Facebook is super-fun for political arguments. Well, maybe not if you’re not me. Me, I like to use social networking tools to have vigorous, vitriolic disagreements, apparently even in the comments section of posts made by friends from college whom I’ve barely seen since. Which occasionally leads to interesting questions. Earlier this week, a friend said the following: “Keeping the Democrats out will at least lead to less growth in government spending.” I like this assertion, because it’s totally testable. Let’s test it!
I had so much fun with the “are high taxes better for California businesses?” blog entry that I thought I’d do the same here. Now, the first step was to turn that statement into something we could test: that is, the rate of change of government spending should be lower when the Republicans are in charge than when the Democrats are in charge. To me, it also made sense to look at the deficit: you could set spending to $100 but taxes to $0 and still run a deficit, despite that low spending.
I grabbed some data from FRED and set to it. They have spending, receipts, and GDP information going back to 1951 on a quarterly basis. (Their earlier data isn’t quite the same, so I just started in 1951. Nearly 60 years seems long enough to me.) I started by taking spending as a percentage of GDP, in order to make things somewhat comparable year-on-year as the economy grows and the value of a dollar decreases.
This does, however, introduce a big problem: spending as a percentage of GDP spikes whenever there’s a recession. Why? Well, to get spending as a percentage of GDP, you just take (spending)/(GDP). In a recession, GDP — the denominator — shrinks. So, even if spending stays the same, that will result in an increase of the final value. (Just like, if you have 2/4ths, and subtract one from the bottom, you get 2/3rds, which is larger!) I can’t think of a trivial way to work around this, but it’s worth being aware of.
Making Data Look Funny
Yeah, that goes uphill. We could all figure that out! What we really want to know is: when do they try to make it less, and when do they try to make it more? That will answer the question at the beginning of this entry.
So that means we don’t so much want to know the level of spending in a given period — we want to know the change — that is, is the spending lower or higher than the previous period? To get that piece of information, I took the slope of the line each quarter. If the slope’s going up, then things are getting more. If it’s going down, then things are getting less.
The dataset did look very erratic, so I took a moving average to decrease the size of the peaks and valleys while still keeping the same overall data profile. The data are quarterly, and so may bounce up and down within a year as a big expenditure hits or as taxes come in — timing effects that may not actually reflect how responsible (or not) the budgeting is over the full year for which the budget is approved. In the graphic below, you can see the un-smoothed data in blue, the smoothed in pink.
When you look at this graph, values above 0 mean an increase, values below 0 mean a decrease. It’s not the case that a downward slope means that spending is decreasing — a downward sloping line with values over 0 means spending’s getting larger slower. An upward sloping line with values under 0 means spending’s getting smaller slower.
Putting That in Context
Here’s the same data overlaid on charts that show if we had a Republican or Democrat in the White House and also which party controlled the House of Representatives. (Spending legislation originates in the House, so I looked only at these two entities here.)
As you can see it’s really all over the map. It’s not obvious that there would be any relationship here. We’ll look into that further in a minute.
Now, it’s not just about spending — it’s about the size of the deficit, too, or so we’ve heard so much over the past few months. We can do the same sort of graph for the deficit here. (The deficit is calculated on a quarterly, not cumulative, basis here, so if it’s over 0 then they’re adding to the deficit in that quarter, if it’s under then they’re subtracting from it.)
Again, there are large fluctuations in both directions. Are there any trends? Time for some statistics.
So the question is: do you do better if you pick a Republican House or President at controlling spending or the deficit than you do if you pick Democrats for those positions?
I looked at this by doing a series of t-tests. A t-test basically says: we have two samples we’ve pulled from the world. Are these two samples part of the same larger group, or are they drawn from different groups? If Republicans are better at controlling spending or the deficit than the other party is, or vice versa, the t-test will show that the deficit or spending levels each year are drawn from different groups, depending on who’s in power. If not, then it’ll all look like one big, equivalent level of spending, regardless of who’s in power. Guess which way it looks?
If you guessed that there are no statistically significant differences, you’d be right! I looked at these numbers both from 1951 on and starting in 1980, when Reagan was elected and the modern Republican Party came into being and: there’s no statistical difference in either period. (I tried to look at before 1951 or just with split-party government, but there were too few data points for Republicans controlling the House in the former and the latter overall.)
What does this mean? This means that if you vote for one party or the other in the belief that this party will lower spending or the deficit (or raise it, for that matter!), you’re most likely to be disappointed.
However, there were some trends. This chart shows these trends. Again, the percentages along the side are the percentage change quarter-to-quarter:
Also again, none of these trends are statistically significant, so you can’t count on anything repeating itself if you vote one way or the other — but it’s interesting to note that it’s Democratic Presidents and Republican Houses that lower the deficit. We can see that for Democratic Houses and Presidents, there’s a clear effect of taxes, keeping deficits from growing as fast as expenditures do. The opposite is true of a Republican House: they cut taxes enough that their spending cuts don’t impact the deficit as much as they could. Interestingly, Republican Presidents look pretty much like Democrats.
All of these differences are small. Two-tenths of a percent down? Three-and-a-half-tenths of a percent up? It hardly makes a difference over the long term. Small mean effect sizes plus wide variance here means, again, that there’s no statistically significant difference between Democratic and Republican policies when it comes to spending and the deficit — you’re drawing from the same set of ideas, either way, and getting the same end results. Which party spends more? They both spend a ton.