Keynesian Stimulus and Japan's Lost Decade

Two historical examples of fiscal policies that Keynesians like to point out are the US in the 30s and Japan in the 90s.  I have already looked at the 30s, and the numbers did not suggest a Keynesian conclusion to me.  I am speaking here of the Keynesian theory of stimulus spending. 

Today I will look at the numbers for Japan in the lost decade, supposed to be almost exactly the 1990s.

The chart above shows Japan's government consumption, both in absolute terms and as a percentage of the economy.  All numbers are from stats.oecd.org.  The prime minister at the time purported to be Keynesian, and as the cart shows, they did increase spending significantly throughout the decade.  Total spending, in inflation-adjusted Yen, grew by about 50% during the decade.  As a percentage of the economy, government grew from 31% to 39.5%, spiking at 43% during the 1997-98 Asian crisis.  I am sure some Keynesians will say that no single year constitute a model Keynesian stimulus, but I have heard some of them claim that the Japanese economy supposedly got better when the government was keeping up spending.  Let us look at this claim next.

This chart shows GDP growth; unemployment; government consumption, change over year earlier; and taxes, percent of GDP, change from year earlier.  These measures are all in percent, so that enables us to compare them in one chart.

One thing to be careful about is that the slope at any given segment matters less than the distance from 0.  The last good year of the Japanese economy, 1989, is included, and the GDP line drops dramatically from there.  That just means the economy stopped growing, it did not actually shrink until the 97/98 crisis. 

Taxes are included in the chart, though it stays within 1 percent of change over the previous year (except for 1992).  Apparently the Japanese government did not pursue tax relief or revenue-based austerity over the period.  Without making a comment about the effectiveness of that, it is a bit fortunate from an examination standpoint, because it is then an important variable that is held almost constant.

I make three observations from this chart:

1. Government spending growth does seem to lead to some nominal GDP growth.  It is not a silver bullet: it did not seem to help much in 89-90, where I would have expected stimulus to be most effective, from what the Keynesians tell me.  Neither is it required: Japan's economy had some of its best years from 2004-2007, though government spending levelled off.

2. Japan's attempt at stimulus was enourmously costly, yet its best year gave the Japanese a paltry 2.8% economic growth.  For this, the Japanese government grew permanently from ~30% of GDP to ~40% of GDP, and its debt to GDP ratio went from ~20% to ~80%.  Despite all this increase in spending, the economy never did as well as it had done in the 80s, and neither did it achieve consistent growth.

3. Most importantly, the unemployment rate did worst during years with the most spending.  Unemployment rose steadily during the 94-95 years, when the initial stimulus spending kicked in.  It then levelled off when spending slowed to a more reasonable 1% in 97.  As the 98-99 stimulus spending took off, unemployment went up with it.  Not until spending growth slowed after 2003 did unemployment start to come down, slowly. 

Even if unemployment had stayed neutral, the tradeoffs seems dubious, at best.  With the third point above, one of the best arguments normally presented in favor of stimulus spending seems to be exactly on its head.  If we take the Japanese example as a case in point, stimulus spending is a terrible idea.  Of course, it is possible that this period is not a good example of stimulus spending, but then Keynesians need to stop using it as their poster boy.