A Tale of Three Countries - Recovery After Banking Crises: A Comment
Unlike natural sciences, macroeconomics lacks pure laboratory experiments. Once a particular policy is implemented, the outcome of alternative policy is never known for sure. In a political debate this gives an opportunity to convincingly argue whatever policy had glorious / disastrous consequences without being debunked from the opposite camp.
Instead one may study the story of "twin" countries and then link the differences in performance to differences in policy. Indeed, few countries in a real world are so similar to being regarded as "twins". I must disagree with Zsolt Darvas that Latvia, Ireland and Iceland could be regarded as "twins". Just to mention one difference, income level in Latvia is 2 – 2.5 times lower (in PPP terms), so huge wage differentials between Latvia and Western Europe led to emigration that was significantly underestimated by official statistics (used by Zsolt). Therefore, do not be surprised by the sharpest decline in employment in Latvia.
Here are some of my comments to Zsolt's paper.
First, Latvia's output level in 2007 was clearly overheated and beyond potential, therefore it should not be considered as a relevant base period. Although all three counties were somewhat overheated, Latvia took the first place in the EU regarding real GDP growth rate in 2004, 2005 and 2006. When the crisis came, first from the top became first from the bottom: "crisis hit Latvia harder than any other country". Should we link the GDP fall during the crisis solely to policy mix during the crisis (as Zsolt does)? I wonder whether any policy mix could avoid the "the-bigger-the-boom-the-deeper-the-bust" outcome. That is why the conclusion "Iceland seems to have the right policy mix" actually seems quite ambiguous. Furthermore, Zsolt mentioned that "in all three countries output fell back to its early 2005 level" therefore cumulative GDP drop during the crisis is just a reversal of the preceding, unsustainable boom.
Second, I do not think that a conclusion that "internal devaluation in Ireland and Latvia through wage cuts did not work, because private-sector wages hardly changed" could be made solely on the average wage data. Although it is true that the AVERAGE wage in the private sector hardly changed, wage reduction was large and widespread. How was it possible? If low-skilled employees are fired, the share of low-wage earners will decrease; the average wage in a country may even increase despite all employees who saved their jobs experiencing significant wage cuts (see numerical example). Just ask people on the street – it will be difficult to find someone who did not experienced a significant wage cut during the crisis.
Third, the paper underestimates the role of labour productivity increase to regaining competitiveness. According to the paper, labour productivity growth took place "only because employment fell more than output". If ONLY, of course, such labour productivity increases have rather few beneficial consequences. However, in this case we would not see labour productivity increase during the period of rising employment. Albeit, in Latvia, number of hours worked y-o-y grew by 3.1% in Q3 2011 while GDP per hour worked (labour productivity) increased by 3.4%... Although part of the average level of labour productivity rise was driven by an increase of the share of high-skilled employees, rapid advances of labour productivity allowed entrepreneurs to restore competitiveness (based on unit labour costs) without significant drop of average hourly labour costs.
Fourth, although Zsolt mentions that "recovery has started in all three countries – with the fastest pace in Latvia" he argues that "it has not yet brought many new jobs". How much is MANY? We will not return to overheated 2007 level of employment because no overheating is predicted (and is desirable) in a foreseeable future. Therefore, any employment increase will be small compared to 2007. Actually, Eurostat report tells that Latvia is currently witnessing the second highest job creation rate in the EU, trailing only Estonia (which, I presume, might be a better candidate for Latvia's "twin" than Iceland). Thus, no evidence of jobless recovery so far.
Fifth, Zsolt's paper compares official emigration figures concluding "Iceland experienced the largest net emigration […] even though official data on migration may not be reliable". In the case of Latvia, official emigration figures are significantly underestimated: alternative estimation done by professor Hazans and my own results suggest that about 200 thousand people left Latvia (net emigration) during the last decade (10% of population) while official numbers are six times smaller. Emigration is one additional reason why Latvia would not revert to 2007 level of employment. However, despite decreasing pace of emigration in 2011 and some evidence of intensifying return migration, the unemployment reduction is rather fast, so in Q3 2011 unemployment rate in Latvia was already lower than in Ireland.
Sixth, Zsolt's paper compares the changes in poverty rates during 2007 – 2010 concluding that "poverty was already high in Latvia and has increased, while Iceland was not impacted in this regard, and in Ireland a gradual improvement up to 2009". In reality, the situation is just the opposite: during the crisis, poverty rate decreased in Latvia and increased in Ireland. How it is possible? Eurostat figure for 2010 reflects the survey that was made during the 2010 about the household's wellbeing in 2009. Thus, all numbers in a graph for a particular year represent a situation a year before. Therefore, poverty rate in Latvia actually increased in 2007 but decreased in 2009. Despite significant wage (and tax revenue) cuts the government was able to save old-age pensions, thus, the poverty rate which during the good times was particularly high among elderly, decreased.
Finally, it should be noted that Bank of Latvia calculations on ULC-based REER show appreciation by 8% (Q2 2011 compared to the Q1 2000) that is substantially (not "somewhat") lower than according to the Eurostat data used by Zsolt (30%; the main difference is that Eurostat excludes Russia which is one of the most important Latvia's trading partners). And yes, regarding manufacturing, ULC-based REER was by 2% lower in Q2 2011 than in Q1 2000. Latvia's export shares are on an increasing path, leading to a historically record-high export volumes both in nominal and real terms. In the recent two years, Latvia's exports are increasing faster in real euro terms than that of Ireland or Iceland: the deeper the loss, the greater the recovery it seems.
To summarize, the conclusion of the paper that "Latvia would have likely been better off with devaluation" is mainly driven by two arguments that I showed are not reliable:
- Output and employment loss during the crisis in Latvia was larger than in Iceland.
However, I showed that this outcome by a large extent was driven by the difference in initial conditions ("The Higher They Climb the Harder They Fall" argument) and a significant extent of emigration from Latvia (10% of the population during the last decade, six times more than according to official statistics). The differences in initial (pre-crisis) conditions do not allow us to credibly link cross-country differences in performance during the crisis and differences in policy mix during the crisis.
- Internal adjustment through wage cuts in the private sector did not work.
However, I showed that when the crisis particularly hit low-wage earners (as was the case of Latvia), significant wage cuts could not be reflected in the average wage data. Actually, significant wage adjustment was taken place in Latvia; combined with robust labour productivity advances it helped Latvia to restore competitiveness and achieve record-high export volumes.