Hacker News Reader

Comment by AnimalMuppet | Hacker News Reader

Which, arguably, shows that economists were not as useless as a group over 2008-2018 as they were over 1929-1935.

Why not? Minimising these impacts it's about finances, not economics. I.E Economics is about predicting these things (which they absolutely didn't), finances is about putting mechanisms in place so that if they happen they will be minimised.

The vast majority of economics is not macroeconomic forecasting. That's somewhat akin to saying that computer science is about making email work.

Sure. But the ones that were about macroeconomic forecasting totally fucked up. That's somewhat akin to saying that the IT department responsible for managing email is about making email work. Not such an unreasonable complaint any more. Particularly galling that we had "crazy" economists claiming that we were in a massive housing bubble which we should expect to burst back in 2006, while at the same time the economists at the Fed were saying, "Look at interest rates! Now is a great time to buy a house!" and coincidentally their mates from college were selling mortgage backed securities. So really its more akin to the IT department responsible for managing the email deliberately sending Nigerian fishing scams to their fellow employees along with the public message of "No, outlook is fine, just click on anything in your inbox!"

Economics is not about predicting, it's about understanding things.

It's basically impossible to do the former at scale for time periods over 6 months, we're slowly getting better at the latter.

One big reason economic forecasting is so hard is that agents in the system you are forecasting are taking your present period forecast into account when acting in future periods.

Plenty of people I know were predicting the crash of 2008 at least a year out - sometimes more.

Fortunately, none of them were qualified economists.

Macro is actually pretty straightforward to do.

But economics is politics by other means. It is not a science - it's a branch of rhetoric and persuasion (i.e. propaganda) and is used to disguise and rationalise purely political decisions that would otherwise be impossible to justify.

Indeed. Take Brexit. Government economists stated as fact that there would be 400,000 job losses following a vote to leave (note: the vote, not actually leaving itself). Absolute bloodbath. Reality: the economy grew.

Everyone knows economists say what they're paid to say, which is why they're so often ignored. Oddly, the people who tend to understand that best are often the ones derided as the "uneducated" voters and the people who continue to cling to their faith in economic modelling despite all the evidence it's worthless are often the "educated".

If you want to find an economist to support some statement that's easy, but does not mean that it's actually what most economists actually believe.

It's like saying 4/5 Dentists recommend X by asking them if it's better to brush with crest or not brush at all. Or pay some group of 5 dentists that work for you and ask a more neutral question.

If you cannot predict, you do not really understand.

A coin flip with 50/50 odds is understandable even without predictions.

This is a technicality / edge case of 'pure randomness', rather than a rebuttal.

Furthermore, I can predict the distribution of results for a coin quite accurately over long time timeframes. Indeed being the 'understanding' of a successful gambler means being able to make predictions about what will happen with a random process over time.

Roulette for example is analogous to the coin toss; most casinos seem extremely able to predict what will happen over time quite well.

Or are you hinting at a counterargument that most of the world is a truly random process analogous to a coin throw?

In which case, your experience with reality must be terrifying. Will the sun come up today? Who knows? Will gravity work in a minute? And so on.

You can model a coin flip you can't predict it. Next year the US GDP with be the same +/- 50%. That's been true for the last 200+ years, further there is positive bias so it's more likely to be + than -, but the down tail is thicker so it's more likely to be down 9% than up 9%.

Similarly casinos will sometimes lose lot's of money on Roulette when someone makes a big bet. But, the their long term average is positive money maker. Further, they don't pick specific returns just a range with some probability aka more than X not 10,203,556$.

What you want is a prediction for next year, not a model for next year.

> One big reason economic forecasting is so hard is that agents in the system you are forecasting are taking your present period forecast into account when acting in future periods.

Meh, that's not terribly convincing, as far as excuses go. For GDP predictions, for example, there are many competing institutions publishing forecasts for any number of countries, sectors, etc. It'd be impossible to take them all into account, especially when they don't agree.

I'm also not sure if "taking it into account" wouldn't actually lead to the opposite: self-fulfilment of those prophesies. After all, the prevalent reaction to an expectation of high growth would be to invest, thus creating that growth.

Right. What I said is generally true for Central Bank forecasts, because they have an effect on interest rates (which they set). But the expectation of change in interest rates changes lending and borrowing behavior now, etc.

That said, it's generally doable to do serious forecasts for a few months in advance, but the error bars explode after that.

Reply on news.ycombinator.com