Saturday, November 19, 2016

Ambit Capital Prediction of Sharp Drop in GDP Growth is a Joke

Ambit Capital Prediction of Sharp Drop in GDP Growth is a Joke

Vivek Dehejia

A widely publicized report (see here and here) by Ambit Capital has predicted a sharp drop in India’s GDP on the heels of the currency swap exercise. The report is available here.

These large estimates are at variance with other predictions and back of envelope or common sense predictions that the temporary drop in aggregate demand caused by the short run liquidity crunch might shave on the order of a quarter to a half percentage point of GDP growth over the relevant time horizon.

So what is going on?

The answer is that Ambit has completely changed the methodology by which they make predictions of GDP growth! They write:

In the light of the disruption caused by the demonetisation decision triggered by the Central Government on November 8, 2016 we suspend our current supply-side GDP model.

The model to replace it?

To quantify the impact of the demonetisation on GDP growth we rely on the ‘quantity theory of money’ (QTM).

One really doesn’t know whether to laugh or to cry. The QMT derives from the following accounting identity, known as the equation of exchange:

MV = PY

Basically, this says that the stock of money (M) multiplied by the velocity of its circulation (V) is identically equal (as a matter of accounting) to nominal GDP (PY), which may be broken into real GDP (Y) multiplied by the nominal price level (P).

This equation only becomes a theory if one assumes that one of the four components is fixed or is determined outside the equation. Thus, the usual interpretation of this equation in macroeconomics is that, when converted into growth rates, gives an estimate of long run inflation -- not of short run changes in GDP growth!

Thus if you assume that V is stationary in the long run and log differentiate the equation, you get the following:

% Change in M = %Change in P + %Change in Y

If you re-arrange, you get:

%Change in P = %Change in M - %Change in Y

In other words, assuming that velocity is stationary in the long run, the equation predicts that long run inflation is determined by the long run growth of the money supply once one adjusts for growth in real output.

Yet another interpretation is to take this as a basis for Milton Friedman’s “k percent” rule of classical monetarism. If you want to target growth in the monetary aggregate, and you would like (say) 2% inflation over time, then you should allow the money supply to grow at 2% plus whatever is the expected long run growth rate of real GDP.

Please note that nowhere in the above is the equation of exchange taken as a theory of real output determination in the short run! It may be seen as a short-cut theory of aggregate demand. In fact, in a research paper with my colleague Nicholas Rowe, published in 1998, we interpreted the equation in exactly this way. (See here.)
In a stylized theoretical model (that too, 1990s vintage!) it may be acceptable to use the equation of equation as a theory of aggregate demand, although that is not the current state of the art, which would employ some contemporary beefed up version of the textbook IS/LM or Mundell-Fleming model, or the optimizing DSGE type models.

The crux is this: Ambit gets its extreme predictions by an extreme and problematic modelling strategy, which is to throw out their supply side model, and assume that the dynamics of real output are entirely determined in the short run by the equation of exchange.

Thus, with a predicted sharp drop in the velocity of circulation, the “model”, if one even wants to call it that, predicts a sharp drop in real GDP, since, presumably, there is assumed to be no impact on the price level. Strictly speaking, though, note that their prediction should be of the impact on nominal GDP (PY), unless they assume absolutely rigid prices (P). In other words, even within their posited modelling framework, part of the predicted drop in nominal income may occur through a temporary drop in the nominal price level rather than drop in real GDP. They do not explain, so far as I could tell, why they assume all the action in their simplistic model comes from Y and not from P.

The bottom line: Ambit’s “model” from which their extreme prediction is derived is non-serious and would be laughed out of court by anyone serious about macroeconometric modelling of economic dynamics. It is frankly surprising that is has received so much play. Were it submitted to my upper division or graduate seminar, it would rate a B+ or so for effort.

Vivek Dehejia is an economics professor at Carleton University in Ottawa, Canada; a resident senior fellow at IDFC Institute in Mumbai; and a columnist, Mint. Follow him on Twitter @vdehejia.