The advantages of using hedonic regression to measure the value of standardised technologies

The recent deployment of 5G and other standardised technologies means that measuring their value has become a question of central importance, both for industry participants and courts around the world. Courts typically rely on various methodologies to value standardised technologies, including comparable licences and public declarations on royalty rates. Hedonic regression has proven useful in many recent litigations, including that in the High Court of England & Wales between InterDigital and Lenovo.

Hedonic regression is a type of regression analysis, where one seeks to measure the relationship between two variables (called an independent and a dependent variable), while holding all other factors constant. For example, one may ask: “How much more income do workers earn as they age?” – the goal being to characterise the relationship between income and age. However, because income is affected by many other factors (eg, education, talent and occupation), a regression analysis is needed in order to filter out the effects of these other confounding factors.

Hedonic regression seeks to characterise how the price of a product varies based on product attributes and other factors relevant to the sale (eg, the time and location of the sale). A common example is the analysis of house prices: all other things being equal, buyers pay a higher price for a house if it has more bedrooms. Hedonic regression helps precisely identify what is meant by “all other things being equal”. To this effect, it measures the value of each feature making up a product, controlling for other factors that also affect prices.

It is especially useful in helping to explain ‘fair market value’ – the price at which a product changes hands in an informed, willing, arm’s-length  exchange. Hedonic regression utilises the data taken from such transactions to explain the prices that emerge from them. Thus, it enables a factfinder to identify the factors that determine fair market value, in particular, in ascertaining whether a given factor (eg, the presence of 5G), is a statistically and economically significant determinant of a product’s price.

Combining these ideas, in running a hedonic regression we seek the best equation that describes the variable we are trying to explain – for example, the price of a cellular handset. Suppose that a court is tasked with evaluating the fair market value of 5G cellular technology over 4G. We can consider the following hedonic regression of cellular handset prices:

(log price) = 0.11 (screen resolution) + 0.16 (5G) + 5.1

A handset price is explained by many other features – although, for demonstration purpose, we only show results of two features. In addition, 5G is, still in the early development stage - the estimated 5G premium given here may well change, as it continues to develop and more data becomes available. In the above equation, the variables “screen resolution” and “5G” have been used to explain (the logarithm of) handset price. The values 0.11 and 0.16 are, respectively, the effects of screen resolution and 5G on handset prices (often called coefficients). Our coefficient of interest is that of 5G. Here, screen resolution (and other similar features) are included as control variables, playing a similar role as the variables of occupation and education in a regression of income on age.

In this example, the presence of 5G raises handset price by approximately 16%, as compared to the same handset without 5G capability. The 5G coefficient reveals how much more (in terms of percentage of handset price) customers pay for a 5G handset than for an otherwise identical 4G handset. Hedonic regression isolates the value of 5G from that of newer hardware and other technologies that may accompany the standard in the handset. We call this the 5G premium.

Hedonic regression also addresses the apportionment requirement specified in Ericsson v D-Link: the value of the standardised technology must be apportioned from the value of standardisation. It achieves this goal by comparing the prices of handsets with two different standardised technologies (in our case, 5G and 4G) that both already incorporate the benefits of standardisation.

In hedonic regression implementations, we have found the 5G premium to range from 15% to 22%. The exact estimate depends on the specific regression configurations, such as the data source used, the time period and set of features included as control variables. Even though these results are not considered final and are subject to change in the future, especially as new data becomes available and as the 5G standard matures, it can serve as a reliable starting point for a top-down analysis. In particular, this top – which is backed by market evidence – is higher than other values sometimes used by a court (eg, the 6% aggregate royalty burden in TCL v Ericsson).

 This methodology has proven useful for valuing standardised technologies in multi-feature products. It has also been shown to be a more reliable top to use in a top-down analysis. Here, we focused on valuing the 5G standard in handsets. However, the applications of hedonic regression could be extended to many other products and different standardised technologies. The scope of 5G applications can increase to include less conventional products such as automobiles, vacuum cleaners and smart faucets. There would be other complexities to navigate in implementing this methodology to other applications. Notwithstanding these, the hedonic regression is still one of the most reliable and objective ways of estimating the value of any standardised technology.