Daniel Rathbone – Technology, Travel & Adventure

The snowboarding season has arrived in the Pacific Northwest and I’m hoping to buy an SUV to help me get up to the mountains.  As a good computer scientist, I want some data to help inform my selection of a used car.  To that end I wrote a little program that scraped some pricing data from Seattle’s craigslist, and here’s what I found.

Craigslist car ads last for one week, and I narrowed my search to three models of SUV.  There were about 1700 listings that fit these criteria.  My little program, written in F#, was able to extract pricing and model-year data on about 1500 listings.  Mileage data was missing from many listings, but I managed to parse out about 280 usable numbers.  The data shows some interesting trends, but there’s a caveat: this data displays listing prices – the asking price for these vehicles – which is different (probably higher) than the eventual sale prices of the vehicles.  Additionally, craigslist ads are not externally verified, so there’s no guarantee that the sellers listed mileage data truthfully.  With that said, here’s some pretty charts:

Model-year vs. Price

Naturally, vehicles depreciate as they get older.  This data indicates that these SUVs depreciate about $2500/year for the first 10 years, and after that they retain their value.  One can extrapolate that the buyers and sellers in this market expect that these SUVs will operate reliably for the first 10 years, but after that the value of a vehicle must account for recent or expected replacement of mechanical components.  Indeed, many of the listings for older vehicles mention rebuilt engines or other major maintenance.

It appears that a Toyota 4Runner retains its value better than the Nissan Pathfinder, which in turn retains its value better than the Jeep Grand Cherokee.

Mileage vs. Price

Vehicles depreciate as they’re driven.  This chart indicates that the relationship is roughly proportional.  The three models under consideration seem to depreciate $7500 for every 50,000 miles, but the correlation is not exact.

To return again to my personal situation, I’m simply hoping to find the best value for my money.  I see that the chart of price-depreciation-per-year regresses much more neatly than price-vs-miles, so I’m wondering if I look at older vehicles with low miles, I’ll find a better-than-average value.  Based on this data, my strategy is to find a 10 year old 4Runner with less than 100,000 miles for around $11,000.  If I’m lucky, the thing will run for a few years without mechanical failure and retain most of its value, which would yield very good value-for-money.  If I’m unlucky and I need to drop in a few thousand dollars on mechanical repairs, then I still expect this will yield better value-for-money than a newer car that depreciates more rapidly as the years go by.