David Austin and Molly Macauley
Resources for the Future
This paper illustrates how the ATP has extended the state-of-the-art of evaluation for programs which support the development of new technologies. An important caveat to keep in mind is that the paper does not claim the estimated benefits are realized, or will be. It does provide a means to estimate the potential returns to R&D investments, in this case, if the new ATP-funded technologies are adopted.
This study generates these estimates by drawing on previous economic modeling to develop and apply a novel new method for estimating consumer benefits from advanced technology under development or under consideration.
The resulting cost index model is tested by estimating expected consumer welfare gains from two ATP-funded innovations in digital data storage (DDS)—both expected to offer dramatic price/performance improvements compared to existing tape drives if they are successfully commercialized.
The analysis yields expected benefits to consumers from the optical tape technology in excess of $1 billion, and from the linear scanning technology, $2 billion, both taken over a five-year period, and conservatively estimated.
Based on experimental use in assessing ATP projects, the method appears suitable for use by other federal and state government programs to develop performance measures for programs which support the development of advanced technology.
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 239
Page 239 Estimating Future Benefits from ATP Funding of Digital Data Storage David Austin and Molly Macauley Resources for the Future KEY POINTS This paper illustrates how the ATP has extended the state-of-the-art of evaluation for programs which support the development of new technologies. An important caveat to keep in mind is that the paper does not claim the estimated benefits are realized, or will be. It does provide a means to estimate the potential returns to R&D investments, in this case, if the new ATP-funded technologies are adopted. This study generates these estimates by drawing on previous economic modeling to develop and apply a novel new method for estimating consumer benefits from advanced technology under development or under consideration. The resulting cost index model is tested by estimating expected consumer welfare gains from two ATP-funded innovations in digital data storage (DDS)—both expected to offer dramatic price/performance improvements compared to existing tape drives if they are successfully commercialized. The analysis yields expected benefits to consumers from the optical tape technology in excess of $1 billion, and from the linear scanning technology, $2 billion, both taken over a five-year period, and conservatively estimated. Based on experimental use in assessing ATP projects, the method appears suitable for use by other federal and state government programs to develop performance measures for programs which support the development of advanced technology.
OCR for page 239
Page 240 OVERVIEW This papers develops and applies a new method for estimating consumer benefits from advanced technology. The new method takes into account changes in the quality of service provided by the new technology. We apply the method to evaluate two ATP-funded innovations in data storage technology. One innovation, undertaken by LOTS Technology, Inc., is an optical tape read/write technology representing a dramatic increase in data storage capacity. The second innovation, carried out by Imation Corporation, is to develop underlying technology for linear scanning of magnetic tape that, at a fraction of the cost, can match or exceed the performance and capacity of a helical-scan system, a competing technology led by offshore competitors. Both technologies promise dramatic price/performance improvements compared to existing tape drives. Applying the new evaluation method, we estimate the expected benefits to consumers from the optical tape technology to exceed $1 billion, and from the linear scanning technology, $2 billion, both taken over a five-year period, and conservatively estimated. This study focuses on consumer benefits from early commercialization of the technologies, and ignores benefits accruing to the innovator, or to other manufacturers via knowledge spillovers, as well as benefits from second-generation products. DEVELOPING A MODEL FOR ESTIMATING BENEFITS TO CONSUMERS In any R&D program, a key challenge for public and private managers alike is to assess what the outcomes and longer run impacts will be. Government programs, such as the ATP, are particularly concerned about how consumers will benefit from the innovations being developed. Traditionally, forecasts of how consumers will benefit from an innovation have been difficult, particularly when the innovation is a high-tech link in a chain that ends with a benefit to the consumer. The quality of service is not readily observable and is difficult to measure quantitatively. Work by Stanford's Timothy Bresnahan 1 made it more feasible to estimate consumer benefits from quality of service improvements. He developed a cost-of-living index that, under certain general assumptions, makes it possible to compare observed price and performance for an innovated product against hypothetical, best-available price and performance had the technical advance not occurred. Bresnahan's method was aimed at retrospective evaluation; that is, for 1 Timothy Bresnahan, “Measuring the Spillovers from Technical Advance: Mainframe Computers in Financial Services,” American Economic Review, 76(4):742-755, 1986.
OCR for page 239
Page 241estimating consumer benefits from already existing innovations rather than for forecasting consumer benefits from innovations still under development. A Prospective Approach To meet the challenges facing R&D managers, we extend Bresnahan's method to make it suitable for prospective assessments of consumer benefits from proposed R&D projects and those that have not yet resulted in products on the market. Our approach allows for the gradual diffusion of the new technology, and we express the model's parameters as probability density functions to reflect uncertainties over future or estimated parameter values. We also extend Bresnahan's method to have it reflect consumer preferences for specific product characteristics—speed, for instance—and to take into account the fact that those preferences may affect the product's success in a competitive marketplace. Differentiated product characteristics may provide benefits not fully reflected in product prices. A Tool to Estimate Potential Benefits The resulting method can be used for estimating the potential benefits of a technology under development or under consideration. Thus it has potential as a tool to guide the allocation of both private-sector and government R&D investments to projects with high potential for consumer benefits, as well as to evaluate potential consumer benefits per se. For federal government programs like the ATP, this methodology is suitable to meet the requirements of the Government Performance and Results Act for performance measures. Illustration of an Expected Gain in Consumer Benefits Figure 1 illustrates the expected gain in consumer benefits from a technological innovation, such as an improvement in the ability to store large amounts of data. The left-hand panel of the figure shows the pre-innovation baseline, where only a defender technology is available, whose supply is represented by. The downward sloping line, D, represents demand for data storage. The right-hand panel of Figure 1 shows the ATP-sponsored innovation—a combination of cost reductions and quality improvements—that occurs in a subsequent time period. The innovation is represented as an outward shift in the supply curve to . In the meantime, the defender technology may also have improved and shifted the baseline supply curve to . The shaded area represents the consumer welfare gain (“consumer surplus”) at a point in time, due to the innovation. It is measured with respect to the hypothetical, future curve rather than the observed . As long as lies to the right of , the innovation offers an expected improvement over the defender technology. Cost Index Model Measuring the gain is straightforward if the demand curve can be estimated using econometric techniques. But this is difficult to do in
OCR for page 239
Page 242 ~ enlarge ~ FIGURE 1 Derived demand for new technologies: Illustration of net surplus change service sectors, where real output is not readily observed, yet where much of the demand for high technology is located. In this case, the Tornqvist cost-index approach pioneered by Bresnahan is attractive, because it does not require estimating a demand curve. To paraphrase Brasnahan, the method substitutes economic theory for (unobservable) data. With reference to Figure 1, the cost index will be greater than unity, meaning costs would be higher under the baseline scenario and consumers will be better off (gross of R&D costs) if the innovation occurs. The index is an estimate of the change in the cost-of-living (in the sense of consuming digital data storage-using services) under the innovation scenario, relative to the baseline. To construct the index, we adjust nominal unit prices of off-the-shelf devices to reflect consumer preference for faster data transfer rates, larger capacities, and lower file access times. 2 We assume that these “shadow values” decline over time, reflecting consumers' declining marginal utilities. For example, an extra gigabyte of storage capacity is more valuable to a consumer who has only ten gigabytes storage than to a consumer with 100 gigabytes. The value of a given increase will decline over time as performance further improves. 2 A discussion of quality-adjustment methods employed by the Bureau of Labor Statistics in the construction of the consumer price index (CPI) can be found in B. R. Moulton and K. E. Moses, “Addressing the Quality Change Issue in the Consumer Price Index,” Brookings Papers on Economic Activity I, 1997, pp. 305-366. Unlike the CPI, which compares prices over time, the index used here compares prices in a single period: expected, future prices given the innovation versus hypothetical, future prices assuming no innovation.
OCR for page 239
Page 243In our applications of the technique, we almost always adjust the prices of the defender technologies. Their usually lower capacities and transfer rates, and longer file access time, impose real user costs relative to the innovations. The price adjustments reflect consumers' willingness to pay to achieve the relatively superior performance of the innovations. More specifically, the index is constructed as the geometric mean of a Laspyres index—measuring consumer willingness to accept compensation to give up the gains from the innovation—and a Paasche index, measuring their willingness to pay to receive the gains from innovation. Both are measured relative to the baseline, and neither is theoretically superior to the other. The Tornqvist index is an equally weighted geometric average of the two. 3 USING THE MODEL TO ESTIMATE CONSUMER BENEFITS FROM TWO ATP-FUNDED PROJECTS We apply the cost-index model to estimate expected consumer welfare gains from two innovations in digital data storage (DDS), both funded by the ATP and proposed and implemented by two U.S. companies. Imation Corporation and LOTS Technology, Inc. are the innovators. Both technologies are expected to offer faster writing and retrieval of digital data, and one of them would offer a large increase in storage capacity as well. One of them would pioneer the use of optical tape, and the other would replace helical with linear scanning of magnetic tape. The Imation Joint Venture: Tape and Cartridge System Imation Corporation of Oakdale, MN, is leading a research joint-venture project begun in the fall of 1995, and scheduled to last five years. The joint venture received $10.4 million from the ATP, to which member companies added $10.7 million. Other members of the joint venture are Pergrine Recording Technology of St. Paul, MN; Seagate Technology, Inc. of Costa Mesa, CA; and Advanced Research Corp. of Minneapolis, MN. The joint venture seeks to develop technologies required for a small, reliable, affordable tape recording and cartridge system that will record data at rates greater than 30 megabytes per second, with an ultimate goal of 100 megabytes per second. The first major goal in the development of that system is to develop a linear tape drive that can match or exceed the performance and capacity of helical-scan systems currently in use. 3 The theory of index numbers tells us that no single index satisfies all “desirable” properties or tests, e.g., tests related to scalability, transitivity, symmetry, and proportionality. The Tornqvist index satisfies many of the tests. For the mathematical formula we use to construct a Tornqvist cost index see our larger report from which this version is drawn, D. Austin and M. Macauley, Estimating Future Consumer Benefits from ATP-Funded Innovation: The Case of Digital Data Storage, NIST GCR 00-790, April 2000.
OCR for page 239
Page 244 The LOTS Award: Optical Technology LOTS Technology, Inc., of Sunnyvale, CA, received a single company award and began its two-year project in the fall of 1995. The company received nearly $2 million from ATP for its project, and put in approximately another million. The project goal was to develop an optical tape read/write technology capable of storing 1 trillion bytes (a terabyte) of information, and of transferring that data at a rate of at least 100 million bytes (100 megabytes) per second. This represents a 1,200-fold increase in capacity compared to the industry standard and a 100-fold increase compared to the next generation of cartridge storage tape drives currently being introduced. ESTIMATING BENEFITS We estimate how much better off consumers may be with these two innovations, as compared to staying with the existing or defender technologies which, as noted above, are assumed to continue to improve. We define the cost index relative to an aggressive baseline scenario where we assume that the best available performance of the defender technologies improves at the same rate as the innovations improve. To calculate the index, we use a simulation model containing 18 parameters, all but two of which are drawn from estimated probability distributions. We directly observe current prices and performance of the defender technologies, but we must forecast the initial values they will take in the model because the innovations, as of late 1999, had not yet been introduced, and the model's initial period is linked to the introduction of the new products. Our price and performance forecasts for the new products reflect the innovators' target, both at introduction and two-to-five years ahead. We assume that these targets reflect some “pioneer project bias”—a tendency for innovators to be overly optimistic about their projects. We make allowances for this by putting extra weight on potentially “disappointing” outcomes with respect to expected growth in market size, adoption rates, prices, and performance of the innovations. In contrast to our conservative treatment of the innovations, we use non-conservative assumptions in estimating the performance of the defender technologies. We make our forecasts for the parameter values on the basis of recent trends in leading DDS devices. To further provide a conservative estimate of consumer benefits for the ATP-funded innovations, we assume that the prices of the existing products will decline at the same rate as for the innovations. We estimate the shadow values by hedonic regression analysis of recent retail prices and performance characteristics. These regressions also produce estimated standard errors, which we use to construct the initial-period probability distributions for the shadow values. For the other parameters, we must use ad hoc rules of thumb for estimating the uncertainties. Again for a more conservative
OCR for page 239
Page 245estimate, we assume less uncertainty for the existing-product parameters than for innovations, even in the later years. Uncertainties Uncertainties derive from three main sources: (1) manufacturing and market conditions can vary; (2) the data are imperfectly observed; and (3) future outcomes cannot be predicted with certainty. Sensitivity tests reveal those parameters that are most important and indicate where it is necessary to test the results using alternative values. RESULTS Our analysis shows that both the linear scanning technology and the optical tape, if successfully introduced, are expected to generate large consumer benefits. Estimated Benefits Linear Scanning We estimate that the linear scanning technology, if successfully introduced, will yield $2.2 billion (median value) in present value consumer benefits over a five-year period, relative to the best existing technology. (Benefits from the probability distribution ranged from $1.3 billion to $3.2 billion between the 5th and 95th percentiles.) Figure 2 shows how the gains from linear scanning technology are expected to accumulate over our five-year forecasting window. 4 Optical Tape We estimate that the optical tape technology, if successfully introduced, would yield $1.5 billion (median value) in present value consumer benefits over five years relative to the best existing technology. (Benefits from the probability distribution ranged from $1.1 billion at the 5th percentile, to $1.9 billion at the 95th percentile.) Figure 3 shows how the gains from the optical tape technology are expected to accumulate over the same five-year forecasting window. 5 These estimates of benefits are very large even though it is assumed that the existing (defender) technology improves at a rate faster than historical rates. Furthermore, the estimates do not include benefits resulting from knowledge spillovers or from second-generation projects. On the other hand, if disk drive arrays 4 Under our assumption that the new DDS technologies would not have been developed within this five-year period without ATP assistance, these benefits can be attributed to the ATP. 5 Please note the difference in scale.
OCR for page 239
Page 246 ~ enlarge ~ FIGURE 2 Consumer benefit given successful introduction: Linear scanning innovation vs. defending products ~ enlarge ~ FIGURE 3 Consumer benefit given successful introduction: Optical tape innovation vs. defending products continue to make inroads into traditional tape storage markets, actual benefits will likely be lower than expected. We find the benefits estimates to be relatively insensitive to changes in the performance parameters, and somewhat more sensitive to price and rate of adoption data. However, within the expected range of parametric values, the results are robust.
OCR for page 239
Page 247 TABLE 1 DDS Innovations, Present Value of Consumer Welfare Gains over Five Years ($ billions, 2000) Percentile Linear Scanning Optical Tape 5th 1.25 1.05 25th 1.79 1.30 Median 2.16 1.45 75th 2.53 1.62 95th 3.17 1.88 CONCLUSIONS Our analysis has shown that gains in consumer benefits from ATP investments in digital data storage are likely to be substantial. If these DDS technologies would not have been developed without ATP assistance, our estimates measure gross early-commercialization benefits attributable to the ATP. Even if other, failed investments by ATP in DDS technology were included, just one success on the scale of our forecasts would outweigh the ATP's total annual investments in all areas of technology. Sensitivity analysis, in which we test the results to changes in the value of parameters, demonstrates the robustness of the basic conclusions. Where greater precision is desired, model simulations reveal the most important sources of uncertainty in the final benefit estimates, suggesting where additional research on the true values of individual parameters might be cost effective. Uncertainties and omissions may cause actual outcomes to differ from the forecasts. We did not include benefits from knowledge spillovers or follow-on improvements in this study. If knowledge spillovers occur, our benefit estimates here may be low. On the other hand, if disk drive arrays continue to make inroads into traditional tape storage markets, our benefit estimates will be high. A Useful Tool for Comparison The cost-index approach is a potentially useful tool for resource allocation by R&D managers in both the private and public sectors. Two of the model's strengths are that it incorporates uncertainty and it varies all of the parameters simultaneously, so that the implications of changes to the parameter values and assumptions can be seen within a unified framework. The model can be used to evaluate a single project's consumer benefits over time, and also to compare the future benefits of projects that may be competing for private-sector or government R&D funds.
OCR for page 239
Page 248 REFERENCES Austin, D. and M. Macauley. 2000 . Estimating Future Consumer Benefits from ATP-Funded Innovation: The Case of Digital Data Storage . NIST GCR 00-790. Timothy Bresnahan. 1986 . “Measuring the Spillovers from Technical Advance: Mainframe Computers in Financial Services.” American Economic Review . 76(4): 742-755 . Moulton, B. R. and K. E. Moses. 1997 . “Addressing the Quality Change Issue in the Consumer Price Index.” Brookings Papers on Economic Activity I . pp. 305-366 .