Let’s say you are obsessed with GDP and get so darn excited for every new release of GDP that you just can’t stand to wait a whole three months for every new quarterly update. That means you are probably either an economist who needs a hobby or someone who for some weird reason writes a newsletter about things economists care about.
In any case, there is hope! Because, somewhere along the line, a really impatient economist decided to start making estimates on what GDP was going to be before it was released. Over time, more and more people built increasingly complex ways to essentially predict the future and figure out what GDP was going to be any given quarter.
Like all predictions about the future, they are completely made up. However, some use a lot of math and people with PhDs go “wow would you look at all that math” and hype them up to their PhD buddies.
Some of these forecasting models are actually pretty good when it comes to predicting the future (i.e. not catastrophically wrong). The idea being, the future is impossible to know precisely but some guesses are better than others.
So, here is one particular system of guessing (aka a lot of math) done by the Federal Reserve Bank of Atlanta called “GDPNow” and it’s basically the hottest thing to come out of Atlanta in recent memory if your an economist and like predicting things:
The dark line is quarterly change in GDP, which was just covered in the last post. The lighter line is the final estimate for GDPNow, which is made typically one day before the actual release. So, the way to think about the GDPNow model is, with all the information available in the market at a point in time, what is the best guess for GDP?
Making a guess one day before the release of the actual isn’t that fun or interesting. So, what the model does is update throughout the entire quarter. As soon as one GDP release comes out, GDPNow starts guessing what next quarter will be. And, as new information comes available throughout the quarter, the model updates its guess. Somewhere along the line, some savvy marketing insight inspired someone to call these kinds of forecasts “Nowcasts” because, well, it’s a forecast right now.
As can be seen based on the time in the chart, this model has been around for several years, so there is a good amount of data to test its performance. The model has shown to improve the accuracy of its guesses as the date of release draws closer. That is something you would definitely expect from a model like this but is actually kind of hard to achieve quantitatively.
Ultimately, the two lines above aren’t perfectly in sync but it’s not bad either. It is definitely better than nothing! And, it’s kind of fun to watch. When good or bad economic data comes out, it’s interesting to see how the GDPNow guess fluctuates in response. Well, it’s interesting to me at least. And a bunch of economists (and they are a tough crowd to impress).
I’ll talk a little bit about how I’m going to track GDP going forward in the next post. Because, good news, the next GDP release is just a few days away so we can talk about the latest things happening in the economy!
Related content:
What’s Next for Endless Metrics
Quarterly Changes in US Real GDP