What is missing in Krugman's structural unemployment story?
Paul Krugman proposed an easy test to check whether a country suffers mostly from structural or cyclical unemployment. In the case of structural unemployment, "job losses should be concentrated in the bloated sectors, […] employment should if anything be rising elsewhere". Next, Krugman shows us some nice graphs for the US (see here and here) where job losses (compared to the pre-crisis level) are evident in all sectors/professions, asking "who’s bidding for workers?" and concluding that "evidence is strongly inconsistent with a structural story, and quite consistent with a demand story".
First, let's follow Krugman to test sectoral mismatch in Latvia. As I mentioned previously, in 2007 Latvia was one of the most overheated economies in the world, so job losses were evident in all major sectors: agriculture, industry, construction and services (see Figure 1).
Figure 1. Employment changes in Latvia by sector; 2011 vs. 2007; %
A disaggregated breakdown shows, however, there were some (mostly small) sectors that still increased employment in 2011 as compared to 2007 (see Appendix for details). But in this blog I wanted to concentrate on a much more important point that was missing so far.
Structural mismatch (e.g., labour shortage of salespersons while there are plenty of construction workers) is not the only form of structural unemployment (but the only one Krugman has tested). Thus, even if there is no structural mismatch, the economy could still suffer from structural unemployment. I have argued many times before (for instance, here) that in Latvia (at least), structural unemployment primarily takes the form of skills mismatch (i.e., labour shortage of skilled workers in almost every sector and profession, and low demand for unskilled ones). Skills mismatch is difficult to measure since it only partly relates to education and includes many interpersonal characteristics that remain unobservable by any statistical office. However, one proxy for this variable could be the educational attainment (see Figure 2).
Figure 2. Employment changes in Latvia by educational attainment; 2011 vs. 2007; %
Note that I use unadjusted LFS data here because data adjusted for population census are available only for 2011 and are not comparable with the earlier periods. But even if adjusted data of 2011 were compared with the unadjusted (overestimated employment) data of 2007, the employment among those with higher education is still growing.
As compared to the overheated 2007, the number of the employed with a higher education increased by 16%. I doubt whether this could be explained by Krugman's cyclical (i.e., demand) unemployment hypothesis according to which everybody is equally hit by the crisis.
At the same time, the employment decline for those with merely a basic education and less was dramatically sharp (more than 30%). In the overheated 2007, when more than 40% employers in construction claimed labour shortage as a main obstacle for doing business (and about 35% in manufacturing; European Commission business survey data) as the wage growth exceeded 30% (Central Statistical Bureau of Latvia data; which was far beyond labour productivity rise), enterprise had no another opportunity but to hire the low-skilled.
Which sectors were the engines of employment growth during the boom years? As employment peaked in the second quarter of 2007 and the earliest enterprise survey data are available for 2005, let's compare employment in the second quarter of 2007 with the respective indicator for the second quarter of 2005. The number of the employed increased by 14% in just two years (by 129 thousand): the general picture suggested that the tradable sectors (agriculture, industry) were stagnating while the domestic consumption sectors were prospering (see Figure 3).
Figure 3. Employment increase in Q2 2007 as compared to Q2 2005; thousands
Nearly a half of the employment increase was concentrated in construction and trade (relatively low-skilled labour-intensive non-tradables). In just two years, the number of the employed in construction surged by 52% (in construction of buildings – by 84%), in trade – by 14% (in trade and repair of motor vehicles and motorcycles – by 37%). The real private consumption growth (30% year on year at the end of 2006) required even more salespersons. The private consumption boom was unsustainably import-based (the current account deficit reached 27% of GDP) and was financed by strong credit growth (credit stock to households more than quadrupled in January 2005 – January 2008).
The increase in employment in other sectors was also mostly cyclical. For instance, employment in legal & accounting activities as well as in advertising & market research (both belong to the sector of "professional activities") in just two years increased by 56% and 68% respectively. The respective numbers for the employment activities and administrative & support service activities ("administrative and support") are 94% and 105% accordingly…
Despite Latvia becoming one of the world leaders with respect to the higher education institution students per capita ratio, the employment growth during the boom was particularly strong among the less educated group (see Figure 4; its share in labour force was declining fast; thus, this group strongly increased its participation and employment rates).
Figure 4. Employment changes in Latvia by educational attainment; 2007 vs. 2004; %
When the party has ended, less-skilled workers (particularly in the previously booming construction and trade sectors) were first to be fired, facing difficulties to find jobs in export-oriented sectors. First, the bubble has eroded the competitiveness of the exporting sectors. Manufacturing had to compete for workers with construction and pay wages far beyond the level commensurate with labour productivity. Unsurprisingly, manufacturing was stagnating when everything else was prospering (see Figure 3) and was unable to recover from the global demand slump until the wage-productivity gap closed at the end of 2009. Second, there was a mismatch – both related to the necessary education/skills and to high reservation wage of former construction workers.
After all, it is a bad idea to use 2007 as a base year as Krugman does. Neither employment nor unemployment rates are expected to revert to the overheated levels of 2007 (there is even no such a policy objective). By calculating an index (2007 year = 100) and not showing the previous overheating, there is an illusion of how big the negative output gap (and cyclical unemployment) is nowadays. In reality, however, the employment rate now is similar to that observed in "normal times" of 2002-2004 (the unemployment rate is higher owing to a higher participation rate). This suggests a rather small (if any) negative output gap and a large share of structural unemployment in total unemployment (which should be fought with structural – education policies).
Seems like Krugman's structural unemployment test has failed in Latvia?
Sectors that increased employment in 2011 as compared to 2007 in Latvia (16 out of 81 sectors according to CSB enterprise survey data; its share in total employment in 2011 was 11%):
- Forestry and logging;
- Mining and quarrying;
- Manufacture of motor vehicles, trailers and semi-trailers;
- Repair and installation of machinery and equipment;
- Electricity, gas, steam and air conditioning supply;
- Water transport;
- Air transport;
- Warehousing and support activities for transportation;
- Computer programming;
- Information service activities;
- Legal and accounting activities;
- Activities of head offices; management consultancy activities;
- Veterinary activities;
- Employment activities;
- Office administrative, office support and other business support activities;
- Social work activities without accommodation.