February 20, 2007
Hardly any concept has caused as much dispute in the German economic debate over the past years as the idea of the “Bazaar economy”, the claim of Hans-Werner Sinn that the German export performance is sign for economic weakness. According to Sinn, the raising German share in world exports is a pathological problem stemming from the fact that companies are increasingly using cheap intermediary inputs for their export goods while the domestic share of value added is shrinking. The fault lines run across the political spectrum, with critics from the left (the union-financed think-tank IMK) up to the right (the employer-financed think tank Institut der deutschen Wirtschaft) questioning this analysis while TV talkshows and politicians have readily picked up on the idea.
As the support for the theory in Germany has ebbed with a strong turn-around in the labour market, the concept still seems to have a large number of supporters abroad: Just a few weeks ago, the IMF published a new working paper, supposedly given support to the Bazaar hypothesis, which was quickly picked up by several websites, among others Wolfgang Munchau at Eurointelligence.
Wolfgang rather enthusiastically wrote: “As intutitive the [Bazaar] theory may be, it is has not been supported by much empirical evidence so far. This working paper is to our knowledge the first serious contribution that claims to have established solid empirical evidence in favour of Sinn’s theory. […] Wage moderation and the structural shift in export demand, possibly the most widely cited arguments in the policy debate, play only a minor role in explaining the growth of Germany exports.“
However, if one looks closer, a number of problems arise with the IMF paper, very much questioning the conclusion (to the authors’ defense, they have been very careful in wording their results – much more careful than those reporting on it).
The problem starts with the time frame for the IMF analysis: The authors look at the years 2000 to 2005, when the German world market share has expanded strongly and ask what kind of effect the wage moderation had. During this time, the euro appreciated strongly in nominal terms and thus basically counteracted the wage restraint. The authors consequently find that wage restraint in this period has not caused a significant increase in world market shares.
However, it is reasonable to assume that the export growth in this time was a lagged effect of the large increase in competitiveness from 1995 to 2000. After all, Germany’s export market shares in the years before had not been lost overnight, but continued over an extended period even after the real appreciation had been stopped. The IMF’s analysis does not allow for such a lagged causality: In the econometric analysis, the paper only allows for an implausible short lag of only a few quarters between exchange rate movements and change in export volumes, even though former studies have found much longer lags (Bayoumi found in an earlier IMF working paper a lag of up to 4 years, with the main impact arising in the second and third year). While the authors of the Bazaar study argue that in Germany the exchange has reacted quicker as there were spare capacities after 2000, this is most likely does not cover the cases in which companies rethink the location of investment in new production units which can be expected to happen when there is a pronounced increase in competitiveness as it has been the case between 1995 and 2000. After all, the Bazaar phenomenon is supposed to be a structural, not a short-term cyclical issue.
The next problem is the data basis for the variable which measures the share of domestic value added in industry. The authors somehow constructed this series from data from the statistical office, claiming that there has been a fall in the share of value added in manufacturing production from 2000 to 2005. However, if one looks at the Bundesbank figures (here and here – “produzierendes Gewerbe ohne Bau”), one sees that from 2000 to the third quarter 2005, industrial production has increased by 13.5 percent while real value added in industry also has increased by 13.5 percent (data from February 14, there are likely to be revisions published next week). Thus, at least over this period, there has been no change in the relation between the two. While there might have been some fluctuations over the past year, from 2000 to Q3 2006 overall, the contribution of any Bazaar effect to export growth must have been zero. (While the difference between this measure and the authors’s measure might be due to some differences in seasonal adjustments or data revisions since the study was conducted, these Bundesbank data seriously questions the results.)
Besides these data questions, there is a methodological problem with the IMF paper. While it tries to assess which part of the German export share in world trade is due to the Bazaar phenomenon, it completely neglects the fact that growth in world trade itself may be distorted by some Bazaar-like developments. A number of factors point towards the fact that an increase in the share of intermediary goods in industrial production is a worldwide phenomenon. According to the WTO, from 2000 to 2005, world trade in manufactured goods has increased by 28 percent, while world GDP has only increased by 14 percent. Almost three fourths of all world trade is in intermediary goods. Logically, this can only go together with a significant reduction of the domestic share of manufacturing – not only in Germany, but in all or at least most important trading countries. And if one looks at single country data, one almost always finds indications of an increasing share of imported intermediary goods – even and especially in low-cost countries such as China and India.
If there were not the Bazaar phenomenon worldwide, world trade would thus have grown much slower over the past years (and thus the German export share faster than what now has been reported). If one wants to analyse what part of the increase of the German share in world exports is due to the Bazaar phenomenon, one cannot simply assume there were only a Bazaar effect in Germany but none in the other countries – or at least should provide some estimate about the magnitude of this phenomenon in the rest of the world. In its present form, the paper leaves us completely in the dark whether the effect of an increased division of labour on the Germany is bigger or maybe even smaller than in other countries.
Thus, in the end, the paper does not bring us any closer to understanding whether the Bazaar economy is a pathological phenomenon of the German economy or just another – prerogative term – for the growing international division of labour.
The stunning thing is that the Bazaar hypothesis survives even though there are a number of easy indicators which actually points toward a very healthy development of the German manufacturing sector which are incompatible at least with the early versions of the Bazaar economy (such as the one presented in Sinn’s “Ist Deutschland noch zu retten?”, but which has been modified in “Die Basarökonomie” – see below):
- Since 1999, domestic value added in manufacturing in Germany has grown stronger than in any other G7 country (see figure 1 – data from the European Commission’s AMECO database)
- Germany is the only G7 country in which the share of value added in manufacturing in GDP has actually risen significantly since 2000 (see figure 2)
- Employment in manufacturing is actually increasing in Germany, in contrast to other OECD countries
Of course, one could start considering all these indicators as further evidence that the Bazaar economy hurts Germany, as Hans-Werner Sinn is doing in his latest writings on the topic (the book “Die Basarökonomie”). According to him, capital in Germany is “fleeing into the export sector” because investment in the domestic (service) sector is unprofitable due to overly high wages for low qualified. This leads to a national specialisation in the capital-intensive export sector. This then, he believes, shows up in rising value added in manufacturing.
However, even Sinn cannot explain why this should be the case even though almost 20 percent of full-time employed Germans work for less than 7,10 € an hour (according to the Sachverständigenrat) and some for even 4 € an hour while Great Britain does not suffer under a similar pathological “Bazaar effect” with capital fleeing into the export sector but virtually full employment with a legal minimum wage of roughly 8 € per hour.
This post has been also published on Eurointelligence.