عوامل مؤثر بر ارزش ثبت اختراع: دیدگاه از نظر تجزیه و تحلیل شبیه سازی

This paper contributes to understanding the determinants of patent value. By drawing on a real options approach, we develop a theoretical model of patent value, which explicitly considers the uncertainty about future value. On this basis, we rely on Monte Carlo simulations with data from a case study in a large chemical firm to estimate patent value according to our model. In the simulation analyses, we compare an R&D project with patent protection and the same project without patent protection. The difference of the values of the two projects is the surplus in profit that may be expected from having a patent covering the project. This surplus is regarded as the value that is directly attributable to the patent. The results of the simulation analyses indicate that the development costs and expected net cash flows of a patent-protected project are higher than of an unpatented project. The higher net cash flows outgrow the increased development costs, and patent value is positive. However, this value is smaller than the overall project value of the patent-protected R&D project.

The valuation of patents is a major managerial challenge, and it has attracted considerable interest of researchers and practitioners for decades [1], [2], [3], [4] and [5]. Starting with the first studies in the 1960s, researchers have examined a variety of determinants of patent value [6], [7] and [8]. For instance, prior studies have examined patent breadth, novelty, disclosure, and inventive activity [9], [10], [11], [12], [13] and [14]. On this basis, several recent studies provide excellent overviews of the literature on patent value determinants [6] and [15]. Despite the theoretical and empirical advances of this growing body of knowledge, however, patent valuation remains a highly relevant topic for several reasons. First, a lot of the literature on patent value focuses exclusively on theoretically modelling the patent system, e.g., [11]. Thus, additional studies are needed to link this line of research to patent evaluation in practice [6]. Second, many studies have focused on assessing patents by means of value indicators, such as renewal information [16], claims [17], legal arguments [18], and many others [19]. Here, prior research has shown that the impact of particular patent characteristics differs with the specific utilization of patents [8], [20], [21], [22] and [23]. Third, practitioners are constantly occupied with optimizing their patent portfolios [24]. The need for thoroughly measuring patents is particularly important in light of recent updates on financial reporting standards, which partly require firms to present the fair value of their intangible assets on the balance sheet [25]. Despite this strong need for detailed patent evaluation, the applicability of patent valuation methods still constitutes a major challenge in many industrial firms. “Few of the companies that could benefit from patent valuations have the capability to perform them in-house” [26], p. 135. In a similar vein, the patent valuation expertise of consultancies and financial institutions seems very limited [24]. As a patent grants protection for a technology that will be used in the future, there is considerable uncertainty in the patent's value [18] and [27]. In light of this uncertainty, many existing valuation techniques either seem to excessively simplify reality or they are so complicated that the possibilities of practical application are limited [28] and [29]. Moreover, most current valuation approaches focus on determining the value of a patented technology rather than the value directly attributable to the patent [6] and [26]. To deepen our understanding of patent valuation, we therefore adopt a perspective that differs from many prior works and complements earlier studies. Thus, we contribute to the literature in several ways. First, we explicitly take into account uncertainty about future value by applying a real options perspective on patent valuation [30], [31] and [32]. Second, we rely on Monte Carlo simulations with data from a case study in a large chemical firm to estimate the value of a patent according to our model. At the current state of research into patent evaluation [3] and [15], the particular characteristics of simulation analyses, e.g., internal validity and experimentation benefits [33], provide major opportunities for gaining new insights. On this basis, our study identifies major starting points for future empirical research into the determinants of patent value. Third, despite using real options and Monte Carlo simulations, we take a managerial perspective that is directed towards developing valuation methods for firms and policy makers [24]. The application of our methods may be directly implemented by practitioners. Thus, this paper attempts to bridge the gap between closeness to reality and managerial applicability in patent evaluation [34]. Fourth, we focus on a patent's financial value as the surplus in profit that derives from comparing a specific R&D project with patent protection and the same project without patent protection. Specifically, we concentrate on a patent that is used for protecting a technology, which is currently developed and will subsequently be applied in a firm's own products. Fifth and related to the last point, we focus on valuing a patent per se rather than estimating the overall value of a patented technology [35] and [36]. The remainder of this article is structured as follows. In section 2, we present an overview of different approaches to patent evaluation. In section 3, we develop a dynamic real-option-based model of patent value. In section 4, we use Monte Carlo simulations based on real data of a patented R&D project in the chemical industry to illustrate the practical usefulness of our model. In the simulations, we compare an R&D project with patent protection and the same project without patent protection to determine the value of the patent per se. Detailed sensitivity analyses deepen our understanding of the determinants of patent value. In section 5, we analyze the study's theoretical and managerial implications. Finally, we discuss the study's limitations and present opportunities for further research.

5.1. Theoretical implications The findings of the simulation analyses have major implications for research into the evaluation of patents and R&D projects. In particular, this article has shown how patent value can be defined as the difference of two simulations. Whereas prior real-option-based research often stopped after determining a monetary value of a patented R&D project, we have explicitly presented a method to value patents per se. In regard to this, the model is capable of determining the value of both a patent that is a prerequisite to a project and a patent that can provide additional sales for a project. The analyses have shown that firms often exercise the option to exit [29]. Thus, many projects are stopped during development [92]. As expected, the average development costs of the unprotected project are considerably lower than the average costs of the protected project. This indicates that firms stop unprotected projects more often than patent-protected projects. Consistent with these findings, the present value of net cash flows of the unprotected projects is considerably lower than the 45% that were used as an input variable for the sales market share. Only in about 20% of all cases, a positive present value of net cash flows is achieved. Here, the large number of projects with values at zero or close to zero indicates that many projects are abandoned in relatively early phases with limited investment. Patent-protected projects tend to involve higher profits, but firms may also incur higher losses with these projects than with unprotected projects. As the sales outlook is more beneficial with patent protection, firms often stick longer to projects with patent protection. In support of this decision, the probability of having a positive project value is much higher with patent protection than without. By contrast, the probability of not realizing any losses is higher for the unpatented project. As it is more likely to immediately stop unprotected projects without any investments, it is also more likely to have a project value of zero. Thus, patent protection seems to offer greater opportunities, but it also appears to involve greater risks because of sticking to relatively unsuccessful projects [93]. Biased decisions and management errors, such as project persistence, can be especially damaging in fast-paced environments with path dependencies of technologies and markets [94] and [95]. Many existing studies have focused on valuing patents that are a prerequisite to a project [29]. By contrast, our model can also value patents that are responsible for additional sales rather than for all sales. On this basis, the simulation analyses have shown that the mean value that may directly be attributed to the patent is €1.10 million in light of the assumptions of our model. Thus, the value of the patent itself amounts to 67.0% of the value of the patent-protected project. This focus on valuing the patent per se besides valuing the overall R&D project extends prior studies [28]. In particular, this examination emphasizes the importance of tacit knowledge that may hardly be codified in patents [96]. This tacit knowledge is crucially important for successful R&D projects, and it helps to explain the difference between the value of the patent itself and the value of the patent-protected project [97]. Accordingly, caution is required in comparing patent value and R&D project value. Here, our detailed sensitivity analyses have shown that the total cost to completion of R&D projects seems to be an important determinant of patent value. This finding reflects that major financial investments, which likely involve a relatively high degree of novelty, increase patent value [6]. Based on a similar logic, annual net cash flows are an important determinant of patent value. Even more important seems to be the high sensitivity of patent value to variance. An increase in the uncertainty of cost and sales strongly reduces the value of the patent and the value of the overall R&D project. This finding emphasizes the critical role of market uncertainty and technological uncertainty in estimating the value of an R&D project, which may involve considerable levels of uncertainty [32] and [60]. On this basis, the additional benefit from having a patent relative to the unpatented project decreases. While we have relied on objective historical data for estimating variance, this parameter generally is a subjective value, which offers considerable opportunities for influencing the results of the simulation analyses. Accordingly, researchers and managers need to put particular emphasis on calculating realistic values for variance, e.g., by means of historical data from prior projects. Concerning the additional antecedents of patent value, the simulation analyses indicate that the weighted average cost of capital and market share are relatively important determinants of patent value. By contrast, the drift factor, the annual probability of failure, and the sales multiple do not have a major influence on patent value. From an integrated perspective, our analyses have shown that net cash flows, variance, total cost to completion, the weighted average cost of capital, and to a limited degree market share have the strongest influence in our model of patent value. While some of these variables have received considerable attention in prior research into patent valuation and R&D project valuation, others have been relatively neglected [32] and [98]. Thus, our findings point to major opportunities for further research. In addition, the results have major managerial and policy implications. 5.2. Managerial and policy implications Despite some limitations, our model is built close to reality, and it considers the influence of uncertainty on patent value. Despite some general difficulties of applying real options logic in practice, which may be overcome, e.g., by means of scenario analyses [90], we have shown the model's practical usefulness by means of the case study. On this basis, the application of the model seems to give practitioners and policy makers the possibility to value patents more accurately than with many other valuation approaches. However, further empirical work is required with a larger sample to confirm the model's appropriateness for examining patent value. Thus, this paper constitutes a step towards facilitating the use of real options methods for valuing patents and R&D projects in practice [34] and [99]. The model as presented and applied in this article hence focuses on applicability, and it aims at providing a balanced trade-off between closeness to reality and usability. In doing so, the model can be transferred to a software platform that may be directly used by executives, analysts, and governmental experts. In contrast to other real-option-based models, its practical application is relatively easy. A particular advantage of our model is the use of historical data as input parameters for volatilities. As such, the susceptibility of our model to personal biases is relatively limited. In general, the strong influence of variance on patent value is somewhat troublesome because variance generally is a subjective value. As Amram (2005) writes, “a model is deemed relevant when it is used by experts and non-experts to attract and speed business transactions” [28], p. 74. To arrive at realistic assumptions and results, we therefore strongly suggest using historical data from previous projects to estimate variance. Moreover, universities, research institutes, and patent offices could conduct empirical studies to estimate the variance of patent value. Prior research suggests that the variance of patent value strongly differs according to the particular field of technology. For instance, the very high value of some critical pharmaceutical patents has often been highlighted [40] and [72]. The results of these studies may provide helpful starting points for practitioners to estimate the variance in the analyses of their organization's R&D projects and patents. Hence, these studies may increase the effectiveness of capturing value from science and technology at a national level [100]. Based on the general estimates of variance, practitioners could specify variance in patent value for their particular organization. According to prior work, some firms tend to have higher average patent values with limited variance relative to other organizations [24]. Besides estimating variance of individual patent values, these studies could examine the effects that patent portfolios have on patent value. Furthermore, there may be considerable differences in the variance of patent value between research institutions, e.g., universities, and industrial firms [98]. Finally, the results of the simulation analyses have underscored the high importance of patent protection. The estimated value of the patent per se relative to the value of the overall patented R&D project emphasizes the benefits that innovators may gain from patent protection [101]. In particular, patent protection broadens the strategy space in capturing value from an R&D project [102], [103] and [104]. For instance, the study of Arora and colleagues suggests that “stronger intellectual property rights can enhance the efficiency of technology transfers, and hence encourage the diffusion of technology, including parts of the technology that patents do not protect” [40], p. 117. While patents facilitate the externalisation and codification of technological knowledge, the remaining part of the project value exceeding the patent value highlights the critical role of additional tacit knowledge [105] and [106]. To appropriate the returns from a patented technology, organizations need additional knowledge, e.g., market know-how, which is not embodied in the patent [107] and [108]. Together, these components determine the value of a patent-protected R&D project [98]. 5.3. Limitations and outlook As with all models and simulations, this article relies on several assumptions, and it can only capture a simplifying view of the real world. Thus, the typical limitations of real-option-based simulation analyses apply to our study [33], [34], [56], [83] and [109]. In particular, future studies could rely on decision tree analyses to move towards analyses of continuous-time settings [34] and [110]. In addition, scenario analyses and options games can be helpful to facilitate the applicability of real options logic in practice [90] and [91]. Nonetheless, our model is built close to reality, and its practical usefulness has been shown by means of the case study. However, this case study is only based on a single development project in a large chemical firm. Although it has been a relatively typical R&D project, further research is needed to broaden the analysis beyond the context of a single firm. Another strong limitation of the model is the high sensitivity towards variance and market share. This is a critical condition because variance and market share are two variables that are subject to personal bias by the model's user. In contrast to other determinants, these variables are not based on ‘facts’. They have to be estimated, whereas, for instance, the weighted average cost of capital can be calculated from company data. In this estimation, the user can willingly overvalue or undervalue the variables in order to distort the results of patent valuation. Therefore, we strongly recommend using historical data of previous projects for the variance and historical data of sales after patent expiration for market share. Although these values can only serve as approximations, they likely reduce the tendency of personal bias. In further research, it could be worthwhile to use this model in additional empirical analyses that exceed our context of one single firm. Accordingly, these studies could examine R&D projects across patent characteristics, across companies, and across industries in order to explore specific differences in the value of patents and R&D projects. These studies could also provide insights into additional environmental determinants that might affect patent value. The competitiveness of the product market, the size of a firm's marketing budget, and the imperfections in the markets for technology are examples of additional determinants that exceed our model. As much remains to be explored, there are great opportunities for further research into the determinants of patent value, particularly regarding additional empirical studies. However, the data collection for these studies might prove difficult because companies tend to be very reluctant in providing the necessary information on R&D projects. Nevertheless, in light of increasing theoretical, managerial, financial, and political interest in the valuation of patents and R&D projects, further research is required to offer a better understanding to firms aiming at profiting from their technology assets.