sábado, 7 de junho de 2008

On The Impact of Demand Shocks on Inflation

On The Impact of Demand Shocks on Inflation
Franklin Serrano (Instituto de Economia, Federal University of Rio de Janeiro, Brazil)
[june,2008]
Joan Robinson (1982) liked to tell the story that Kalecki once defined orthodox economics as “the science of confusing stocks with flows”. Nowadays in discussions about fighting inflation it seems that it is more about confusing levels, rates of increase and acceleration. Here we are concerned with a single question (for more details see Serrano, 2007): What is the impact of a demand shock on prices? Does it increase the price level, causes a positive rate of inflation or does it accelerate permanently the rate of inflation?
After Friedman and Phelps most people think that usually a single aggregate demand shock, which for our purpose we shall define as something that generates a positive output gap leads to a permanent acceleration of inflation, that is a step increase in the rate of inflation.
Formally, p being the current proportionate rate of increase of the price level , u being the ratio of actual output to capacity output (or employment to capacity employment) and and b a parameter we would have:
p-p-1= b(u-1)
It is not very clear why people think that way since there is overwhelming statistical evidence that almost everywhere in the world inflation persistence coefficients add up to less than one at least since great moderation of the early nineteen eighties in the rich countries and since the mid nineteen nineties in most of the periphery.
Now if they do add to less than one (in Brazil the highest estimate I have seen is 0.6) the impact of a single demand shock cannot be fully persistent. Therefore the impact of a single demand shock is on the level of inflation and not on its acceleration.
Formally, let a be the coefficient of lagged inflation p-1 (for simplicity we assume a one period lag only) . We then have:
p = b(u-1)+ a p-1
that will quickly tend to :
p = b(u-1)/(1- a)
which means that inflation will tend to zero as soon as the demand shock is over and the output gap is closed.
Does that mean then that demand shocks cause a level effect on inflation rather than its acceleration, like in the days of good old Phillips curve ?
Well, not really. As it happens, there is another thing. There is a large amount of statistical evidence which shows that the rate of capacity utilization and the rate of unemployment tend to be mean reversing (the former much more certainly and quickly than the former) . However there is a bigger amount of statistical evidence that the levels of output and employment exhibit strong or full hysteresis , which means that whatever causes their cycles is also what causes their trends or to put it in the words of someone who already knew that many decades ago “In fact, the long run trend is but a slowly changing component of a chain of short-period situations: it has no independent existence”(Kalecki, 1968, section I).
Ok, but so what ? The thing is that this overwhelming evidence of strong hysteresis shows that any increase in output (or employment) that is sufficiently persistence will bring wih it an endogenous increase in the levels of potential or capacity output (and employment). If and when this happens the output gap will close itself, not by a decrease in the growth of aggregate demand but by an induced increase in the growth of aggregate supply (as it is explained in models of flexible accelerator induced investment such as my own supermultiplier).
Note that every single econometric model that uses the now trendy (no punt implied) statistical filters to measure the output gap will exhibit these strong hysteresis effects on potential output.
But if we take these two pieces of statistical evidence together (on which see Braga,2006), i.e., the partial persistence of inflation and the strong hysteresis of potential output, the total long run impact of a permanent demand shock will be first a temporary level effect on the rate of inflation that will not last , even if demand growth is not checked , because later the level of potential output will catch up with the higher levels of demand and the output gap will close. When this happens the gap falls back to zero and there is no more demand inflation. In the end, the permanent impact of the demand shock will have been just a permanent increase in the price level. Note that is the sole long run effect of demand on inflation even for places where it is easy to find a significant impact of the demand shock variable on Phillips curves , which by the way is definitely not the case of Brazil (Serrano,2008).
References
Braga, J. “Raiz unitária, inércia e histerese: o debate sobre as mudanças da NAIRU na economia americana nos anos 1990”, IE-UFRJ, Ph.D. dissertation
Kalecki, M. “Trend and Cycle”, Economic Journal, 1968
Robinson, J. “Shedding darkness”, Cambridge. Journal of economics”,1982; 6: 295-296
Serrano, F. (2006). “Mind the gap: hysteresis, inflation dynamics and the sraffian supermultiplier” IE-UFRJ, mimeo
Serrano, F (2008) “Juros, câmbio e o sistema de metas de inflação no Brasil”, IE-UFRJ, mimeo