Shadow-boxing with DSGE Models
November 4th, 2016
by Philip Pilkington
Lars Syll has recently linked to a post by Noah Smith criticising DSGE models. Criticising DSGE models is the latest fad in mainstream macroeconomics — hey, it’s easy to use the model that was in fashion just before the crisis as a scapegoat to distract the profession from the fact that they still have no idea how to begin to explain the crisis or its aftermath.
I’m somewhat miffed that Syll gave Smith a pass on this one, to be honest. These criticisms of the DSGE models are so transparently self-serving that they really need to be called out. I can show this quite clearly by breaking down Smith’s argument and reconstructing it in a somewhat different way. Once this is done such criticisms of DSGE models can be seen as what they are: a rearguard defence mounted to prop up a rapidly decaying intellectual program.
Okay, so Smith’s post focuses on the fact that a key component of DSGE models called the Euler Equation is empirically wrong. Basically the Euler Equation states that when interest rates rise consumption will fall. The idea here is that the interest rate is a reward for forgoing consumption — i.e. saving — and so when it rises saving increases and consumption falls.
As Smith notes this simply doesn’t square with the data. In actual fact we see that in reality when interest rates rise consumption also rises. Smith makes no attempt to explain why this is the case — mainstream economists these days, obsessed as they are with their little models, don’t care much about causality and explanation — but it might be worth noting that this is probably, as Nicholas Kaldor noted in his study of the monetarist interest rate hikes in the late-70s and early-80s, due to the fact that higher interest rates add to savings which then produces a sort of wealth effect where the interest income stimulates consumption.
Now, let’s break down the steps implicit in Smith’s argument to see what, for him, constitutes a viable argument.
- Step one is that a hypothesis is created. In this case it is the idea that consumption should fall when interest rates rise.
- Step two is that this hypothesis should be tested against the data.
- Step three is that when we find that the data contradicts the hypothesis then the hypothesis is dropped.
Okay, this all seems pretty obvious, right? Well, why don’t we apply the same criteria to one of the marginalist sacred cows?
You see, in his post, Smith notes that the argument he lays out also relies on the idea of a utility-maximising agent whose preferences can be fixed. But of course, we all know that this idea does not stand up to empirical scrutiny. After all, Daniel Kahneman won a Nobel Prize in economics showing just that. So, not only do we have data on this hypothesis but we also know that this data is accepted by the mainstream economics community.
Well, by Smith’s own criteria shouldn’t that make the idea of a utility-maximising agent just as dodgy as that of the Euler Equation? Of course, it should but Smith marches on regardless without criticising the utility-maximising agent even though it would be falsified using the same criteria as he uses to falsify a theory that he doesn’t like — that is, DSGE models.
This is what mainstream economics has turned into. Economists bicker amongst themselves over models that are obvious nonsense. They will use criteria to prove or to falsify such models that when turned around on beliefs they all hold in common would demolish these too. It is a bizarre show.
Basically, this is a group of people vying for the Throne of Macro all the while secretly terrified that anyone will push the scientific criteria they claim to uphold too far and expose the fact that the Emperor is naked. How long can this carnival last? Well, if we look back on the last time such a discourse was firmly entrenched — that is, during the Scholastic era in the Middle Ages — we can be confident that it will last as long as those in power put up with it. With regards to mainstream economics, that day might well be coming to an end.