Do you get ripped of when buying a car in Munich?

(Used) Electric cars in front of BMW’s largest german plant Dingolfing.

Ever thought you got ripped of after buying a used car? Well in case you ever did buy a used car, it is quite likely you felt that way. One question, that might come in mind is, whether it’s worth to travel farther to buy a car if you live in a congested area. This and other questions will get answered in this article.

So, as you already might know, Germany is a car-nation. Not only it is home to several of the worlds largest car manufacturers, like BMW, Mercedes-Benz and VW, there is also a huge number of car owners. According to the Kraftfahrt-Bundesamt, the german authority responsible for cars, in the beginning of 2019 64,8 million cars and trailers with an average age of 9.5 years were registered. A huge market like this also has a big market for used cars. In the following, this market is being analyzed and several questions shall be answered by having a deeper look at the german used car market by analyzing the “Used cars database” from Kaggle, supplied by Orges Leka in 2016. The dataset contains over 370.000 used car adds scraped from ebay-kleinanzeigen, which is kind of the german Craigslist. Let’s get started, shall we?

1. What are the most common vehicle types on the german used car market?

Sedans (Limousine, 29%) have the biggest market share in Germany, for somebody from the U.S. this might be surprising, yet in Germany sedans are quite a common view. They are followed by compact cars (Kleinwagen, 23.3%) and station wagons (Kombi, 20.4%). Less common are busses, convertibles (Cabrio, 7%) coupes (5.7%) and SUVs (4.5%). But, judging by a look on the streets, the SUV share has to have increased dramatically over the last few years.

Sedans have the biggest market share in Germany.

2. Which brands are predominant on the german used cars market?

As mentioned before, Germany is home to quite some manufacturers from the premium and medium segment. BMW, Volkswagen and Porsche are only three examples. But do the germans prefer those brands over others, and if so, what are the favorites? Well, it does look like the germans prefer their own cars:

Brands on the german car market, sorted by number of adds.

Germany’s five most common brand on the used car market are, as a matter of fact, german brands. Nr.1 is Volkswagen (21.4%), followed by BMW (11.3%), Opel (10.5%) — which at least used to be german — Mercedes-Benz (9.7%) and Audi (9.3%). Those german brands are followed by american Ford (6.7%), french Renault (4.5%) and Peugeot (3%), italian Fiat (2.4%) and spanish Seat (1.9%).

3. What is the average age of cars on the german used cars market?

As mentioned before, cars on the german market are in average 9.5 years old and are thus relatively young. But is this the same for the used cars market? Well, no. Used cars on the market are in average 12.9 years old, while the 25%-percentile is at 8 years, the median at 13 years and the 75%-percentile at 17 years. The maximum age that could be found is a lot older though: 105 years.

Distribution of car ages on the used car market with an age younger than 30 years.

Interesting to note is that there is quite a lot cars with an age of less than 1 year. An explanation for this is that a lot of cars are sold as one-year-old used car or annual employee’s car, for the reason of tax benefits.

4. Are there local differences? How are the offers distributed over Germany?

As you might suspect, in larger cities there is a greater number of cars offered. Here we will focus on the 20 cities with the greatest number of cars offered.

The 20 german cities with the most used cars in the market.

Unsurprisingly, Germany’s largest cities Berlin (4%), Hamburg (1.9%) and Munich (1.3%) have the most cars on the market. What is a lot more interesting is, that there is also quite big differences in prizes, motorization and age. But what about the pricetag?

Munich’s used cars are a lot more expensive than those of of other cities.

As can be seen above, Munich’s used cars are in fact more expensive than those of other cities. But it’s not easy as that, there are other differences: While for Munich the median pricetag is at 5490€, the median age at 11 years and the median power at 136 horsepowers, in Saxony’s Leipzig the median price is at 2999€, the median age at 14 years and the median power at 116 horsepowers. So Munich’s cars are not only more expensive, they are also newer and faster. The complete data can be found in the table below.

Table with median prices, ages, motorization and mileage for the 20 german cities with the most cars on the market.

5. Concerning motorization: Which fuel-systems are preferred? What is the average motor-power?

For quite a while, there is a discussion about diesel-fuel going on in Germany. But let’s not focus on the discussion, but the data. What is the actual share of diesel-cars?

Gasoline (benzin) is by far the most common propulsion system in Germany.

The data is more clear than the debate: By far gasoline (benzin) engines with 65.7% are the most common propulsion system, followed by diesel engines with a share of 32.3%, But wait, that‘s not all! Liquified petroleum gas (lpg) surprisingly has a share of 1.6%. Less common are natural gas (cng), with a share of 0.2%, and plugin hybrid electric vehicles, electric vehicles and others with a share of less than 0.1% each.

What about the gearbox? Other to drivers in the U.S., german people highly prefer manual gearboxes (77.5%) over automatic gearboxes (22.5%).

6. How are sale prices distributed on the used cars market?

There is a high spread when it comes to prices, yet the cheaper prices are significantly domiant: The median price of a used gar on the german market is at 3393€, the min at cheap 0€ (okay…) and the max at not so cheap 999999€.

Distribution of sale prices on the used car markets, cut at 50000€.

7. Putting it all together: is it possible to predict prices?

To get back to the start and your feeling you got ripped off: Wouldn’t it be great to have a tool, where you can enter some basic probabilities of a car and get back an estimated price? Is that possible?

In fact, it is. And using Data Science, regression algorithms and the dataset above — I coded such a tool in Python and made it available to you just below this paragraph. You just have to press the green “run” button and follow the instructions in the command line. The first time this might take some seconds until the necessary packages are loaded.

So every car seller will use this tool in the future to set prices? Well, maybe not. Since this tool does not work magically, but was trained on the available data, it’s accuracy is not 100%, but is at R²=0.76, that means 76% of the variance in the data can be explained by the model. And it should be possible to further improve the model, for example by using more data or adding additional features. But it definitely is good enough to give you a qualified estimate, which price is about justified for a car and prevent you from getting ripped off. And that is something, isn’t it?

If you are interested in more detailed Information, you can find it on my Github. There you can download the full Jupyter notebook with the analysis as well as the code and model for the applet above.