Seven technical reasons why ‘(Real) Unit Labour Costs’ are not a valid macro-indicator of competitiveness

Update: a revised, empirically extended  and improved version of this blogpost can be found here.

Unit Labour Costs (ULC) are one of the most used metrics during this crisis. Generally, it is stated that countries with a high increase of ULC have to get these down, at least compared with other countries, to become more ‘competitive’. The way to do this is to slash wages, as this will lead to a lower ULC and therewith to an increase of competitivety. The European Central Bank even considers ULC to be so important that they are included in the list of eleven core economic metrics, together with inflation, GDP growth, the current account, unemployment and some others (graph 1).

Graph 1. Development of the Eurozone ULC, Year on Year increase. Source: ECB.

For an economist this is quite embarrassing, as it is well-known that competitivity is only to a very limited extent dependent on the wage level (see this very recent ECB task group study or Felip and Kumar (2011) ). And though ULC might be useful to compare individual companies they are not fit to compare countries, for a whole bunch of reasons (see below). Felip and Kumar do a good job debunking the use of this metric. Their list of complaints is however far from exhaustive and I’ll try to complement it. To do this, I’ll however explain how they are calculated:
1) The RULC are calculated by Eurostat, based upon the next operational definition (emphasis added):
“This derived indicator compares remuneration (compensation per employee in current prices) and productivity (Gross Domestic Product (GDP) in current prices per employment) to show how the remuneration of employees is related to the productivity of their labor. It is the relationship between how much each ‘worker’ is paid and the value he/she produces with her work. It’s growth rate is intended to give an impression of the dynamics of the participation of the production factor labour in output value created. Please note that the variables used in the numerator (compensation, employees) relate to employed labour only while those in the denominator (GDP, employment) refer to all labour, including self-employed”.
Ehm… yes, that’s nothing else than the share of labour in total income. Which is not really any kind of indicator of competitivety. There are more ways to calculate RULC but these all boil down to using a complicated way to estimate the labour share. See the introduction of this IMF study.

2) Aside from RULC we have NULC, Nominal Unit Labour Costs or nominal wage level divided by real GDP per worker. The way growth rates for this hybrid indicator are derived is taking time series of ULC and deflating GDP (in the denominator) with the GDP deflator while compensation per employee is not deflated by any deflator(Felip en Kumar, 2011). It is therefore an indicator with a numerator in current prices and a denominator in fixed prices. Why does this lead to problems, on the macro-level?

A) First, it is an indicator which contrary to many statements should increase, considering stated policy goals. Imagine a country with (like the Eurozone) an inflation target of 1,8% but which (unlike the Eurozone) does not target consumer price inflation but the GDP deflator (which also includes the prices of individual government consumption, like education, collective government consumption, like street lights, government investments, like roads and private investments, like houses or machinery). Such a level of inflation can not be sustained when, in the medium run, wages do not increase with at least the same percentage (in fact: a slightly higher percentage, assuming a stable RULC and some increase of productivity). Arithmetically this means that unless the rise of deflated GDP (per employment) is in the medium run larger than the rise of nominal wages (per employee), which in a situation of even 1,8% positive inflation is a sheer impossibility, the ULC is bound to increase . Looking at the 2009 value of the increase of the Eurozone ULC shown in the graph it can easily be shown that the 6% increase is the consequence of about 2% increase of nominal wages and a 4% decline in real production per employee (but a lower decrease of nominal production). But did this really affect competitivity? It was caused by an unexpected decrease of production which affected profitability – but not competitiveness.

B) The Eurostat caveat: in the definition above the phrase ´Please note that the variables used in the numerator (compensation, employees) relate to employed labour only while those in the denominator (GDP, employment) refer to all labour, including self-employed´ was emphasized. This phrase means that countries which had a large share of self-employed which left this status to start to work for wages (I.e: a shift out of peasant farming and into tourism in Greece) will see an increase if the ULC, because of structural modernization!

C) A related problem exists because of differences between economic sectors. Ireland is an example. The bust of the building boom led to the demise of lots of industries with relatively high ULC’s while industries with a low ULC and a high share of capital, like the pharmaceutical industry, were much less affected by this bust. As a result, the average ULC of Irish industry declined – while the ULC’s of subsectors of Irish industry did barely change. Talk about a fallacy of composition!

D) Non-tradeables. Between 2000 and 2011, average German wages did not rise too much which led to a low increase of the German macro ULC. But to quite some extent this was caused by stable nominal wages (and falling purchasing power) of teachers. Wages in industry rose a little above average and Germany still is one of the very few countries where industrial wages are higher than the average (even despite very high wages in the financial sector). But does decreasing real wages of teachers really increase the international competitivity and exports of a country? It might affect the current account as German teachers will have had to restrain consumption of, among other things, imported products. But at least I do not see the relation with gross exports!

E) Felip en Kumar (2011) mention the Kaldor-paradox: the empirical evidence about the ULC and competitivity in fact suggests that high increases of the ULC do not cause a decline of competitivity but are, to the contrary, a sign of succesful export performance.

F) Another fallacy of composition. An individual firm can increase its competitive position by cutting the wage level as the wages it pays (almost) do not affect the demand for its products. But when every company decreases wages total demand will suffer. Greece (where nominal wages have decreased with about 20%) is an extreme example of this.

G) Global supply chains. The share of ‘domestic’ labour in the cost price of tradeable products shows sustained declines. Giordano and Zollino mention that the ‘domestic labour share’ of Germany declined from 27% to 21% of gross output (not the same as value added, the concept of GDP!) while the Italian share declined from 21% to 18%

Summarizing: ULC are a  lousy macro-indicator of competitiveness, much better ones are available, like the one proposed in this ECB study. And yes, it is embarrassing too that the ECB does not take its own studies serious and still uses the ULC as a competitivety metric