The Failure of Keynesian Criticism
Today realclearmarkets.com linked to this piece by Scott Grannis, the Calafia Beach Pundit. In it, he prominently shows a graph plotting federal government spending as a percentage of GDP against the unemployment rate, then argues that unemployment is directly caused by government spending.
Grannis appears to show understanding of the irrefutable arguments against his position, then says “So I want to be careful with the causation/correlation argument here.” But then he walks all over his brief foray into the realm of common sense for the rest of the piece, concluding (with emphasis but not evidence): “Since more government spending has hurt the economy, less government spending should help the economy.”
Let me put this bluntly: nobody believes that government deficit spending causes the economy to contract in the here and now (well, almost nobody, I guess).
Taxation reduces private consumption and saving (investment). Government borrowing tends to crowd out private investment by causing higher interest rates, which should result in lower future GDP growth. In today’s economy, not so much.
Government tax-derived spending tends to be less efficient than private sector spending; it causes less economic activity than would have occurred if that money was spent by the private sector instead. This efficiency can be characterized by the multiplier; a dollar spent here is then spent there, and so on and so on and so on. The multiplier for government spending tends to be lower than that for private sector spending. Debt-derived spending is different, however, since it takes money that wasn’t being used for direct consumption, but rather money that was being saved. This gets complicated fast because of the interrelationship between investment and consumption, but the end result of debt-derived government spending is that it borrows spending from the future and injects it into the economy now.
But what Grannis is arguing is not that government spending is not just less efficient, but that the government multiplier is negative. That government deficit spending is causing GDP to decline and unemployment to increase.
Consider the mechanism. The government borrowed money from the private sector, causing interest rates to… do nothing at all. The cost of borrowing money and the return on investment changes from 50+ year lows by… zero. Due to (no) changes in incentives, private sector investment reacts by… continuing at the same level. So how much investment was crowded out? Meanwhile, the government spends the money on teachers and soldiers and high-speed rail construction workers, all of which are positive to GDP and employment (even if it’s not all wise spending).
I’ve seen studies arguing that the government multiplier is near zero (which means that each dollar of government spending causes almost a dollar less private sector GDP), but negative? Giggle.
Grannis says that most of the increase in government spending over the last three years was secondarily safety net spending, and primarily additional stimulus spending. “And,” he writes, “it is clear that this virtually unprecedented spending boost coincided with the biggest and fastest rise in the unemployment rate, and the deepest recession and slowest recovery in many decades.” 1.: no causation claimed. 2: let’s see, last claim first.
The unemployment rate for January 2009, when Obama entered office. was 7.8%. The American Recovery and Reinvestment Act (ARRA) was signed in February, when the unemployment rate was 8.2%. Here’s what happened for the rest of the year
|During||Stimulus Spending||Ending Unemployment Rate|
Simple, right? Since unemployment increased with stimulus spending, there must be causation! Stimulus causes unemployment!
Simple, alright. Let’s see some analysis? Certainly the Congressional Budget Office hasn’t agreed; in its quarterly report of January 13, 2010, CBO stated “Following implementation of the ARRA, the trajectory of the economy changed materially toward moderating output decline and job loss… decomposition of the GDP and employment change by component or sector suggests that the ARRA has played a key role in this change of trajectory.”
Looking at the most recent such report, CBO states that on average in 2010, the impacts of the stimulus were: higher real GDP by 1.5-4.2%; lower unemployment by 0.7-1.8%. That’s between 1.3 million and 3.3 million more people employed, and many more people moving from part-time to full-time.
In other words, the stimulus had a positive impact. As much as expected? No – but CBO says “that outcome reflects greater-than-projected weakness in the underlying economy rather than lower-than-expected effects of ARRA.”
But before we yell silly names at those few people who believe government spending leads directly to increased unemployment, let’s look more closely at the argument.
“The big rise in spending as a % of GDP in the past three years was mostly driven by a forced increase in spending, and only partially by the fact that automatic stabilizers (e.g., unemployment insurance, food stamps) kick in as the economy weakens.”
Well let’s see.
Using the 2011 budget data from omb.gov, federal spending is projected at $3.82 trillion for FY11, up $836.3B (28%) from 2008. First, the biggies:
- National defense and veterans benefits: +$208.9 billion (29.8%)
- Social Security: +$131.3 billion (21.3%)
- Medicare: +$103.6 billion (26.5%)
These three cover 53% of the entire expansion of government spending. The increase in defense spending was certainly discretionary, but I don’t think much of it was envisioned as economic stimulus, and most folks making arguments like Grannis would never dare touch military spending. Of Social Security and Medicare, some of the increase is due to people needing to start drawing on these programs prematurely, but more of it is simply due to an aging population and increasing medical costs.
Now let’s look at some other safety net spending:
- Health care services (Medicaid): +$99.3 billion (40.1%)
- Unemployment compensation: +$89.5 billion (197.4%)
- Food and nutrition assistance: +$46.6 billion (76.7%)
- Housing assistance: +$28.8 billion (71%)
That’s another 31.6% of the total increase, purely on safety net spending related to the recession. Grannis might characterize some of the increase in unemployment compensation as discretionary, but nonetheless it is in response to persistently high unemployment, not a cause of it.
We’re at about 85% of the total increase of the last three years. So much for the idea of a majority of the last three years’ increased spending being “forced” discretionary increases. Unless you count that $200B+ of military and veterans benefits in this category.
“At the very least this is prima facie evidence that Keynesian pump priming doesn’t work.”
Grannis is partially right; at most it is evidence at first glance. Any deeper analysis, and I do mean ANY, reveals it to be nothing of the sort.
“[I]t’s potentially strong evidence that a big dose of pump priming not only doesn’t work, it makes things worse.”
Except that I haven’t seen any evidence — certainly none in this story — demonstrating this. There are better arguments here against the conclusion than for it.
“Furthermore, it’s evidence that a significant reduction in government spending as a % of GDP does not prevent a significant strengthening of the economy: consider the 1993-2000 period in the above chart, when both spending and the unemployment rate experienced significant declines.”
Grannis continues to put the cart before the horse. The economy grows, causing increased employment and lower spending as a percentage of GDP. He himself states the reasons why it’s the unemployment rate (both in itself and as a proxy for economic growth) that drives government spending, and yet he can’t seem to let go of the absurd notion that government spending is the cause of the correlation.
Grannis wrote an article where the balance of his arguments support the opposite of his conclusion.
Who at Forbes is selecting these things?