In my last post I decided on looking for a sedan out of the many different types of cars out there. Now I'm going to turn my focus to the different manufacturers to see which one is best for me. The decision is going to be made based on 3 criteria:

  • Efficiency
  • Durability
  • Value


What does fuel economy look like across all makes?
There's some variation here, but what does it really account for? A one unit change in MPG only accounts for a fraction change of price at the pump. In fact, a growing cadre of car enthusiasts like using a different metric, gallons per 100 miles, to gauge fuel economy in terms of cost of fuel, which makes it slightly more tenable and, thus, more relevant:
(If you can't see much of a difference, you're not blind. The two graphs show the exact same information, the second is in terms of cost.)


This one is a tough quantity to gauge in general, once inside a specific type it becomes even tougher. Like my previous post we are going to depend on two sources of information: supply over time and seller confidence.

In an ideal world we could judge durability based on the life of a given set of cars. But since there is no recorded data on when cars' engines give out, we have to settle for the lives of cars on the market:

Some interesting things emerge here. One, Hyundai, Nissan, and Honda take the cake as far as market share goes. Two, they lack proportional representation in later years (past 6 years old), maybe indicating that owners of older cars either don't want to sell them or the opposite, that they get bought fairly quickly 1. A third conclusion is that these 3 makes don't last long past 6 years of age, but if we take into account how their mileage scores relative to the median, we can assume that this third conclusion is unlikely.
They hug the median for the most part except for the oldest cars, where things get kind of chaotic. We look into this more later.

Let's recap this idea. A high supply by itself doesn't necessarily correlate with durability, but a relatively high change in supply could say something about buyer and seller confidence. Furthermore, if the makes whose change in supply is greater than the others have aggregate mileages that are at or above the median durability can't be outright guaranteed, but there's a good case against the opposite.


If there was a metric vaguer than durability, its value. Just like in part 1, I'm going to make the assumption that the aggregate price of a good contains both the buyers' and sellers' appraisal of value. As long as the most inclusive scope of the market is examined, we should have a pretty good approximation of "true value". Median price might be a good place to start:
The biggest difference seems to be among American cars (Chevy and Ford). Volkswagen also has a pretty big gap. Mazda, on the other hand, has the highest resale value out of any car make (We'll take a closer look at Mazda in a bit). Honda, Nissan, and Hyundai all lag near the middle.


For a lengthier explanation of how I examine depreciation, read this part of my previous post. To make a long story short, I'll be defining depreciation as the inverse relationship between price and mileage over time. Basically, the value loss that occurs by driving.
First off, look at that huge drop in value that happens in the first 2 years of ownership. If the accumulation of miles over time is linear (which it is according to the mileage graph earlier) then that means price shrinks by a larger amount in the first 2 years than any other interval of time. In other words, a car that is younger than 2 years loses value faster than any later age.

Second, lets look at the change in PM ratios each year. We can do this by taking the first differential--plotting the slope of each interval:
Each point is the rate of depreciation at that year (look at that Mazda curve, again an anomaly). See how at 5 years the rates converge? We can take that as a sign that cars older than 5 years depreciate at the same rate. So if you find a car you like that is 5 years or older, depreciation doesn't need to be a huge part of your decision since all cars of that age lose value at the same rate.

Chasing a Deal

When I started this project I wanted nothing more than a clear cut winner at the end. Unfortunately the data doesn't reveal one but it does offer insight on how to a spot a good deal. Check out the graphs for price and mileage -- see how the data points become erratic for older cars? To me this indicates a market where sellers lack information both about what other sellers are basing their prices on and about how buyers make purchase decisions. The used car market is fragmented, so there isn't always information to guide pricing decisions. Markets stocked with older cars of relatively low supply could have sellers trying to free up inventory and setting prices low just to clear their lots and make way for more profitable cars. They might sell based on intuition rather than solid research. These are the markets where good opportunities hide. What classical economists call "market inefficiency" is a synonym for lucky break. With enough patience you can find deals like:


High supply might tell us something that other car owners know. At the least it means there are more cars to choose from, so potentially more deals out there. A fragmented market means that sellers don't necessarily know what other sellers are doing. You might end up with a good deal like this one:

or this:

Cult Cars

Remember those anomalies we saw with Mazda? In case you don't, they were

  • Lowest difference between new and used cars
  • Consistently low rate of depreciation
  • Comparable price as other makes

Looks like a cult car to me.

Let me explain. If we accept that price is a measure of value for the majority of market participants but the price of a certain car changes as a different rate despite being comparable to other cars, that means that the people buying this car gauge value on the basis of information not contained in the price. Like a painting or a fine wine, cult cars have an intangible value that only a subset of the market perceives.

On the one hand, a cult car could be a hedge against market uncertainty. Where other cars sell for much less than what they were bought for, Mazdas have selective appeal that buoys the price as the years go by. But buying into a cult means taking a risk on a metric that you may not perceive or understand.


So what do we have at the end of all this? To me, the Accord, Sonata, and Altima are the winners. Fuel efficient, durable, keep their value over time. There's enough supply on the market where a good deal can pop up.

But... is it even worth getting a car at all? Stay tuned for part 3 where I take a look at what you gain and lose on ownership.


  1. Its unfortunate that the same data can lead to polar opposite conclusions. But we luck out here in that both reasons lend themselves to the idea that all 3 makes rank high as far as durability goes.