Scott Sumner  

Germany's mysterious recovery

Ramblings on Piketty... Civil Disobedience: King versu...

In the past 10 years Germany as gone from being the "sick man of Europe" to the star of the eurozone. This partly reflects the strong job creation that preceded the recession, perhaps due to the labor market reforms of 2003. However the post-2007 performance is even more amazing. There was almost no increase in unemployment during the recession, and the unemployment rate has fallen to relatively low levels during the recovery. Here's some data for the US and Germany between the last quarter of 2007 and the last quarter of 2013:

Country ------- United States - Germany

Real GDP growth: +6.3% /// +4.0%

Nom. GDP growth: +16.3% /// +12.8%

Un. rate change: +2.2% /// -3.2%

Employ. change: -0.7% /// +6.0%

Total labor comp: +12.2% /// +19.2%

Labor comp./GDP: -3.6% /// +5.7%

Ave. weekly hrs: -0.4% /// -1.1%

[Update: Marcus Nunes pointed out that NGDP growth for the US was misreported in the original post.]

Let's start with the two GDP numbers. The US did a couple points better, but we also had modestly higher growth in our working age population. The US population age 16 to 64 grew by 3.2%, while the German working age population declined a few tenths of a percent (quarterly data is not available.) Thus in per capita terms Germany did 1 or 2 points better. But the real action lies elsewhere.

Initially I thought that the sharp fall in the German unemployment rate might have reflected falling German productivity and slow population growth. But that doesn't explain the NGDP figures. Germany actually created far more jobs than America, despite lower NGDP growth.

The next step is to look at compensation per worker. I couldn't find hourly compensation data for Germany (outside manufacturing) but if one compares total compensation to total employment it looks like compensation in Germany rose by just over 12% per worker. The US figures appear to be just under 13% per worker. So it does not look like the German employment miracle was caused by low wages.

So what is the explanation then? As far as I can tell, the only plausible explanation (at least in an accounting sense) is a breakdown in my "musical chairs model." In this model the level of hours worked reflects the interaction of NGDP and (sticky) nominal wages. It is assumed that total labor compensation is a stable fraction of NGDP. Arnold Kling found this assumption so reasonable that he once called the model an "identity."

In this case, however, the share of income going to labor behaved very differently in Germany and America. Let's start with Germany. If NGDP rose by only 12.8%, and compensation per employee rose nearly as much, then how could employment have increased substantially? The answer is simple. Workers grabbed an extra 5.7% of national income (up from 48.7% to 51.5%.) That roughly explains the 6% rise in total employment in Germany.

In the US, NGDP grew a bit faster than compensation per employee. Thus one might have expected a small increase in total employment. Instead employment actually fell by 0.7%, as workers grabbed a 3.6% smaller share of NGDP (from 54.5% to 52.5%.)

Are there any lessons here? Yes, but not the ones you might imagine. The "solution" is not to try to increase the share of national income going to workers. Why not? Because the most likely methods (higher minimum wages, stronger unions, etc) are also likely to boost compensation per employee. So any gains flowing from labor receiving a higher share of NGDP will be offset by losses in jobs from higher wage rates.

Instead, the lesson is that NGDP in America should have grown faster than 14.2% 16.3% over those 6 years. A few years back Bill Woolsey recommended reducing the trend rate of NGDP growth from 5% to 3%, in order to stabilize prices. But even Woolsey's proposal (viewed as being quite conservative at the time) would have meant considerably more than 14.2% 16.3% NGDP growth over 6 years. Had his proposal been followed, unemployment in America today would likely be closer to 5%.

In Germany, the 6% rise in employment was partly reflected in a lower unemployment rate, and partly in a higher labor force participation rate. The labor market reforms encouraged the creation of many low wage jobs (Germany has no minimum wage) and helped low income workers with wage subsidies. It is interesting that the one shining labor market success of the 21th century occurred in a "social market economy" like Germany. Even more interesting is the fact that this approach (enacted by the Social Democrats) has almost no support among American progressives, who tend to prefer the more rigid southern European labor model.

Is it possible that (paradoxically) the flexible-wage German model actually boosted the share of income going to labor? Yes, but it's equally possible the increase occurred for some unrelated reason. In any case, the focus should be on jobs, jobs, jobs, not the share of income going to labor. To get there you need stable growth in NGDP (or total labor comp.), no artificial wage floors, and generous subsidies for low wage workers. It's a pity that there is no one touting that model anymore. Even the Germans have decided to abandon the labor market reforms of 2003 and adopt a minimum wage.

Indeed even Hong Kong now has a minimum wage. America recently increased the minimum wage by 40%, and there are calls for another 40% increase--at a time of slow NGDP growth and high unemployment.

Comments and Sharing

COMMENTS (13 to date)
TravisV writes:

Prof. Sumner,

Great post but what about the future?

Germany controls the ECB, right? Why is the ECB increasingly considering QE?

Could it be that weak NGDP growth in rest of the Eurozone is finally pulling down Germany's performance?

The divergence here cannot be sustained:

How will Germany respond as German NGDP growth heads south?

Max writes:

If your model holds true than we will see a sharp increase in unemployement next year, when Germany enacts its minimum wage law of 8.50 €/h.

Exactly your reason for job growth and labour participation will then be extinguished (except for the myriad of strange exceptions the "Great Coalition" has enacted).

I think it will be interesting to see, what will happen or you could just resign yourself to read the letters and comments of Don Boudreaux on minimum wage and the gap between non-payed internships and minimum wage.

marcus nunes writes:

Scott, I´m glad to say, despite appearences, that there was no breakdown in the Musical Chairs Model!

Anand writes:

Perhaps work-sharing is a part of the reason of Germany's low unemployment?

Dean Baker has been pushing this idea for some time:

Andrew_FL writes:

Calling a model an identity seems more like calling it trivial than calling it reasonable.

At any rate, distressing news about Germany there at the end. And after hearing so much good, too. :(

Scott Sumner writes:

Travis, It may be that the Germans would like a bit more growth, or perhaps they realize the high inflation they were worried about is not going to happen.

And they only control the ECB if the 17 others let them.

Steve Sailer writes:

Keep in mind that the Germans very quietly cracked down on immigration from places like Turkey. Compare population growth figures from 2000 to 2008: in Germany, population dropped 1 million, in Spain it rose 5.6 million. Is there much surprise who has higher unemployment today?

Philip George writes:

I do not see anything mysterious in these figures. Perhaps there is such a thing as NGDP fixation.

Anyway, let me try and make sense of the figures. Workers receive all their income as wages. So if employment goes up by x% one would expect the ratio of labor compensation to GDP to go up by x%, other things being equal. The figures show that this is indeed so. In Germany, employment went up by 6% and the share of labor compensation in GDP went up by 5.7%.

In the US both these variables again moved in the same direction. But while employment fell by 0.7%, the share of labor compensation in GDP fell by a larger percentage, 3.6%. This needs to be explained.

The difference between NGDP and real GDP is much higher in the US than in Germany, suggesting that money growth in the US has been higher than in Germany.

But one also needs to compare the proportion of money growth spent on real goods and services to the proportion spent on financial assets. In the US the Fed has conducted a mammoth QE operation. When the central bank buys bonds from an investment banker one does not expect him to use the money to increase his purchases of real goods and services; it is commonsense to expect him to buy another financial asset. And that this has indeed been the case can be seen from the steep run up in financial asset prices in the US. The larger amount of money spent on financial assets means the return on capital has surged. It is this which explains why the share of labor compensation in GDP in the US has fallen even faster than the fall in employment.

Jim Rose writes:

The key phase of the Hartz labour market reforms rolled-out on the eve of the GFC and reduced the German natural unemployment rate by several percentage points.

Christian writes:

You can still see a split between Western and Eastern German wages that also played a role in the discussion that we had about the minimum wage over here. My prediction is that negative effects of the new rate of 8.50€/h will be seen primarily in Eastern Germany where many jobs in the service sector are well below that rate.

Scott Sumner writes:

Steve, I'm more perplexed by the rise in German employment. I'd expect faster job growth in a country with more immigration.

Philip, I don't see why you'd expect the labor share to rise with employment. Employment in the US has trended upward for more than 200 years.

Jim and Christian, Those comments sound plausible.

Philip George writes:

Prof Sumner,
If no workers were employed, their aggregate wages would be zero, and labor's share of GDP would be zero. I hope you agree. If you do that explains the figures in your posting.

However, if all workers were employed that would not make their share of GDP 100% because others would receive incomes from interest, rent, profits, trading financial assets and so on.

QE has driven up the prices of financial assets and therefore proportionally driven down the share of labor even below what it would have been. It helps that managers' compensation is often tied to the price of financial assets and that a large proportion of the income of wealthier people comes from the trading of financial assets.

Petar writes:

Regarding the german minimum wage, DIW did the math ( It turns out that the minimum wage, that would now "benefit" some 5mil workers will have a really small impact since when it gets enacted, a big chunk of those 5 million will already get that amount per hour - in 2012 500.000 people got wage raise that got them more than 8.5Eur. So, only ones who will be left are low skilled people working in (mostly hotel etc) services in former Eastern Germany where paychecks are lower anyway. Help the poor, right.

Anyways, Deutche Bank research came out projecting some 450k - 1mil in employment losses, with aggregate wages shrinking about 8%.

Another thing is that collective bargaining did very good job for Germany allowing for more bargaining power for the workers while still focusing on competitiveness. Uniform minimum wage may change a lot there, too.


Comments for this entry have been closed
Return to top