Chapter 5: Valuation and rating methods for patents and patent portfolio

Valuation and rating methods for patents are intended to make it easier for economic agents to discriminate efficiently among a large set of patents and to early detect the more valuable ones. The econometric literature proposes indirect assessment methods based on observable and objectively measurable characteristics of patents, referred to as patent metrics. These metrics are assumed to condition the rent that may be extracted from patents. Assessment methods are said to be indirect in the sense that the level of the rent is not observed but inferred from surveys, observed behaviours or economic results. Most of these methods were initially designed to characterise the overall distribution of values within a population of patents rather than to assess values at the patent level. Indeed, they initially aimed at getting some insights into the pace of innovation at a macroeconomic or sector level and from a more qualitative point of view than simple patent counts. This chapter reviews research articles that adapt them for the purpose of valuation of individual patents or of portfolio of patents. It is argued that none of the three types of methods described in this chapter is perfect. Methods of the first type, stated value approaches based on an econometric treatment of survey data, are costly to implement and to update and may potentially be subject to declaration biases. As regards the second type, methods based on the valuation by stock markets and more specifically Tobin's Q studies, criticism is based on a background note: in essence, no additional information compared to that available to financiers is produced. Resulting patent scorings cannot therefore be thought of as tools that can help other economic agents than the patent holder to better discriminating among patents. For their part, methods of the third type do not account for the strategic component of the value of patents. Patent renewal methods are particularly illustrative of this third type that gathers revealed value approaches. For the time being, none of the described methods is able to convincingly address the distinction between embodied patents and disembodied patents. Similarly, none of the described methods currently tackles with the valuation of a portfolio of patents, though some of them could theoretically do it. Nevertheless, a striking feature of these different methods is that they all point to the same cautious conclusion as regards the feasibility of an automated patent scoring. Indeed, they all conclude that some patent metrics, and more specifically forward citations, have a significant impact on patent value but that the role of patent metrics in explaining the total variance of patents value is rather too limited compared to that of unobserved sources of heterogeneity.