Hacker News Reader

Comment by pinkmuffinere | Hacker News Reader

You're reading a lot into the wobbles in the plot, when you probably want to be looking at the total change over the last 5 years. Here's the change by city, Q4 2019 - Q4 2024:

Austin: 346 -> 510. 510/346 = 1.47

Dallas: 192-> 295. 295/192= 1.53

NYC: 203-> 318. 510/346 = 1.56

SF: 271-> 361. 510/346 = 1.33

LA: 291-> 443. 510/346 = 1.52

These numbers are all pretty close imo, clustered around 50% growth in the last 5 years. Austin in particular grew 47% over the last 5 years, for a annualized growth rate of ~8% -- that is some _crazy_ growth for real estate, when the long term average in the US is about 3% annualized. IMO SF is the only remarkable datapoint, with 33% growth -> ~6% annualized growth. Even that is still high though.

It would be easier to visualize if we can plot them all on top of each other, but I didn't find an easy way to do that.

View on HN

The problem with such calculations is that they're very sensitive to the chosen starting point: compare the growths in the most recent 3 instead of 5 years:

Austin: 502-> 510 growth: 1.6%

Dallas: 263-> 295 growth: 12.2%

NYC: 255-> 318 growth: 24.7%

SF: 349-> 361 growth: 3.4%

LA: 384-> 443 growth: 15.4%

If we're looking for a trend that follows a specific city instead of specific year and city, then I agree. For that purpose I think I'd like to run this same calculation for the 5 year interval 2019-2024, and 2018-2023, and 2017-2022, etc, to see if there is a trend that is unique to the cities, as opposed to the cities in a certain year.

However, my interpretation of the original claim about Austin is "Austin house prices have grown less in the last 5 years than other places". That claim is pretty specific, and can be examined by looking at just the start and end points. Looking at the end points, it seems incorrect.

I ran into this problem yesterday trying to calculate changes in federal spending over time. One user claimed that inflation-adjusted spending had sharply declined between 2000 and 2010 - which was true. But I had already calculated that spending had sharply increased from 1990 to 2023. I couldn't understand how we had reached such different conclusions until I looked at the IRS data book and saw that there was a slump in 2010 (presumably due to the effects of the Great Recession) and a huge spike starting in 2020.

"Lies, damn lies, and statistics."

You're looking at the start and the end of the plots and totally ignoring what happens in between. If Austin had followed the linear trendline from Q4 2019 to Q2 2022 (increase of 22.7/Q) then it would have been at 800 in Q4 2024.

800/346=2.3

2.3 >> 1.47

That's not a wobble in the plot.

Reply on news.ycombinator.com