that either makes visitors lose their critical faculties, or non-visitors like myself and Paul Krugman lose theirs. In my earlier post, I was not that surprised that a member of the ECB’s Executive Board would trumpet Latvia’s ‘success story’, because that fitted the party line. I was a little more surprised to find the head of the IMF saying similar things, because the IMF has been rather more realistic about the consequences of austerity and internal devaluation. (Although the contrast between the IMF’s recent UK assessment in writing, and what Lagarde said in public, was widely noted at the time). What really surprised me was this post from Dani Rodrik, who has also just visited Latvia.
Dani Rodrick’s post goes into more detail about recent Latvian macroeconomics. In particular, he suggests that despite the massive decline in GDP and huge rise in unemployment, the economy may still have not completely regained the competitiveness they lost in the preceding boom. So, in terms of the evidence, he sees what I see if not worse. But then he says this.
“The main lesson I take from all this is the need to avoid easy generalizations that do not respect country peculiarities. Fiscal austerity missionaries are surely off base when they say Latvia’s experience decisively proves Keynesians and advocates of currency depreciation wrong. It is too early to judge the Latvian experience a success. But it is also too early to say Latvia has been a failure. Growth may continue, in which case the country will look better and better.”
When someone like Dani Rodrik looks at the same facts, but comes to rather different conclusions than me, I get worried that perhaps I’m wrong. It is also wise to heed Brian Ashcroft’s warning, that small countries do have different characteristics to larger countries. With this in mind, let me go through some of my own macroeconomic reasoning carefully.
Was the depth of the recession inevitable given the preceding boom? Paul Krugman is perhaps a little too dismissive on this point. The economy clearly was overheating in 2007/8, which is why it became uncompetitive. At the very least, inflation had to come back to a reasonable level. In principle this need not require any subsequent deflation if we have a totally credible macroeconomic policy, a suitable devaluation and a completely forward looking Phillips curve, but that is an idealisation. So some recession was probably inevitable, as it has been for many Eurozone countries. But in 2008/9 Latvia suffered the worst loss of output in the world!
The key issue is not that Latvia had to get inflation down, but that having done that it also needed to regain competitiveness. It is here that the macroeconomics of a short sharp recession with a fixed exchange rate looks so bad. To see why, think of the following little experiment.
We have an economy, which through overheating has become uncompetitive. It needs to get its prices in Euro terms down by, say, 20%. The government has dealt with the overheating: the output gap is now zero and inflation is at its competitors’ level. But prices are still 20% too high.
The least cost option is to devalue the currency by 20%. Now it would be foolish to believe that is all you need to do. Restoring competitiveness will boost demand, so to prevent the overheating starting up again you need to undertake deflationary policy of some kind. You may also need some negative output gap because higher import prices will raise price inflation. However you do not need a 20% decline in output.
But suppose we take the fixed exchange rate as given. Even in this case, a short sharp recession makes no macroeconomic sense. Suppose that, for given inflation expectations, a 1% output gap will reduce inflation by 1%. A short sharp shock of output 20% below potential for a year will give you -20% inflation in that year, so you will have restored competitiveness quickly, but at great cost. Now think about spreading the correction over two years.
To see how that works, we need to say a bit more about the Phillips curve. As we discovered in the 1970s, inflation today depends not just on the output gap, but also expected inflation. Suppose our Phillips curve is of the modern New Keynesian kind, so this year’s inflation rate depends on next year’s expected inflation, and let’s assume people are pretty rational in forming their expectations. What if we have an output gap next year of -10%. This will mean that inflation next year will also be -10%. But this year we do not need any output gap at all. Inflation will still be -10%, because expected inflation is -10%. So we get our competitiveness adjustment over two years, but at half the cost in terms of lost output.
Is my assumption about rational expectations critical here? No. Imagine instead that the Phillips curve is of the old fashioned kind, where expectations are naive and backward looking, so current inflation effectively depends on past inflation. In this case we need an output gap of -10% this year, but then nothing next year, and we still get our correction over two years at half the cost. (We have to do something to get inflation back up after two years in this case, but that is not important to the point I’m making.)
Now this is all very stylised and partial equilibrium, but there is one important message that will survive complications. The Phillips curve tells us that reducing the price level gradually over time is more efficient than doing it quickly. So even if you believe that you have to stick with a fixed exchange rate, a short sharp recession is much less efficient than a more modest but prolonged recession. Thinking about the convexity of the social welfare function reinforces this point.
As a result, even if output growth this year and next year was over 5% p.a., and the country achieves a sustainable level of competitiveness, I would not call the Latvian experience a success story. The competitiveness correction will have cost the economy a huge amount in wasted resources and unemployment misery, when it could have achieved this correction at a much reduced cost.