up, people believe the economy is doing better.  This will cause them to go out and consume more goods and services which will make the economy better.  Mass psychology plays a huge role in the stock market but then again, when our consumers are real people and not data points in a program, mass psychology has a role to play in the economy at large as well.

Again, adjusting for inflation seemed like such a good idea that I did it again here.  My first instinct, when I look at this chart, is from the 1920’s through about 1990, I feel like the stock market was used as a long-term vessel for investment.  There were some shocks (obviously, that time range includes the Great Depression) but for the most part, the ups and downs could likely be explained by the business cycle and the economic picture of the time.  That is simply not the case anymore.  With the proliferation of computers and this whole internet thing, it has become easier and easier for average people to put their money into the stock market and execute their own transactions.  This opening of the markets to the masses has partially led to the huge swings that can be seen in the 2000’s and 2010’s.  Along with their being more money in the market, the average know-how of a person executing these trades has gone down.  The experts have been diluted by the average people looking to get rich on their own.

Thankfully, as can be seen from these graphs, data is available for GDP, the DJIA, and the CPI from 1928 to 2014.  I have to be honest, when I think about the amount of data that is available to anyone and everyone, I kinda geek out a little bit… but that’s just me.  Anyways…

CONGRESS

The next step in this analysis was straightforward.  I’ve already identified my economic indicators and gotten them into exceedingly usable forms.  Next, I have to line up some more independent variables; the makeup of Congress.

Thankfully, this is actually an even easier task given that we have records of just about every Congress that has ever served.  I readily found data going back to the mid-1800’s and personally, I always like it when I find more data than I need.

From 1927 to 1929 (the 70^th Congress), there were 96 senators and 435 representatives.  This was very close to our current makeup but to account for minor fluctuations in the size of Congress over the past eight or nine decades, I decided to simply take the percentage of total members that were either Democrats or Republicans.  For instance, in the 77^th Congress (1941-1943), there were 66 Democrats and 28 Republicans in the Senate and 267 Democrats and 162 Republicans in the House.  These figures became 68.8% Democrats in the Senate and 61.4% in the House and then 29.2% and 37.2%, respectively, for the Republicans.  Those two years, there were a total of 8 members (2 senators and 8 representatives) from other parties but I don’t take them into account given their tiny minority.

From there, I simply created a number of dummy variables to describe who controlled which aspects of the government, including whether or not a single party controlled both houses of Congress, whether or not a single party held all three (House, Senate, and White House), and still another to signify whether or not Congress was divided.  Interestingly enough, out of the 44 Congressional sessions in the sample, there have only been 6 instances of a divided Congress.

During Reagan’s first six years in office, he dealt with a Democratic House and a Republican Senate.  In George W. Bush’s last two years in office, he had a Democratic House and a split Senate.  However, Republicans and Democrats had 49 seats a piece so even with a Republican vice-president (who would be the deciding vote in the event of a tie), I couldn’t give either party the edge in that case.  Lastly, in the past two Congresses, President Obama has had a Republican House and a Democratic Senate.  In other words, in all six instances of a divided Congress, the Senate and the White House shared a party.  There is your interesting tidbit for today.

Lastly, I included dummy variables for several key features of economic history that occurred in the past eighty years.  For instance, I included a dummy variable to indicate when Congress serves in a time of open war.  Rather than getting too bogged down in what should be defined as a war, I accounted for the major conflicts that happened during the time frame; World War II, the Korean War, and the Vietnam War.  I also included a dummy variable to indicate the periods of time encompassed by the Great Depression and the Great Recession.  Lastly, I included another variable for the 107^th Congress, which served from 2001 to 2003.  This was to account for both the dot com bust as well as the 9/11 attacks, both of which had serious repercussions on the economy.

The purpose of these dummy variables is simple; they attempt to account for something in the model that other variables do not account for.  In other words, by including a variable that indicates the nation was going through the Great Depression, I’m attempting to determine if the poor economic performance was due to the makeup of Congress or due to the Great Depression.

RESULTS

The largest problem with this analysis was simply a matter of sample size.  There were 43 Congresses in my data set and while 86 years (technically 85 years; my data does not go through the end of the current congressional session, perhaps because it has not yet ended) might feel like plenty of time to draw statistically significant conclusions, at this level of analysis, that is not the case.

Below are the output statistics from an OLS regression in Gretl.

Model 3: OLS, using observations 1-43
Dependent variable: GDP_Capita___Ch
Heteroskedasticity-robust standard errors, variant HC1
Omitted due to exact collinearity: Rep_Pres Div_Con_
	Coefficient	Std. Error	t-ratio	p-value
Constant Term	90.7396	337.831	0.2686	0.7901
Democratic % (House)	0.839877	3.5372	0.2374	0.8139
Republican % (House)	0.478972	3.61369	0.1325	0.8954
Democratic % (Senate)	-1.48576	1.43847	-1.033	0.3099
Republican % (Senate)	-1.59213	1.61083	-0.9884	0.3309
Democratic President	-1.16063	8.48723	-0.1367	0.8921
Democratic Congress	-6.16873	4.80762	-1.283	0.2093
Democratic Government	2.71519	9.96121	0.2726	0.787
Republican Congress	4.77725	4.92264	0.9705	0.3396
Republican Government	-6.32027	9.30869	-0.679	0.5024
Depression or Recession	-15.1617	4.44234	-3.413	0.0019
War	7.27271	4.45552	1.632	0.1131
Other	0.774011	2.95253	0.2622	0.795
Mean dependent var	4.33168	S.D. dependent var	9.049915
Sum squared resid	1704.174	S.E. of regression	7.536962
R-squared	0.504578	Adjusted R-squared	0.306409
F(12, 30)	33.53435	P-value(F)	4.60E-14
Log-likelihood	-140.1265	Akaike criterion	306.253
Schwarz criterion	329.1486	Hannan-Quinn	314.6962
Excluding the constant, p-value was highest for variable 4 (House_Rep)

This model says volumes about the matter in question while simultaneously saying relatively little.  The key figures to look at are the p-values in the right hand column and this is a situation where lower is better.  In fact, to the majority of statisticians, a correlation isn’t statistically significant unless the p-value is less than 0.1.  By that standard, the only variable in the regression that is statistically significant is the one indicating the presence of a recession or depression.

From this model, it would appear that the makeup of Congress is not an indicator of economic success or failure, according to the changes in GDP per capita.  A model substituting the performance of the DJIA as the dependent variable yielded similarly poor results.

So what are we to make of these results?

My take is that the economy is complicated and is affected by an unbelievable variety of factors, just one of which is the makeup of Congress and the laws they enact.  To that end, you can also notice in the table above that the adjusted r^2 value for this regression is 0.306.  That means that just over 30% of the variability in the dependent variable (change in real GDP per capita) was due to the variability in all of the other variables.  In other words, after crunching these numbers and finding this data, I was able to explain just 30% of what’s going on.

As a professor of mine would say, don’t get too caught up in that pesky r^2 value; it’s not the be all and end all.  If we tried to create a model that entirely explained the changes in real GDP per capita for the past 86 years, it would require a supercomputer to execute… my poor little laptop would just not be up to the task.  The more important thing to look at is whether or not any of the independent variables were statistically significant; that is where the real meaning of this regression analysis comes into play.  Unfortunately, none of the variables that I wanted to be significant were; I am forced to reject my hypothesis.

REFLECTIONS

Then again, I’m hardly surprised by this result.  Our economy is enormous and even with the government spending truly obscene amounts of money during the past 4-6 years, government spending still only accounts for 20-25% of the size of our economy.  Even though they have enormous power to enact laws that directly affect the economy, it appears (at least from this analysis) that they are far more bystanders than they and the media would have us believe.

After all, was it Hoover’s fault that the DJIA dropped 43.5% from 1929 to 1931 and then another 55.3% from 1931 to 1933?  He was faced with one of the worst economic crises the modern world has ever seen and he was doing exactly what classical economics said that he should.

Was it the fault of George W. Bush in particular and Republicans in general that the dot com bubble burst after he took office and then the 9/11 attacks rocked the country?  Is it the fault of either party that the banking industry started the largest economic downturn since the Great Depression?  From the little reading I’ve done, that particular crisis can be attributed to both parties over the course of fifteen or twenty years of deregulation of the banking industry.

Just because a president was in office during good or bad economic times does not mean that they caused those good or bad times.  Presidents get far too much credit when the economy is booming (Bill Clinton, for instance) and far too much blame when the economy is stagnant (George W. Bush).  Statistically speaking, the same can be said for the occupants of Congress.

Essentially, at the end of the day, the picture on the right is flat out wrong.