Inflation And Unemployment Conundrum
As I was browsing through my bookmarks, I bumped into that blogpost on the Citizen’s Watchdog, on unemployment and growth in Morocco. Just as he rightly pointed out, growth does not generate employment in Morocco. Or, to be precise, there is no direct correlation between growth in output and unemployment. Because the causal effect has to be investigated beyond the regular correlation device at hand. My interest in this post extends a little beyond mere output growth and unemployment; it looks into the other ingredient in the mix: inflation. And if I may add a fourth component, expected volatility.
The standard Philips curve (with or without the NAIRU rate) nor Okun’s law apply in Morocco, at any point of history, that is. Remember that Morocco has come a long way in fighting inflation. The second graph shows a steady decline in 10 and 15-years average in CPI inflation, a year-to-year decrease of .3% since 1989. However, the other end of the bargain has not come out exactly the way policy-makers would hope it would: unemployment has not dropped until the late 1990s, and for all the straight growth figures since 1999, an average GDP growth of 4.3% resulted only in a cut of 4 points in unemployment, even as the size of the labour force increased only 1,7 Million over the last decade. This means the economy grows only so much to post vacancies to new labour force, and only marginal openings to on incumbent unemployed. Obviously, we should be expecting some comprehensive study from the Labour ministry, but as far as aggregate data shows, I would bet growth in Morocco creates enough jobs for the new labour force only, and higher levels of growth do not automatically expand to tap into the unemployed population. Besides, Bank Al Maghrib did not provide a clear evidence as to how Okun’s law can apply to Morocco: their 2009 annual report stated the obvious with (statistically) insignificant regression line drawn through a cloud pairs of GDP growth and unemployment rate (p.47) On the other hand, when one looks at lagged aggregates and their historical volatilities, the picture becomes perhaps a little clearer.
I would however direct the reader’s attention to the more intricate relationship between output, inflation and unemployment: in fact, I contend it is wrong to compare a long-term aggregate variable -unemployment- to the more cyclical aggregates that are GDP and inflation. By now, we can safely assume much of Morocco’s unemployment is structural for many reasons, among others labour legislation and collective bargaining mechanism designs. Again, some study has to show if the unions are helping, or if workers are protected in their rights. So, annual unemployment rate should not be compared to annual CPI or growth: respective correlations over 1976-2010 are -.194 and -.079, too weak a result to tell any meaningful story. On the other hand, when annual unemployment is compared to say, the 10-year growth GDP, correlation is much strong (-.47) and equally the 15 years average CPI inflation observes tighter correlation with unemployment (.608) which tells a story: an effective policy designed to fight unemployment cannot expect to yield immediate results. In that the current government can be reasonably excused if they fail to cut unemployment to 7% by 2016 as they promised. On the other hand, a steady policy aimed at keeping both inflation and growth within the long-term figures do contribute more effectively seem to do better in bringing down unemployment.
In fact, two elements seem to display large correlations with unemployment: 3-years average output volatility and 15-years average inflation are both tightly linked to unemployment: lower medium-term output volatility and long-term inflation rate. The relationship is initially described in linear terms such where is a weighting parameter for inflation on unemployment, and the premium put on GDP volatility, and the NAIRU. At a 95% confidence, the formula above provides good indications the benefit of cutting GDP volatility: every 1% subtracted from GDP volatility with respect to the 3-year average yields a 2.01% decrease in unemployment. Admittedly, it is very hard to control GDP volatility over 3 years, but if the right policies are devised, a sustained decreasing output volatility all the way to 2016 can bring unemployment as low as 7.2%, and 6% by 2021, i.e. Morocco’s full employment. In that respect, it is not the level of output growth that matters, but rather how stable it is with respect to medium term targets. (incidentally, estimated values for and are respectively .41 and 2.01 (The formula itself can be adapted to encompass target levels similar to the Taylor rule) (results are pretty much the same) but this is a body of evidence reliable enough to observe that stability in output and inflation, more than higher levels of growth can help to bring down unemployment. We thus consider more institutional arrangements to reduce volatility, chiefly by reducing uncertainty, which calls for large institutional transparency in setting official targets, for instance.
These computations suffer three major shortcomings: first, they do not account for all the cyclical fluctuations in the considered aggregates; because only annual data is available to me, the analysis undoubtedly misses out on quarterly fluctuations especially (perhaps primarily) on inflation. Second, the formula downplays a pretty significant correlation between long-term inflation and GDP volatility, which indicates some additional endogenous effects that remain to be capture in a more specific model; Third, employment dynamics are significantly different in the agricultural sector, which may value output stability differently.
These caveat however would not fundamentally contradict the initial assertion behind these figures: economic stability both in expected inflation and output serves Morocco better than high levels of growth.