The Collectible Car Asset Class
Executive Summary:
Very large and highly heterogenous, the collectible car market is ripe for sophisticated quantitative analysis
Measurements of value can be inferred on a relative basis, much like in the bond market
Diversification can be an expensive trap due to the structure of the negative carry
Demand and supply drivers of scarcity have different long-term effects on values; supply-driven scarcity may provide more durable support
The recent strength of nominal secondary market values has been a function of macroeconomic conditions
Car culture is important in supporting values but introduces a free rider problem
The market structure is complicated by information asymmetries and expensive liquidity, but opportunities arise from this
Introduction
There is much debate about whether collectible cars are an investment. This really depends on your definition of an investment! Intuitively, to the extent that people buy cars for enjoyment and utility, it makes sense that prices are bid up high enough that the expected financial return is negative, or at least lower than other asset classes.
Nonetheless, the collectible car market cap. (total value of all cars outstanding) is very large and therefore the asset class is worthy of some quantitative analysis. A back of the envelope calculation: there are currently around a million Porsche 911s in existence and if we assume they have an average value of $100k, the implied market cap is $100bn. This would be a top 10 FTSE 100 company and larger than 386 of the S&P 500 constituents.
During my ten years working as a bond trader, the similarities between analysing financial markets and the classic petrolhead activity of perusing the classifieds became more striking. Looking through many ads to work out which cars are the best value, or infer what particular options are worth, is a form of relative value analysis. The two car garage conundrum is a constrained optimisation problem.
Commentators in any market love to attach narratives to price movements, however, often these are distinct from the underlying flows that cause them. Many of the dynamics or events affecting car prices are macroeconomic; to take an overall view on the car market is to take a view on things like the money supply, interest rates, inflation, equities and credit spreads.
So I thought it was worth diving deeper into the characteristics of the collectible car market to understand what we can learn from other asset classes.
The pricing problem
Collectible cars are heterogeneous: there are many different models and specifications, so the total number of parameters and combinations is huge. As a simplified example, if there are 25 different options (binary and independent) available when a car is new, then the number of potential combinations is greater than 33 million (225). Multiply that by 15 colours and the total number of potential combinations exceeds 500 million. It is fair to assume that the market is inefficient at pricing specific characteristics because it is computationally hard and rarely is a pure comparison observed (i.e. the contemporaneous sale of two cars which differ by only one parameter). This is the background for considering the relative value of different cars: whether the £80k green car with steel brakes and 15k miles is better or worse value than the £80k white car with ceramic discs that has done 20k miles.
Relative value analysis
Across financial markets it is common to transform prices into ratios which enable different assets to be compared like-for-like. For example, the price to earnings ratio (share price divided by the earnings generated for each share) enables an equity investor to see how much they are paying for a dollar of earnings. In real estate, there is the price per square foot. Of course, these are crude measures, and it takes some consideration to understand whether or not these values are justified by all the other characteristics of the asset.
It is hard to think of a ratio for the collectible car market that is easy to calculate and universally relevant; specific comparisons are often more interesting. To analyse whether a 911 GTS is better or worse value than its GT3 sibling, you could look at the ratio of GT3 to GTS prices across the different 911 generations and consider whether this is consistent with your opinion of which models are most special. Likewise, the ratio of the price of a delivery mileage example to a car with average mileage, calculated for different models and generations, would give an indication of which “collector grade” example is better value.
There is a lot of room to use sophisticated quantitative methods for this type of analysis, which will always remain an enjoyable part of the market: deciding what to buy! To a great extent, whether or not ownership of a car generates a return will be determined at this stage as the relationships between prices change over time. This analysis should also be used to determine when and at what price to sell a car, as well as how to best pitch it.
Modelling the cash flows
In comparing collectible cars to other asset classes, one must consider the cash flow profile of owning the asset. In this sense, cars are most similar to commodities, where there are ongoing costs for storage and insurance. These are of course not the only costs! This negative carry is also negative cash flow, so the total invested capital (the breakeven point) grows over time.
To the extent that the enjoyment of cars in the UK is limited in the winter, the negative carry implies that there should be quite a large degree of seasonality in pricing. This dynamic is also seen in commodities markets: the timing of delivery is important.
There is one important inference to make here which is counter to conventional financial wisdom: it is not at all clear that diversification pays, even in risk-adjusted terms. Most costs associated with owning cars are only partially proportional to the value of the car (e.g. insurance and servicing) while others are totally fixed (e.g. storage). Therefore, the cost of carry for four £100k cars is almost certainly worse than the carry of one £400k car, especially if that car is something like a 911 S/T.
Given negative carry, it is doubly as important that values of collectible cars remain supported. This is where there are some tailwinds.
Value from scarcity
An important driver supporting the prices of collectible cars is scarcity, which can be split into two categories. Firstly, regulated scarcity (supply side), which has emerged not only as an emissions issue discouraging supply of larger engines, but also as a weight and user experience penalty on new cars. Manufacturers are being forced to optimise for increasingly restrictive safety rules which is fuelling the narrative that “peak car” was a decade or so ago. Supply of the most collectible cars that are lightweight and have an analogue feeling is therefore genuinely capped. One must be careful to consider whether this narrative is running faster than reality though; Porsche and Ferrari are currently building more ICE sports and super cars than ever.
The second type of scarcity is driven by consumer behaviour (demand side) and hence the long-term effect on prices is more ambiguous. The most obvious example here is the manual gearbox, which was discontinued by many manufacturers for economic rather than regulatory reasons. The share of cars sold with a manual gearbox started falling quite fast from the mid-2000s with the popularity of paddles. The share of learner drivers in the UK who only learn automatic has risen from 6% to 26% in just over a decade, and many will probably never drive a manual again. While manual sports cars are currently trading with a premium due to demand from enthusiasts, demographic and consumer trends will likely limit their terminal values.
Recent macroeconomic drivers
The supply of collectible cars could be modelled like Bitcoin: increasing at a decreasing rate, tending towards a cap. Mining becomes harder over time just as regulation is making the production of new sports cars trickier. We don’t know the ultimate supply of cars as we know them today, however, it will likely be reached well before 21 million Bitcoins have been mined.
To what extent does finite supply lead to higher prices, though? Everything else equal, nominal prices will rise because inflation is positive, and the money supply continues to increase. This effect is not the real value of the car rising, it is the value of the currency it is denominated in decreasing (the debasement trade). This is the attraction of owning a real asset immune from digital and monetary shocks.
Between January 2020 and December 2025, the broad money supply (M3) in the UK increased by around 29%. The monetary impulse was even greater in the US, where the M3 money supply increased by 45% over the same period. Monetary, fiscal and geopolitical (sanctions) policies have supported the prices of real assets globally over recent years, as demonstrated by the more than tripling of the gold price since early 2020. It follows that collectible cars have also been a beneficiary and could explain why prices in the US have risen faster than in the UK.
The importance of car culture
For the value of collectible cars to remain supported independently of the macroeconomics, the buyer base must grow. As an alternative asset with an uncertain financial return, the aspiration of ownership must be passed down, and the current enjoyment of ownership must offset the negative carry. Here, the short-term trends are very positive, with social media and ever more events enabling the enjoyment of cars to be shared. These need to offset long-term headwinds from the decline in young people’s interest to drive, and owners and the industry must work together to make running the cars as sustainable as possible.
Likewise, the trade-off between preservation and usage must be managed, especially considering the inherent free rider problem. The most profitable (but not necessarily enjoyable) strategy is to hide a car away from the elements and rely on others to champion the asset class. Yet if all owners were to do this, cars would be reduced to mere financial assets with negative carry and presumably trade at very low prices. Speculators may cause short-term price movements, but long-term value is based on an engaged owner base that is willing to enjoy their passion with others.
The market structure
The trading characteristics of bonds and cars are really quite closely matched, perhaps unsurprisingly as both markets are made up of a large number of similar but non-fungible assets/securities. They are both traded over the counter (OTC) rather than on an exchange, exacerbating the inherent information asymmetries both between buyers and sellers and between those aware of the trade and those who are not.
The dealer margin (bid/offer spread) is very wide in this market for a couple of reasons. Whether or not they are leveraged, a dealer must apply a discount rate to the assets on their balance sheet to cover the costs of the sheet or at least manage the opportunity cost of using the capital. If a car is expected to take six months to sell, a 10% discount rate implies a ~5% haircut purely to cover balance sheet costs. This was one of the mechanisms by which the abrupt rise in overnight rates between 2021 and 2023 reduced liquidity in this market.
Furthermore, the job of a trader in any asset class is to manage adverse selection. They are most likely to be offered poor quality cars and will find it easiest to attract buyers of the highest quality cars. Therefore, the best cars will turn over very quickly while the worst ones will linger on the balance sheet — they will be long (own) lower quality stock than the average car they sell, and the positions should be expected to bleed (depreciate).
So the dealer is in a constant battle to overcome these dynamics by charging a margin and making good decisions about which cars to take into stock (or how aggressively to bid any given car offered to them).
Sale or Return (SOR) may seem attractive, however, it is certainly not an arbitrage. To offer a car for sale in this manner is to be an unsecured lender to an unregulated business that in many ways operates like a bank, holding illiquid assets funded by short-term liabilities and often only a thin layer of equity. In general, being a creditor has an asymmetric payoff profile: a large probability of a small profit plus a low probability of (total) loss. The expected return from lending a car in this context is often limited when considered as the product of the potential saving and the probability of sale, which are often inversely proportional to each other. That said, lending to a well-capitalised institution for a reasonable benefit (the credit spread) can of course be a good trade.
There are many other ways of engaging with the market which can simultaneously optimise the price, reduce the cost of trading and help the dealer make high quality (less risky) PnL. After all, a trading desk in the City does not make all its PnL from unsolicited “flow” trades, even if these can initially appear to have high bid/offer spread!