Getting mileage out of motor vehicle sales data

For most people, ownership of motor vehicle represents one of their three biggest financial commitments and is frequently the single most valuable asset they own. It follows, therefore, that motor vehicle sales data holds some significant information about economic momentum – both absolute and relative – as well as consumer sentiment and demand for related inputs such as steel, oil and aluminum.

The broad links between macroeconomic data and motor vehicles sales have been explored from the perspective of the former’s impact on the latter in a number of papers, such as Shoaib and Sanjog’s Automobile Sales Forecasting and Correlation with Economic Indicators: A Comprehensive Intra-Region Case Study. But the question of what those sales mean for investment models has received less attention.

CEIC, an ISI Markets product line, tracks motor vehicle sales in multiple markets. This creates an opportunity to apply tried and tested quantitative frameworks to high quality data while potentially uncovering some investable signals.

Although the data provides a foundation for constructing country-level models whose signals dictate the optimum level of exposure to that market, this Quant’s Corner explores building a cross-sectional model from a universe of countries that optimizes a fully invested strategy.

Both emerging and developed markets will be utilized in this analysis, to capitalize on CEIC’s differentiated data and improve the model’s resilience.

Ten percent down on the weekly model

To create this universe, all markets that were part of the ACWI index during the relevant periods were included. Those countries are Japan, South Africa, Canada, Thailand, the Philippines, Indonesia, Brazil, Mexico, Vietnam, India, Malaysia, South Korea, Argentina and Turkey.

The source of the data for each country is listed below:

For each market, the month-over-month growth in domestic sales of motor vehicles is computed. Those values are subject to adjustment when and if the underlying data is officially revised.

Based on the sales growth data, countries are bucketed in quintiles. Returns are then observed over a range of holding periods, by going long the top quintile and short the bottom quintile.

With the data points refreshing on different T+ frequencies, versions of the model can be run dynamically on both weekly and biweekly cadences. As it turns out, the weekly frequency performs best when back testing, with the signal decaying rapidly beyond the monthly frequency.

When looked at over the past decade, annualized returns have ranged from the eye-catching (2015. 2022 and 2024) to the underwhelming (2018-20, 2025).

In the case of both 2019-20 and 2025, consumers were dealing with higher interest rates than they were accustomed to and external shocks (Covid-19, conflict in the Middle East, tariffs) that affected their perceptions to a greater degree than the purchase of a vehicle.

Fine tuning the engine of growth

Is there more to uncover with this data? A simple charting of sales against consumer confidence for one market, in this case Mexico, suggests that there are correlations that can be explored and modeled at the single country level.

One option is to build upon existing methodology for some of ISI’s in-house index models. These generate signals that can be used for market timing.

There is also scope for more granular analysis, breaking the broad motor vehicle category down to isolate sub-categories such as electric vehicles (EVs), small trucks or luxury vehicles and developing models based on their sales data.