Patent Examination Procedures and Patent Quality1

John L. King

Economic Research Service

U.S. Department of Agriculture

ABSTRACT

This study examines a detailed panel data set of patent examination procedures that affect patent quality. A main conclusion is that the most important of these inputs (examiner hours and examiner actions) have remained largely consistent over time despite an increasing examination workload. Other measures of examination quality (pendency and interference hearings) have declined. Inputs to examination quality are inversely correlated with the rate at which patents are involved in legal complaints, and the expense of increasing examination inputs may be more than offset by the consequent reduction in litigation costs.

INTRODUCTION

Patents grant the exclusive rights to use, manufacture, and sell new inventions and are widely sought legal instruments. The numbers of both patent applications and patent awards have more than doubled over the past two decades. Explanations for this increase might include more innovative activity, greater emphasis on intellectual property rights among innovating firms, reduced requirements for patentability, or a variety of other possibilities. Regardless of the cause, the increase in patent activity has created a greater examination workload and placed a greater burden on patent-granting institutions. This chapter examines the

1  

Portions of this research first appeared as “An Empirical Investigation of the Economics of Patent Institutions,” Ph.D. dissertation, Vanderbilt University, 2000. The views expressed are those of the author alone and do not necessarily reflect views or policies of the U.S. Department of Agriculture. The author wishes to acknowledge suggestions by David Lucking-Reiley, William Lesser, Stephen Merrill, George Elliot, and Wesley Cohen.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 54
Patents in the Knowledge-Based Economy Patent Examination Procedures and Patent Quality1 John L. King Economic Research Service U.S. Department of Agriculture ABSTRACT This study examines a detailed panel data set of patent examination procedures that affect patent quality. A main conclusion is that the most important of these inputs (examiner hours and examiner actions) have remained largely consistent over time despite an increasing examination workload. Other measures of examination quality (pendency and interference hearings) have declined. Inputs to examination quality are inversely correlated with the rate at which patents are involved in legal complaints, and the expense of increasing examination inputs may be more than offset by the consequent reduction in litigation costs. INTRODUCTION Patents grant the exclusive rights to use, manufacture, and sell new inventions and are widely sought legal instruments. The numbers of both patent applications and patent awards have more than doubled over the past two decades. Explanations for this increase might include more innovative activity, greater emphasis on intellectual property rights among innovating firms, reduced requirements for patentability, or a variety of other possibilities. Regardless of the cause, the increase in patent activity has created a greater examination workload and placed a greater burden on patent-granting institutions. This chapter examines the 1   Portions of this research first appeared as “An Empirical Investigation of the Economics of Patent Institutions,” Ph.D. dissertation, Vanderbilt University, 2000. The views expressed are those of the author alone and do not necessarily reflect views or policies of the U.S. Department of Agriculture. The author wishes to acknowledge suggestions by David Lucking-Reiley, William Lesser, Stephen Merrill, George Elliot, and Wesley Cohen.

OCR for page 54
Patents in the Knowledge-Based Economy increase in empirical detail and analyzes its impact on the quality of patent examination. Drafting a patent application is a rigorous exercise in technical language that must accommodate the technology underlying the invention, its commercial significance, and relevant statutory and case law. Patent examiners, who read the applications and ultimately decide whether to allow a patent award, must have at least a basic understanding of these and other factors. A skilled staff of patent examiners with adequate training and resources is essential to maintain the validity of patents that issue according to their decisions. The recent rise in patenting activity raises the question of whether patent examiners have adequate resources to fulfill their increased examination responsibilities in a thorough and timely way. The main contribution of this study is to present data that quantitatively assess the effect of increasing application workloads on the recent examination performance of the U.S. Patent and Trademark Office (USPTO). The study presented in this chapter provides descriptive data about various features of patent examination, largely drawn from detailed Time and Activity Reports of examiners maintained by the USPTO. This data set and information about patent institutions allow inferences about changes in the quality of patent examination. Allison and Lemley (1998) and Jaffe (2000) discuss trends in increasing patenting activity over the past two decades, and Merges (1999) discusses the possible ramifications on patent litigation and incentives to innovate when patent examination standards are unevenly applied. This study extends the empirical analysis of patenting trends in a way that directly addresses issues raised in connection with standards of patent examination. In addition, this chapter presents results of multivariate regression analysis indicating the correlation between measures of examination quality and areas of policy concern, such as patent litigation and incentives for inventors. Although patent examination is only one element in the complex landscape of intellectual property rights, it is a subject with important implications and one that is amenable to empirical analysis. Careful patent examination reduces the need for courts to review patent agency decisions in the eventuality of a patent dispute. Hypothetically, a “perfect” patent agency would never issue a patent that was later found invalid in a court of law. In addition, the scope of issued patents would be extremely clear, providing an easy test—and strong deterrent—for infringement should a dispute arise. Approved claims would be broad enough to reward inventors of significant discoveries but sufficiently narrow to allow patents on competing inventions or further improvements. Conversely, factors that constrain the quality of patent examination cause a divergence of patent agency decisions from the determination a court would make if presented with the same facts. This divergence obscures the true strength of a patent in court. A reduction in the quality of patent

OCR for page 54
Patents in the Knowledge-Based Economy examination therefore increases the uncertainty of enforcing the intellectual property rights that are inherent in a patent award. However, the efficiency of the patent examination process requires that examiners make their determinations without the expense and delay that often accompany litigation. Judging patent applications against the relevant technical, legal, and commercial information requires scarce human capital, which the USPTO must acquire and develop. In addition, examiners play their most active role in the examination process in the course of offering an initial rejection, through which they may require amendments to an application such as more citations to prior art, the narrowing or elimination of a specific claim, or one of many corrections that improve the patent application. Doing this in a thoughtful and competent manner requires time, resources, and inputs sufficient to the task. A concern about increasing application workloads is that they might impose constraints on the resources available to examiners, with the possible consequence of jeopardizing examination quality and creating uncertainty in intellectual property rights enforcement. This is an area of some importance, because uncertain intellectual property rights impose several kinds of costs on the economy. Subsequent legal effort to determine the validity or proper scope of a patent is necessary when examination quality is lower. Legal costs are especially high when patent disputes result in litigation. To the extent that the enforcement costs of patent protection undermine incentives to innovate, low examination quality reduces the amount of innovation in society. Also, when the patent examination process fails to reject patent applications with serious flaws, patent monopolies for inventions with little benefit to society impose additional welfare costs. These costs are more likely to accrue as a greater number of patents are issued each year, underscoring the importance of examination quality in the patent system. DATA AND METHODOLOGY Patent examination quality refers to the ability of patent examiners to make a correct judgment about whether to grant a patent application, meaning that their decisions about validity and scope of protection are consistent with the ruling a court would make after a comprehensive review of the application. Patent examination therefore requires considerable knowledge and skill in the technological area but also knowledge of evolving court rulings. In the aggregate, it is difficult to assess the quality of patent examination directly. Examination quality is a multifaceted concept covering validity, scope, timeliness, and other attributes. The examination effort necessary to make a determination is likely to vary from application to application, because each patent presumably possesses some unique and novel features. Legal rulings on patents occur too infrequently to provide a comprehensive view of whether courts uphold typical examiner decisions. Moreover, the sample of patents that proceed to court

OCR for page 54
Patents in the Knowledge-Based Economy rulings is likely to differ from the population of patents as a whole (Priest and Klein, 1984). However, inputs to the patent examination process can be observed and quantified. The skill and experience of patent examiners, the time allotted to patent examination, and other factors that contribute to examination can be measured. The principal data used to measure the inputs and outputs of the patent examination process in this study come from Time and Activity Reports obtained from internal USPTO records.2 The data set contains detailed information about the types of duties performed by examiners, the intensity of examination effort, and some information about examiners. The Time and Activity Reports summarize examination activities covering the period from 1985 to 1997 (Table 1). Some tables and calculations in subsequent sections include aggregate statistics available through 1998, but results that required data at the examination group level extended only through 1997. Combined with other information about patents, the data facilitate three types of analysis. First, this study provides descriptive data on examination inputs and outputs. Comparison of examination effort to level of examination output measures the intensity of examination, providing some insight into examination quality. Multilinear regressions show which examiner activities are most closely associated with patent awards, establishing a relationship between inputs and outputs. Second, this study analyzes how examination intensity has changed over time, focusing especially on the past two decades of heightened patenting activity. This part of the analysis explores the effects of rising workloads on examination quality. Finally, analysis of a separate data set on patent litigation allows exploration of important consequences outside the patent process itself. In particular, this study relates measurements of patent examination quality with patterns in patent litigation. The combination of these three types of analysis provides an empirical view of examination procedures and allows exploration of their effects on outcomes of policy interest such as patent litigation. Some of the key variables included in the Time and Activity Reports include: • Examination hours. Examination hours are a primary measure of examiner input. Dividing the total number of hours by 2,000 provides a rough estimate of full-time equivalent (FTE) examiners.3 The data set distinguishes between regular hours and overtime hours spent examining patents in the various examination groups. • GS-12-equivalent examination hours. An enhanced measure of examiner effort arises from normalizing examination hours by examiner experience and training. The data set includes examination hours adjusted to a GS-12 pay 2   King (2000) discusses FOIA Request 99-118 used to obtain the data. 3   Assumes a 40-hour work week for 50 weeks per year.

OCR for page 54
Patents in the Knowledge-Based Economy TABLE 1 Summary of PTO Time and Activity Reports, 1985-1997   Examination Group Mean Annual Patent Disposals Mean GS-12 Examination Hours per Disposal Mean Actions per Disposal Mean Rejections per Disposal 1100. General metallurgical, inorganic petroleum, and electrical chemistry and engineering 16,319 18.52 2.48 0.41 1200. Organic chemistry 15,144 16.20 2.69 0.40 1300. Specialized chemical industries and chemical engineering 15,734 18.66 2.54 0.41 1500. High-polymer chemistry, plastics, coating, photography stock materials and compositions 17,829 17.79 2.45 0.43 1800. Biotechnology 15,990 22.41 3.04 0.55 2100. Industrial electronics, physics, and related elements 13,676 19.44 2.27 0.26 2300. Information processing, storage, and retrieval. 11,888 27.52 2.51 0.40 2400. Packages, cleaning, textiles, and geometrical instruments 11,908 17.87 2.35 0.35 2500. Electronic and optical systems and devices 17,191 20.02 2.35 0.32 2600. Communications, measuring, writing, and lamp/discharge 16,445 23.36 2.48 0.35 3100. Handling and transportation media 12,681 16.98 2.26 0.33 3200. Material shaping, article manufacturing, and tools 13,677 15.78 2.23 0.31 3300. Mechanical technologies and husbandry personal treatment information 16,095 16.86 2.44 0.38 3400. Solar, heat, power, and fluid engineering devices 12,259 16.27 2.17 0.24 3500. General construction, petroleum and mining engineering 15,235 15.67 2.29 0.30 NOTE: Examination Groups 2200 (Special Administrative Unit) and 2900 (Design) were not included in this study. grade, meaning that examiner hours of more experienced examiners (at higher pay grades) count proportionally more with respect to differences in pay. This measure takes into account differing levels of experience among examiners and is therefore a better measure of effective examiner input. • Patent disposals. Patent disposals are the sum of patent approvals and rejections issued by examination groups, providing a good absolute measure of total examiner output.

OCR for page 54
Patents in the Knowledge-Based Economy • Examiner actions. An examiner action occurs when an examiner reviews a patent application and responds to the applicant, which usually happens several times before a patent disposal (i.e., acceptance or rejection of a patent application). Each action provides examiners with an opportunity to require improvements before a patent award or give grounds for a rejection. The number of actions for each patent disposal is another measure of examination intensity. • Patent pendency. The timeliness of patent examination is an important factor because of brisk competition in technology markets; also, time spent in examination counts directly against the 20 years of protection that a patent award allows. The ability of examiners to render accurate, thorough examinations in a brief period enhances the benefits of the patent system to inventions. The data source used to analyze the effects of patent examination procedures on patent litigation is derived from a U.S. patent law that requires that courts notify the USPTO when a patent becomes involved in a legal dispute.4 Once the USPTO receives notice of either a complaint or the formal resolution of a case, it is indexed and published commercially by Derwent Publishing in the LitAlert database. This notification requirement creates an opportunity to assemble a data source on which patents are involved in litigation, which cases are settled out of court, and how frequently patents are found to be invalid in court decisions. The litigation data include patents issued between 1989 and 1991 that were involved in legal disputes before 2000. Although some of these patents might have been involved in subsequent litigation, the data show that 95 percent of the disputes arose within 3 years of patent issue, indicating that truncation is not a problem for this sample. A more thorough description of the litigation data used here is available in King (2000); it is constructed from the same primary source used by Lanjouw and Schankerman (2001). The comparison of outputs and workload across the entire population of patents raises a methodological concern about how to treat dissimilar patents. For instance, patents in complex technological areas or areas of relatively rapid innovation such as biotechnology and semiconductors might require additional examination effort. If a complex technology requires more examiner input to provide the same examination quality, regression analysis must account for this difference. Likewise, examiners in more mature technological areas must consider more prior art. Fortunately, this data set allows the use of panel estimation techniques. These USPTO Time and Activity Reports tracked examination effort in 17 different 4   35 USC 290: “The clerks of the courts of the United States, within one month after the filing of a [legal action involving a patent] shall give notice thereof in writing to the Commissioner…. Within one month after the decision is rendered or a judgment issued the clerk of the court shall give notice thereof to the Commissioner.”

OCR for page 54
Patents in the Knowledge-Based Economy technology areas, known as examination groups, over time. To the extent that examination, downstream demand for patents, and other factors affecting patent activity are aligned by technologies, panel estimation techniques allow the isolation of the statistical relationship between examiner inputs and patent quality, holding these other factors constant. Also, examination groups provide a particularly useful unit of analysis for policy makers, suggesting specific areas of patent examination to which resources can be shifted at the margin to improve overall examination performance. Estimates from the panel data are presented with both random-effects and fixed-effects model specifications. The fixed-effects specification might be preferable if relevant aspects of patent examination are strongly correlated with the broad range of technologies represented in the 17 examination groups. If this is not the case, or because the fixed-effects model is more costly in degrees of freedom, then the random-effects specification might be preferable. Because the purpose of this chapter is not a methodological comparison of these estimation techniques, results from both specifications are reported. Unfortunately, a reorganization of the USPTO after 1997 limits the time series of the panel data. The 17 examination groups were reorganized into 6 “technology centers” after 1998, and the USPTO did not release the more finely disaggregated numbers in subsequent years. Notwithstanding this temporal limitation, the data set allows detailed analysis of patent examination procedures at the examination group level through a substantial part of the trend of increased patent activity over the past two decades. TRENDS UNDER INCREASING PATENT EXAMINER WORKLOADS This section presents findings from analysis of the data as they pertain to examiner workload and various measures of the thoroughness and timeliness of patent examination. The general finding is that over the period in which examiner workload has increased, several important measures of patent quality have remained consistent, while some others have suffered. The data do not appear to support a general decline in patent examination quality, although a trend for increasing workload might create delays in the examination process if it continues. Between 1985 and 1998, the USPTO issued approximately 1.4 million patents and 775,000 final rejections, summing to almost 2.2 million patent disposals. On average, a patent disposal received 17.1 hours of examiner time. Adjusting for examiner experience by using USPTO salary calculations, each patent disposal required 19.1 hours of examiner effort paid at the GS-12 level. Figure 1 shows how the average examination time varied across examination groups. The data reflect more variation across groups than intertemporal variation within groups. For instance, mean examiner hours per patent disposal range between 15.4 and 27.4 across examination groups, a difference of 13.0, whereas most individual examination groups varied within a narrower range of about 3

OCR for page 54
Patents in the Knowledge-Based Economy FIGURE 1 Experience-adjusted examination hours per patent disposal. SOURCE: USPTO Time and Activity Reports, 1985-1997. NOTE: Outliers for Examination Group 2400 in 1995-1997 (28, 24.3, 29.7) omitted. hours per patent disposal. Although Figure 1 does not show how mean examiner hours per disposal vary over time, the values tend to be evenly distributed around their means and exhibit neither an increasing nor a decreasing trend. Examination Group 2400—Packages, Cleaning, Textiles, and Geometrical Instruments— showed the most dramatic intertemporal variation, but that was mostly confined to an unexplained jump in the final 3 years of the data (which are omitted from the calculations for Figure 1). An interpretation of Figure 1 that is consistent with the use of fixed-effects modeling is that the time required to examine a patent varies according to examination group. The number of applications and issued patents from each examination group did vary over time, but the amount of examiner effort required to dispose of these applications generally ranged within a fairly narrow band. Panel estimation controls for different examiner input requirements among different technical areas. With this sense of the level of examiner effort necessary to process patent applications, Figure 2 illustrates how the workload of examiners varied over time. Normalizing examiner activity levels to 100 in 1985, Figure 2 shows the increase in examiner workload from 1985 to 1998. In the first half of the period patent awards slightly outstripped the number of hours devoted to examination, but the

OCR for page 54
Patents in the Knowledge-Based Economy FIGURE 2 Change in utility patent workload, examination hours in the United States. SOURCE: USPTO Time and Activity Reports, Annual Reports of the Commissioner of the USPTO, 1985-1998. two variables generally kept pace over the entire span. Although it is not represented in Figure 2, further analysis of the data shows that the same was true within the examination groups as well. Examination groups 2300 (information processing, storage, and retrieval) and 2600 (communications, measuring, writing, and lamp/discharge) experienced temporary decreases in examination hours per patent award, but both reestablished and eventually increased examination intensity. Figure 2 also shows the use of overtime during this period. The number of overtime hours tended to increase slightly from periodic significant decreases. To the extent that overtime hours are imperfect substitutes for examination by fulltime examiners, it would appear that the persistent, rapid increase in patent applications justified employment of more full-time examiners. Indeed, the USPTO hired over 700 additional patent examiners in both 1998 and 1999,5 although 5   USPTO Annual Report, 1999.

OCR for page 54
Patents in the Knowledge-Based Economy FIGURE 3 Average pendency for U.S. patents, 1988-2000. SOURCE: Annual Reports of the Commissioner of the USPTO. some of the additional labor merely replaced attrition. The number of overtime hours might also reflect the difficulty of hiring examiners: The focus of the economic expansion during the late 1990s on high-technology areas may have decreased the availability of suitable examiners. The number of patent applications, another measure of examiner workload, rose at a faster rate than the number of examination hours (Figure 2). Patent applications outnumbered patent awards by 60 percent in 1980 and grew at a faster average rate (7.3 percent compared to 6.4 percent through 1998). The result was a widening gap between the number of patent applications filed with the USPTO and the time available for examiners to review them. The net effect was an increasing backlog in patent applications. Figure 3 shows the average pendency of patents, i.e., the length of time between filing dates and patent disposals (allowances or rejections). Over a period in which examiner workload was increasing, average pendency increased by 25 percent from 20.0 months in 1988 to 25.0 months in 1999. To summarize Figures 1 through 3, examination hours per patent disposal varied significantly across groups but were more or less constant within groups over time. Patent awards issued by examination groups varied quite closely with the number of examiner hours each group employed. At the same time, patent applications increased at a faster rate. Pendency, the length of time between application and patent award, increased over time. A possible interpretation of these facts suggests a patent system applying consistent examination intensity to the

OCR for page 54
Patents in the Knowledge-Based Economy patents it issues but constrained from examining its increasing workload by the examiner resources available to it. In 1996 the GAO conducted two studies in patent pendency.6 Congress requested the studies out of concern for the impact of pendency on effective patent terms, after passage of a law that changed patent terms from 17 years after issue to 20 years after filing.7 Although the effective patent term for the average patent actually increased under the new law, it is clear from Figure 3 that pendency increased substantially throughout the 1990s. The close relationship between patent awards and the availability of examiner hours employed suggests that greater examiner employment might effectively reduce patent pendency. Levin et al. (1987) and Cohen et al. (2000) established that lead time is critically important in protecting intellectual property, perhaps more so than patents. So, even apart from the erosion of patent term resulting from longer patent examination delays, the usefulness of patent protection and therefore its incentive effects for innovation suffer. Longer pendency represents one aspect of examination quality that has been compromised in the face of increasing workloads. Although longer pendency is one effect that can be associated with increasing examination workload, it is important to consider other quantitative relationships in examiner activity. Specifically, data in this study show basic relationships between inputs and outputs of the patent examination process. In the context of regression analysis, the number of hours an examination group devotes to patent examination is strongly correlated with the number of patent awards the group issues. When patent examination output is measured by the natural logarithm of patent disposals (i.e., allowed patents and final rejections) by an examination group, GS-12 examiner hours (Table 2) have the largest and most statistically significant effect on production.8 Because the coefficients on either independent variable increased in magnitude when the other was excluded, examination hours are probably somewhat collinear with the number of examiner actions per patent disposal. However, the sign and significance of both variables were also robust to this variant on model specification. The relationship between examiner actions, another input to the examination process, and patent awards is more ambiguous. On its face, a greater number of 6   GAO/RCED-96-190, GAO/RCED-96-152R. The studies provide pendency detail at the art unit level, but only for a 1-year “snapshot” in 1994, so they could not be integrated fully with the data in this study. 7   Eagerness of inventors to submit applications under the earlier patent term probably explains the 1994 surge and the corresponding drop-off in patent applications the following year, as shown in Figure 1. 8   The log-log specification is used to maintain consistency with results presented later in this chapter. The positive, significant coefficients are robust to linear and semilog specifications. Also, a specification including linear and quadratic terms returned significant coefficients that were respectively positive and negative, respectively, providing the same goodness of fit (R2 = 0.74) as the log specification within the range of the sample.

OCR for page 54
Patents in the Knowledge-Based Economy TABLE 2 Contribution of Inputs to Examination Output (Dependent Variable: Logarithm of Patent Disposals)   Random Effects Fixed Effects GS-12 Hours (log) 0.6670 t = 16.80a 0.6995 t = 17.95a Total Actions (log) 0.2375 t = 6.83a 0.2162 t = 6.37a Constant −1.2842 t = −4.16a −1.4607 t = −4.86a R2 0.75 0.74 aSignificant at 1%. NOTE: 15 examination groups, 1985-1997 (3 missing values). examiner actions per patent disposal suggests greater examiner scrutiny. Indeed, only 13 percent of patent applications were approved on first action.9 Each successive action increases the likelihood of a patent award, with 73 percent of third and subsequent actions resulting in approval. This is consistent with examiners interacting with inventors to improve the application and resulting patent award, a manifestation of examination quality. On the other hand, a greater number of examiner actions might indicate a case in which “the squeaky wheel gets the grease”: The applications receiving greater scrutiny might be the ones closest to being unpatentable. In this interpretation, more actions per patent disposal suggests a greater number of marginal patents and greater examiner effort implies a higher proportion of patents with weak claims. The positive coefficient on total examiner actions, which was robust to several model specifications (i.e., quadratic, semilog, etc.) is consistent with either interpretation. Additional data are needed to clarify the role of examiner actions in improving examination quality. Examination Quality and Patent Litigation Patent protection is asserted through litigation or through negotiation backed by the threat of litigation. In part because of the quality of the patent examination process, courts presume that patents involved in legal complaints are valid. To the extent that the examination process can rigorously clarify intellectual property rights in advance, examination can help parties avoid legal disputes and can promote efficiency in industries where patents are important. The purpose of this section is to bring data on examination practices to bear on this issue, to see what role the examination process might play in patent litigation outcomes.10 9   Practitioner accounts suggest that some patents awarded on first action are actually reworked applications, so that even fewer patents are awarded on their very first attempt. 10   Other authors to address empirical analysis of patent litigation include Allison and Lemley (1998); Jaffe (2000); and Lanjouw and Schankerman (2001).

OCR for page 54
Patents in the Knowledge-Based Economy Clearly, factors other than patent examination also determine decisions to develop and protect intellectual property. The decision to compete and potentially infringe in markets protected by patents and the decision to litigate against infringers who may have patents of their own involve strategic determinations based on many factors. However, the quality of patent examination affects the validity of issued patents and therefore the incentives for patent infringement. For instance, if issued patents were unlikely to be upheld in court, it is likely that more infringement would be observed. Whether this leads to more patent litigation would depend on the specific technological, legal, and economic issues surrounding each case and whether validity is an issue. Meurer (1989) models the strategic incentives facing patentholders and potential infringers when the validity of issued patents is uncertain. The model focuses on the question of whether parties will choose to litigate or settle disputes when infringement occurs. However, a theoretical prediction of the model is that greater uncertainty of patent validity leads to a greater incidence of infringement and litigation. Using the data on patent examination quality to stand in for certainty of intellectual property rights, this study performs an empirical test of this hypothesis. To the extent that the variables identified here are correlated with the quality of patent examination, this approach quantifies the effect of examination intensity on patent litigation. Public records at the USPTO indicate that only 0.22 percent of patents issued between 1989 and 1991 were involved in legal complaints, approximately 200 patents from each year (King, 2000). Patent litigation can be protracted and expensive, which is one reason why so few patents were involved in legal complaints. The cost of litigation also helps explain why many disputes were settled before a final verdict was reached in court: Of complaints involving patents issued between 1989 and 1991 (Table 3), somewhere between 38 and 55 percent of these cases (between 0.084 and 0.121 percent of patents allowed) reached a final verdict.11 Table 4 presents estimates of the effects of various measures of examination quality on the rate at which issued patents were involved in legal complaints. The unit of observation is a patent examination group in a sample year. In addition to GS-12 hours per patent disposal and total number of examiner actions per disposal to measure the quality of examination, the regression includes the mean elapsed time between complaint filing and verdict to control for the cost of litigation, which, ceteris paribus, should reduce the incidence of litigation. Also, the panel nature of the data holds constant effects that vary with technological area. The number of GS-12 examiner hours per patent disposal has the greatest effect on the complaint rate of the variables included in the analysis. This effect is 11   USPTO records did not easily allow a precise determination of litigation outcomes; see King (2000) for details.

OCR for page 54
Patents in the Knowledge-Based Economy TABLE 3 Legal Complaints Involving Patents Issued Between 1989 and 1991 Examination Group Annual Rate of Patents Involved in Complaints per Patent Allowed 1100. General metallurgical, inorganic petroleum and electrical chemistry, and engineering 0.001469 1200. Organic chemistry 0.001303 1300. Specialized chemical industries and chemical engineering 0.001729 1500. High-polymer chemistry, plastics, coating, photography stock materials and compositions 0.001071 1800. Biotechnology 0.001371 2100. Industrial electronics, physics, and related elements 0.001042 2300. Information processing, storage, and retrieval. 0.000535 2400. Packages, cleaning, textiles, and geometrical instruments 0.000050 2500. Electronic and optical systems and devices 0.000000 2600. Communications, measuring, writing, and lamp/discharge 0.001948 3100. Handling and transportation media 0.003137 3200. Material shaping, article manufacturing, and tools 0.003702 3300. Mechanical technologies and husbandry personal treatment information 0.004774 3400. Solar, heat, power, and fluid engineering devices 0.004485 3500. General construction, petroleum and mining engineering 0.004393   SOURCE: Derwent LitAlert, USPTO Annual Reports. TABLE 4 Panel Regression of Complaint Rate on Measures of Examination Quality (Dependent Variable: Logarithm of Complaint Rate)   Random-Effects Model Fixed-Effects Model Random Effects Model Fixed-Effects Model GS-12 Examiner Hours per Disposal (log) −1.1749 −1.0702 −1.325 −1.055   t = −3.48a t = −2.84a t = −2.05b t = −3.16a Actions per Disposal (log) 0.3239 −0.0814 — —   t = 1.20 t = −0.09     Time to Verdict (log) −0.0003 0.0002 −0.0003 −0.0002   t = −0.50 t = 0.36 t = −0.64 t = −0.36 R2 0.40 0.27 0.36 0.30 aSignificant at 1 percent. bSignificant at 5 percent. NOTE: U.S. patents issued by 15 examination groups, observed 1989-1991; 8 missing observations due to absence of verdicts.

OCR for page 54
Patents in the Knowledge-Based Economy statistically significant and robust to model specification.12 This result implies that examination groups that spent more time examining patent applications issued patents that were less likely to be involved in litigation. Conversely, patent litigation was a greater risk in examination groups in which patent examination hours were low. Actions per patent disposal did not have a significant effect on the complaint rate. Collinearity with examiner hours is one possible explanation. Model specifications excluding examiner hours made the coefficient on examiner actions significant and positive in a few model specifications, but not robustly. This might indicate that applications receiving more examiner actions were more contentious in some way. In light of the earlier discussion on examiner actions after Table 2, this result lends some support to the hypothesis that marginal patents receive more examiner actions. An alternative interpretation is that examiners have some foresight about which patents are likely to become embroiled in litigation and interact more extensively with those applicants. Because the data do not allow more detailed analysis of this question, Table 4 also presents regression results excluding this variable altogether. The elapsed time between complaint and verdict, intended as a proxy for litigation costs, had no effect on the dependent variable. The hypothesized negative sign appeared in some model specifications but was extremely weak. Exclusion of this variable did not affect the main result on the effect of examiner hours on the complaint rate. In practical terms, the significance of the negative coefficient on examiner hours is that a 1 percent increase in examiner hours per patent disposal is associated with a decrease in patent litigation ranging between 1.05 and 1.33 percent. Using the range of elasticity estimates from the regression results and evaluating at the means, this suggests that an additional hour of patent examination would be associated with a decrease in litigation rates from 2.21 to approximately 2.07 complaints per thousand patent awards, i.e., perhaps as many as 24 to 26 litigation complaints annually.13 A reduction in the amount of litigation on this order of magnitude would have significant economic impact. Depending on the complexity of a patent, the direct cost of the patent examination process to the applicant is probably on the order of $20,000 per patent, including application fees, attorney expenses, etc. In contrast, an American Intellectual Property Lawyers Association (AIPLA) study (1996) estimates the median cost of patent litigation at $600,000 per side through the discovery phase and $1,200,000 per side if litigation proceeds all the way to a 12   In addition to linear and quadratic specifications, the sign and significance of the examiner hours coefficient were robust to exclusion to one or both of the other explanatory variables. 13   At 186,000 patent awards per year, and a complaint rate of 0.221 percent, an increase in GS-12 examiner hours from 17.6 to 18.6 translates to 24 fewer cases when elasticity is −1.055, and 26 fewer cases when elasticity is −1.33.

OCR for page 54
Patents in the Knowledge-Based Economy verdict. These figures represent only out-of-pocket litigation costs, ignoring the huge opportunity cost of time spent preparing for litigation by managers and R&D personnel. The data suggest that an increase in examination quality would reduce the number of patents that courts must review, decreasing the risk of litigation. This reduction in litigation expense benefits the parties involved but also might generate external benefits in the form of greater transparency and less uncertainty regarding intellectual property rights, which contribute to incentives for innovation. Another benefit of decreased litigation from higher examination quality is the reduction in caseloads and associated public expenses for the court system. Increasing the number of examiner hours by an hour for every patent disposal carries substantial costs, however. A policy of increasing examiner effort across all examination groups by 1 hour per patent disposal would mean an increase in examination costs of roughly 5.5 percent. In 1997, the USPTO employed 4,099,241 GS-12-equivalent examiner hours. By dividing the total number of hours by 2,000 examination hours per full-time equivalent examiner and multiplying the result by an annual salary of $100,000,14 the total cost of patent examination operations to the USPTO in 1997 is estimated at $205 million. Therefore, an increase in patent examination effort by 5.5 percent would cost roughly $11.3 million. Although these costs are significant, they point to an interesting result. Assuming that the statistical relationship between increased examination hours and reduced litigation holds, the 1-hour increase in examination would eliminate approximately 25 legal complaints. Reducing litigation to this extent would release significant resources from litigation of patents. This reduction in patent litigation from increased examiner scrutiny could come about either as a result of reducing the number of erroneously granted patents or from improving the ability of granted patents to deter infringement. Estimating the reduction in patent litigation costs associated with an increase in examiner effort requires certain assumptions. Moore (2001) states that only 5 percent of cases proceed to a litigated verdict, which the 1996 AIPLA estimate of median litigation costs cites at $1.2 million. Another 49 percent of cases were dropped before the start of discovery, which can be assumed to cost much less, perhaps $10,000. The remaining 46 percent of patent cases incur some or all of the costs of discovery. With the 1996 AIPLA estimates of median litigation costs through discovery of $600,000, an expected cost of litigation per case can be calculated as: (49% × $10,000) + (46% × $600,000) + (5% × $1,200,000) = $340,900 14   The U.S. Office of Personnel Management states the full-time salary of a GS-12, Step 5 in the Washington, D.C., area in 2000 as $60,242. Rounding this figure up to $100,000 reflects employee benefits, work space, and additional costs of employment.

OCR for page 54
Patents in the Knowledge-Based Economy By multiplying this figure by 25 cases and two sides per case, an estimate of the reduction in litigation costs achieved by increasing patent examination intensity is equal to $17,045,000. This amount significantly exceeds the $11.3 million cost of increased examination effort, which is based on a $100,000 annual salary for examiners that is probably overstated to begin with. Moreover, the cost of litigation is probably significantly underestimated: Practitioner accounts suggest that the 1996 AIPLA median estimate is low and that skewness in the distribution of legal costs causes the mean litigation costs to be higher than the median. Also, this estimate does not include the public costs of litigation (in the form of courts, judges, etc.) or the opportunity cost of time diverted to litigation by managers, R&D personnel, and other employees of companies involved in litigation. Table 5 illustrates in greater detail the potential benefits and costs of reducing litigation through greater examination quality. Using the estimated elasticity TABLE 5 Ratio of Predicted Benefits and Costs from Increased Examiner Hours Examination Group Estimated Reduction in Litigation Expenses Divided by Cost of Additional Examination Hours 1100. General metallurgical, inorganic petroleum, and electrical chemistry and engineering 1.27 1200. Organic chemistry 1.28 1300. Specialized chemical industries and chemical engineering 1.48 1500. High-polymer chemistry, plastics, coating, photography stock materials and compositions 0.96 1800. Biotechnology 0.98 2100. Industrial electronics, physics, and related elements 0.86 2300. Information processing, storage, and retrieval 0.31 2400. Packages, cleaning, textiles, and geometrical instruments 0.04 2500. Electronic and optical systems and devices 0.00 2600. Communications, measuring, writing, and lamp/discharge 1.33 3100. Handling and transportation media 2.95 3200. Material shaping, article manufacturing, and tools 3.74 3300. Mechanical technologies and husbandry personal treatment information 4.52 3400. Solar, heat, power, and fluid engineering devices 4.40 3500. General construction, petroleum and mining engineering 4.47 All   1.79 NOTE: Assumptions: $50 examination cost per hour, $681,800 cost per litigation, and 1.17 percent reduction in complaint rate per 1 percent increase in examiner hours. Examination costs and litigation rates applied to mean patents allowed per year.

OCR for page 54
Patents in the Knowledge-Based Economy of reduced litigation incidence through increased examination hours and controlling for technology and commercial differences with examination group fixed effects (Table 4), Table 5 presents an estimate of potential returns to a policy of dedicating greater resources toward patent examination. The table expresses the predicted benefits of decreasing litigation through greater examination intensity as a ratio of the estimated cost of increased examiner hours. Because examination groups vary both in the number of examiners they employ and the incidence of litigated patents they issue, not all examination groups have the same expected benefits of increased examination. The examination intensity of groups that employ relatively few examiner hours per patent disposal can be increased relatively cheaply; likewise, the potential reduction in patent litigation incidence varies with the frequency of litigation (from Table 3) and the number of patents awarded (from Table 1). To fully capitalize on opportunities for reducing patent litigation with greater examination intensity, additional examiner hours could be targeted to the examination groups in which the ratio of benefits to costs is especially high. Groups in which the potential benefits exceed the additional costs of examination are better choices for targeted improvements in examination effort, especially considering that reductions in litigation costs are probably underestimated. Of course, such a policy would have distributional consequences and might also have effects on patents not involved in litigation.15 The estimates provided here are extremely rough calculations, but they suggest examination groups with the greatest potential benefits from an increase in examination intensity. DISCUSSION To synthesize some results of the previous sections, this study examines inputs to patent examination in the face of an increasing examination workload. Patent examination contributes to the clarity and strength of intellectual property rights and therefore plays an important role in the patent system as a whole. Although the quality of issued patents is difficult to observe in the aggregate, it is possible to quantify inputs necessary for a complete review of patent applications. A main finding of this study is that inputs into the examination process have remained roughly consistent with the number of patent awards over the past two decades. The most important of these inputs, examiner hours and examiner actions, have kept pace with increases in patent awards and rejections. Although it is possible that the inputs necessary to conduct a thorough examination have increased, examiners appear to devote the same amount of time and effort to each 15   For instance, by changing the strength and clarity of intellectual property rights as a deterrent to infringement.

OCR for page 54
Patents in the Knowledge-Based Economy granted patent despite their increasing application workload. In this sense, empirical analysis suggests that the quality of examination has not declined with the increase in patent activity over the past two decades. Although examination has kept pace with the number of patent awards, it has fallen somewhat behind with respect to the number of patent applications. The duration of patent pendency has increased, so that applications take longer to go through the examination process. These facts might indicate that standards of patent examination are coming under some pressure from increasing workloads. Because the main determinant of patent disposals appears to be examiner time and effort, an obvious way to address problems with the examination process is to ensure that the USPTO has sufficient resources. Currently, the USPTO funds examination activities through user fees; however, Congressional appropriations bills from 1996 to 2000 created total budget rescissions approaching $150 million from an annual budget ranging from $621 million to $911 million. These rescissions could have a significant effect if made available to fund examiners. Clearly delineated intellectual property rights could increase incentives to innovate and could have other benefits as well. For instance, inputs to patent examination quality are statistically correlated with lower patent litigation activity, meaning that more examination could lower transaction costs associated with enforcement of intellectual property rights. Implementation of a policy along these lines could be further refined by concentrating additional examiner resources in technologies or markets in which the incidence of litigation is especially high. The estimated benefits in terms of reduced litigation costs are greater than the estimated increase in examination costs on the whole, but the benefit is greater in certain technological areas. REFERENCES AIPLA. (1996). Report of Economic Survey.Arlington, VA: American Intellectual Property Lawyers Association. Allison, J. R., and M. A. Lemley. (1998). “Empirical Evidence on the Validity of Litigated Patents.” American Intellectual Property Lawyers Association Quarterly Journal 26(3): 185-275. Cohen, W. M., R. R. Nelson, J.P. Walsh. (2000). “Protecting Their Intellectual Assets: Appropriability Conditions and Why U.S. Manufacturing Firms Patent (or Not).” NBER Working Paper Series No. 7552. Jaffe, A. B. (2000). “The U.S. Patent System in Transition: Policy Innovation and the Innovation Process.” Research Policy 29(4-5) (April): 531-557. King, J. L. (2000). “An Empirical Investigation of the Economics of Patent Institutions.” Dissertation, Vanderbilt University, Nashville, Tenn. Lanjouw, J. O., and M. Schankerman. (2001). “Characteristics of Patent Litigation: A Window on Competition.” RAND Journal of Economics 32(1) (Spring): 129-151. Levin, R., A. Klevorick, R. R. Nelson, and S. G. Winter. (1987). “Appropriating the Returns from Industrial R&D.” Brookings Papers on Economic Activity, 3: 783-820. Merges, R. (1999). “As Many as Six Impossible Patents Before Breakfast: Property Rights for Business Concepts and Patents System Reform.” Berkeley Technology Law Journal 14(2) (Spring): 577-615.

OCR for page 54
Patents in the Knowledge-Based Economy Meurer, M. J. (1989). “The Settlement of Patent Litigation.” RAND Journal of Economics 20(1) (Spring): 77-91. Moore, K. A. (2001). “Forum Shopping in Patent Cases: Does Geographic Choice Affect Innovation?” North Carolina Law Review 79(4) (May): 889-938. Priest, G., and B. Klein. (1984). “The Selection of Disputes for Litigation.” Journal of Legal Studies 13(1) (January): 1-55.