Mankiw is right again – this time on prediction
This time on how Economics as an academic discipline will not have to have the wholesale changes some peoples are suggesting.
He is right when he says the focus of economists and economic teaching is not on prediction. However, I would also say that economists HAVE sold the idea that they can predict when talking to people, even if they personally realised this isn’t the primary role.
In some sense this comes back to Friedman. During the positivist revolution in economics he stated that it didn’t matter so much what we assumed – as long as it was predicatively accurate. Furthermore, our “value” for policy analysts and the such has often been tied to predictive accuracy. Here we never agreed with this.
There are two ways to understand current economists methinks:
- Economists want to explain and understand – and that is where the value is: This fits the academic economist view, but often ends up with no predictions.
- The Tarot card view: This fits what economists do when they have to make predictions. They use archetypes (models of aggregate behaviour), historical knowledge (data), and intuition to get a feeling of where the economy will head and the risks around it. Even the more technical models (think DSGE models) have elements of this.
Both these services have value – by building knowledge and understanding. But economics as a discipline should be based on its ability to adequately explain and provide understanding – not its ability to predict (especially given the issues with data).
Economists add value by describing, explaining, and painting risks – but they do not have magic time traveling powers.
This all reminds me of:
http://www.ritholtz.com/blog/wp-content/uploads/2009/09/economics03.jpg
Friedman ’56 isn’t the same as forecasting though. Build model, run comparative statics for predictions all else equal, see how well it tracks in the real world. Like: double prices, expect quantity demanded to drop, check to see if that’s happened.
Totally agreed on forecasting – that’s voodoo. But any decent economist should be able to make predictions like “If minimum wage is tripled, employment will drop by a lot”.
@Eric Crampton
Hi Eric,
True, but Friedmans call appeared (and was taken by many as) very much against the idea of focusing on the realism of assumptions – and also pushed the idea that economists should be judged on their ability to predict. This never sat comfortably with me.
“Totally agreed on forecasting – that’s voodoo”
As a forecaster I hope you aren’t using voodoo in a negative sense 😉
Forecasting has value, but I think the main points to keep in mind are:
1) Errors in forecasting don’t tell us anything about how the teaching of academic economics needs to change – Mankiw’s point.
2) The value of forecasting comes from the information provided, it isn’t all about having the lowest RMSE – as this without some form of consistent understanding and explanation would be of little use.
Comparing forecasting to Tarot card reading isn’t actually an insult – I am just saying that both use forms of archetypes to frame and explain complicated behaviour/phenomenon. As I stated:
“They use archetypes (models of aggregate behaviour), historical knowledge (data), and intuition to get a feeling of where the economy will head and the risks around it.”
“Mankiw is right” posted on April 1st, i thought you must be referring to his latest post http://gregmankiw.blogspot.com/2010/03/taxes-per-person.html
@EbolaCola
Ahhh, I don’t really agree with that article – as surely spending on government services is a normal good and so (all other things equal) a wealthier country should have a higher tax bill. Did you think I was trying to pull an April 1st 😛
yeah i read that Mankiw then saw your headline in the RSS and was hoping for an April Fools 😉
@EbolaCola
For Mankiw to see the headline in an RSS he would have to subscribe – I have no doubt that he would never subscribe to this blog 😀
I wonder what other disciplines/professions are held to the same kinds of forecasting standards that economists are from time to time? Should we lose faith in medical science because it did not forecast swine flu? Are climatologists out of touch because their models vary and they can’t forecast the weather? Is forensic science in terrible trouble due to type II errors on DNA tests in murder cases?
@Hone
That is very true Hone.
I find it runs even deeper than that for some areas that are placed in the branch of science – eg nutritionists. Often they do not properly show causation, do not include obvious exogenous variables, and do nothing about endogeniety – and yet a lot of their work is shown as “truth about food”. It irks me, and a number of health professionals I know 😛
Sometims it feels to me like a large body of NZ economists not only can’t forecast but don’t even try. For example, take this pearler from ANZ’s report this week: “We will need to see the experienced trading activity measure catch-up to the expected reading before we can grow more confident that a full blown recovery is underway.” In other words, “we need to see the recovery confirmed before we can confirm the recovery”. If that’s all you have to say, why not just become a reporter? Oh, right, the salary difference.
@Miguel Sanchez
Bank economists main perogetive is to there traders, so I assume that there discussion will be significantly stronger when talking to what is effectively their client base. When it comes to releasing reports like this it is important to remember that it often functions as advertising – so they are just trying to give a generic, and sometimes yes meaningless, discussion.
Also it is important to try and remember what we mean when we say economists forecast accuracy is not important – we are not saying that the art of forecasting isn’t important, just that it is incomplete. The value should come through the discussion regarding risks and the pointing out of important factors – rather than a blind reliance on figures.