What’s a Used Car Worth?
We want to replace our 2001 Range Rover. The two cars we’re considering are:
- 2006 Toyota Land Cruiser
- 2006 Land Rover Range Rover Supercharged


This past weekend we went to Carmax, then to a local dealer, drove cars, endured the mandatory high-pressure sales pitch, and went away to run the numbers.
So the age-old question is: how does one really know what a used car is worth? There are so many variable (features) to obfuscate the true value of the product (used car). Another confusing factor is the difference betwen Kelly Blue Book (might sell price) and actual auction value (what vehicles actually sold for). Lastly, the dealer has no incentive to share any market pricing indicators so an uninformed buyer has few good options:
- Haggle a bit (makes you feel good but has no material relevance to the deal)
- Finance for a “reasonable” monthly payment (lousy idea)
- Do your homework and come prepared (good idea)
The approach I took is to collect actual sales transactions for the model year and plot them versus mileage. My hypothesis was that a trend line would appear in the data to illustrate a fair-price “zone”. I considered that I could then use this to set a baseline to haggle against, knowing where my acceptable deal guard rails were.
Here is the data and my interpretation of what I found:
Land Rover Range Rover
Pricing appears to be driven by model year and engine type, with little variance for features like DVD or sport wheels. My guess is because this is a luxury, aspirational, premium brand driven by HENRY men. Relevant pricing variables are model year and engine type. Mileage is almost not a factor: for lease trade-in years (2005, 2006) the curves are almost flat between 20 and 50,000 miles!

Toyota Land Cruiser
Pricing data is harder find but appears to follow a simpler trend because of reduced product variance: Land Cruisers have little to no pricing variance by feature set because there are few factory options available (they come fully stocked). My interpretation of the lower transaction count is that Land Cruiser customers buy and keep these cars rather than lease (I still drive a 1997). There is an clear relationship between price and mileage, while model year provides a vehicle grouping function along the curve.

Summary and Conclusions:
More information is allways good, and this approach seems valid. Transaction history and trending seems to provide an indicator of the type of product (aspirational vs practical) plus appears to illustrate an average fair market value.
For my purposes the data illustrates that if I can get the total transaction price for one of the two cards below the curve then I’ll have negotiated a better than average price.
Lastly, if I negotiate a great deal the data also provides an indicator of transaction volume to measure just how well I did.
References:


