Monday, 12 March 2012

Which Macroeconomic Model Best Describes Reality?

Another one of my Quora answers that got traction reproduced here:

1. No economic models accurately describe reality
2. No economic models are very good at making predictions about the future
3. The Smets-Wouters (2003 & 2007) model is considered the best of a bad bunch, the dwarf among pygmies.

Smets-Wouters, 2003 & 2007 is generally considered the most accurate general model of the economy in macroeconomics, when it comes to prediction[1] (note that this is different to 'describes reality most accurately', but I'll get on to this distinction later). It operates in the New Keynesian (or sometimes called the New Neoclassical Synthesis) school of thought, meaning that it takes a 'Dynamic Stochastic General Equilibrium' framework (in other words it works out its equations by examining individual, household maximising behaviour), but incorporates into that framework, multiple frictions, such as wage and price stickiness.

What are the main findings of this model?

While “demand” shocks such as the risk premium, exogenous spending and investment specific technology shocks explain a significant fraction of the short-run forecast variance in output, both wage mark-up (or labour supply) and to a lesser extent productivity shocks explain most of its variation in the medium to long run. Second, in line with GalĂ­ (1999) and Francis and Ramey (2004), productivity shocks have a significant short-run negative impact on hours worked. This is the case even in the flexible price economy because of the slow adjustment of the two demand components following a positive productivity shock. Third, inflation developments are mostly driven by the price mark-up shocks in the short run and the wage mark-up shocks in the long run.[2]

In other words, in the short run, Government Spending, Risk Appetite and technological progress that requires heavy investment to be realised explains output, whilst classical factors like Labour Supply and (to a lesser extent) productivity shocks matter in the long run. Productivity shocks matter less because they are offset by changes in hours worked, so that if people start earning more per hour, they simply choose to work fewer hours.

However, even though this is the "best" model, that does not mean it is very good! In fact, the performance of the model and models of its class are remarkably bad, given their significance in policy making.

The image below shows the performance of these models against the Fed's "Greenbook model" (GB) and against a purely statistical model with no theoretical underpinning (BVAR). Anything below one means that model has more accurate predictions.


Strictly speaking, this has not answered your question. Afterall, you asked "what model describes reality most accurately" - the only answer to that question is that economic models do not describe reality.

The job of an economic model is not to accurately describe reality, but rather, to accurately summarise the relationships between economic variables, such as output and inflation. It tells you what happens to inflation for a given change in output, and it gives you a systematic way of thinking about that relationship, but it does not pretend to give you a "true" story, in the sense that the truth is something along the lines of, well household a,b,c,d had a little bit more money, and they got together and said, lets spend on this on a trip to Paris, and then the airline that provides flights to Paris saw a 7% rise in demand. Since they didn't have the capacity to meet this demand, the board sat down and said "Look, we either raise our prices, or we keep them constant. Raising our prices will earn us more money in the short run, but may ultimately damage our brand". There was 14-5 split on the board in favour of raising prices, so prices went up. And then something similar also happened for groceries, and televisions etc etc. Obviously, no economic model is therefore going to accurately describe reality, in the way that a historical account might do. In general, therefore economists' models are judged on how useful they are, rather than how accurately they describe reality.


No comments:

Post a Comment