# Regulatory Impact Analysis for the Regulation of Microbial Products of Biotechnology Appendix D: Industry Cost Calculations and Options/Sensitivity Analyses

APPENDIX D: This appendix provides an explanation of the assumptions and estimates used to develop industry costs associated with microbial product submissions that would be required under the rule. In addition, it provides information on the options and sensitivity analyses used to generate costs associated with appropriate cases. The appendix is organized as follows: -
__Section A__provides a breakdown of the costs of the various types of submissions (e.g. TERA, MCAN) and recordkeeping requirements; -
__Section B__provides estimates of total industry costs resulting from various regulatory alternatives; and -
__Section C__provides the sensitivity analyses.
A. The steps and assumptions used to generate per-submission costs for TERAs, MCANs, and Tier I and Tier II exemptions are presented here. Costs were derived from data presented by Stanford Research Incorporated, International (SRI) on the amount of time and level of expertise needed to develop and submit information anticipated to be required for biotechnology Premanufacture Notices. The SRI data were originally presented with a probability rating (presented on a quartile basis) indicating the likelihood of the submitter having the information in-house as part of on-going or basic research for product development or that the data are otherwise available to the submitter. The probability for any particular data element is presented as a range (e.g., 51-75 percent). For this Regulatory Impact Analysis, the SRI data have been adapted to estimate the cost of developing and submitting certain information required in the rule. This section describes the cost areas identified from the rule and the method for deriving that information from the SRI data, as well as any assumptions that were necessary to quantify other parameters or to supply alternate estimates for each cost element. Tables D-1 and D-2 relate the SRI data elements to the rule categories of information for TERAs and MCANs respectively. Table D-1 Table D-2 There are several assumptions that lay the groundwork for this analysis. They can be summarized as follows: - It was assumed that for any particular data element identified from the SRI data, most companies submitting notification data would be likely to have at least some of the data of interest and would fall at the upper end of the assigned probability range (ETD Assumption). Probabilities are, therefore, applied to cost elements as the difference between unity and the upper bound of the probability range (e.g., if there is a 51-75 percent probability of having certain data, the cost of acquiring the data element is calculated as 25 percent of the associated unit cost).
- The costs for any data element anticipated as a requirement both for the TERA and MCAN submissions were counted only once. It was assumed that information or data developed in support of a TERA submission would need to be included in a subsequent MCAN package, but that there would be a negligible additional cost of including this information once a submission reaches the MCAN stage.
- Agency estimates of labor requirements and the percentage of cases to which these requirements were applied were used instead of the SRI data for some of the data elements required by the rule (OPTS 1992).
- The submissions will require managerial review of all information to be provided to EPA in order to verify for completeness and appropriateness of the data. (ETD Assumption).
This section discusses these cost components and provides the basis for determining the submission costs for each type of notification. The costs are presented according to the information requirements set forth in the rule for TSCA biotechnology submissions. After submission costs are derived, the number of submissions expected and the incremental costs associated with each submission are used to generate the total industry costs associated with all aspects of the rule including other non-submission factors, such as rule familiarization. These costs are presented in Section B of this analysis. 1. The labor rates used to predict costs based on time requirements provided by SRI have been developed by Kearney-Centaur and are as follows (Kearney-Centaur 1988): - -- $25.00 per hour for clerical employee;
- -- $42.78 per hour for research technician;
- -- $54.14 per hour for junior professional;
- -- $71.35 per hour for senior professional; and
- -- $103.99 per hour for research manager.
These hourly wage rates are used throughout the analysis of the costs to industry associated with the rule. For industry costs, the labor time associated with any given cost element of the regulation is presented under the appropriate labor category. In most cases, this time is presented in total weeks. In order to calculate labor costs, the number of hours was multiplied by the fully loaded hourly rate. In the case of weeks, the number of weeks was multiplied by the fully loaded hourly rate and that product was multiplied by 40. Once labor costs were calculated, the percent of cases was applied to labor costs in order to determine the average expected costs for each element of the TERA and MCAN. The "percent of cases" entry in the detailed cost tables is an estimate of the likelihood that a firm would not have the information readily available and is based on the SRI data. If a firm was likely to have the necessary information, it was assumed that the cost of providing this information for any of the reporting documents was negligible. The average expected cost for each element results from multiplying the expected labor costs by the corresponding percent of cases. The average expected costs for each data element were totalled to yield the average expected cost for each type of submission. In the cost component tables below, some cost elements have been divided down into subcategories in order to identify more clearly all cost components. In such cases, a subtotal of all cost factors: time, labor costs, percent of cases, and average expected costs is provided. For total time and average expected costs, the relevant items were summed. The subtotal for percent of cases was derived by dividing the subtotal of average expected costs by the unadjusted subtotal of labor costs. Totals were derived in the same manner. 2. Many of the costs associated with TERA submissions will be incurred over several years as submitters develop and provide material to EPA. It is assumed for this analysis, however, that all costs associated with developing and preparing the TERA information are incurred in the year in which the TERA is submitted. In addition, although it is likely that a submitter will have much of the information required for a TERA in-house, it is possible that some of the components of the rule will require the submitter to produce data that are not generated from basic research. The probability that the company does not have certain information is the basis for generating the average expected cost of preparing the TERA. The TERA cost components and the estimated costs for reporting are presented in Tables D-3 and D-4. 3. This section presents the components of the rule as they pertain to industry costs for MCAN submissions. It was assumed that each component of the per-submission costs would be incurred by each submitter and that the total costs per submission would include the costs of submitting the information to EPA and the costs of developing the information when necessary. Table D-3 Table D-4 Components of the Cost of Preparing a TERA High Cost Case The cost analysis for MCAN per submission costs is provided in Tables D-5 and D-6. 4. In order to estimate the baseline cost of reporting under thecurrent regulatory environment, it is also necessary to estimate the cost of submissions in the form of PMNs. Tables D-7 through D-10 present the estimated costs for these submissions. It is assumed that the same information would need to be included in these submissions as in a MCAN, but the costs per submission are different than for MCANs, because PMNs for environmental applications would not be preceded by previous submissions. For biotechnology PMNs for environmental applications, the Agency would expect to review all of the information from previous field tests that would have been reported in TERAs under the rule. Each submission is assumed to have been preceded by three field tests for which the cost of collecting information is assumed to be one third the cost of the scientific information collection costs for the PMN. 5. A detailed description of the costs associated with recordkeeping for the rule was presented in Chapter IV. The industry recordkeeping costs for the rule and the other regulatory options are presented again in this section. Tables D-11 and D-12 show estimates for recordkeeping costs for Year 1 and Year 5 for each option. B. Submission reporting costs and ICF Survey (ICF 1988) results were used to estimate the costs to industry associated with the rule. The baseline costs to industry under the current regulatory framework are subtracted to generate the total industry cost attributable to the rule. These costs are presented in Table D-13. Table D5. Components of the Cost of Preparing an MCAN Table D6. Components of the Cost of Preparing an MCAN Table D7. Components of the Cost of Preparing a Biotechnology PMN Table D8. Components of the Cost of Preparing an Biotechnology PMN Table D9. Components of the Cost of Preparing a Biotechnology PMN Table D10. Components of the Cost of Preparing a Biotechnology PMN Table D-11. Year 1 Costs of Recordkeeping Associated with Contained Experiments under TSCA Table D-12. Year 5 Costs of Recordkeeping Associated with Contained Experiments under TSCA Tables D-14 through D-17 present the total industry costs of implementing the 1986 Policy Statement (i.e., the current regulatory framework) through a rulemaking and three other alternative options. 1. There are five categories of costs that contribute to the total industry costs of the rule: rule familiarization; reporting costs; CBI substantiation costs, recordkeeping costs, and post-review monitoring costs. Reporting costs were estimated using the numbers of industry microbial submissions estimated in Appendix C. 2. Tables D-14 through D-17 present the cost analysis for various regulatory options (including the costs of subjecting all microorganisms to the requirements of the rule). Table D-18 presents the important elements of each of these options. C. In this section, a description of the variables examined and a discussion of the basis for alternative assumptions used in the sensitivity analysis are provided. The cases considered in the analysis, including cases in which several assumptions are altered at once, are then provided. Finally, the results of the analysis are presented and discussed. 1. The sensitivity analysis examines the effects of changing assumptions in four areas: the rates of growth in the number of products in research and in the markets subject to TSCA; the numbers of TERA submissions per released marketed product; the numbers of Significant New Use MCANs per released marketed product; and the rates of compulsory monitoring of released microorganisms at the research and general commercial use stage resulting from EPA action. Table D-13. Industry Costs Associated with the Final Rule: Intergeneric Only Table D-14. Industry Costs Associated with Implementing Current Policy with Rulemaking Table D-16. Industry Costs Associated with Regulatory Alternative 2: Include Naturally Occurring Table D-18. Options Matrix for Biotechnology Rulemaking In addition, because all cost estimates are performed using ranges of unit costs for submissions, the sensitivity of the results to changes in unit cost assumptions are clear. a. The number of submissions in later years will depend largely on the growth rate of products in R&D and entering the market. This report uses annual growth rate estimates based on survey responses to questions about current and anticipated numbers of products in R&D. The overall growth rate is in line with the average rate of growth in biotechnology products subject to TSCA anticipated for the next several years according to an industry trade association (IBA 1991). On the other hand, a study conducted by Ernst & Young (Burrill 1989) showed that growth rates can vary significantly across sectors of the biotechnology industry. Growth rates projected in the Ernst & Young study ranged from 8 percent (for medical diagnostic products) to 28 percent per year (for agricultural biotechnology products). To capture this range of uncertainty, this sensitivity analysis examined near-term growth rates of one half and two times the rates predicted by the survey. These changes in growth rate affect the expected baseline costs as well as the costs under the rule. b. This report assumes that each "new" product intended for environmental use will be associated with four TERAs before commercialization -- an initial TERA and three follow-on TERAs. The actual average number of TERAs per released new product is unknown, but prior submissions and conversations with researchers in industry and the scientific community suggest that between two and ten additional TERAs (some covering multiple strains) are likely to follow each initial TERA submission before a product isready for the market. The sensitivity analysis incorporates a range of assumptions about numbers and costs of follow-on TERAs based on characterizations of the industry R&D process (see Appendices A, B, and G). d. This report assumes that monitoring will be required by EPA for research and development environmental applications of reportable microorganisms, but that no monitoring will be required for environmental applications at the general commercial use level. It may be, however, that EPA will not require monitoring for field tests of some microorganisms. Conversely, there may be some microorganisms for which monitoring is required even after commercialization. Consequently, the sensitivity analysis considers cases both with higher and lower rates of mandatory monitoring, resulting in higher and lower unit cost estimates, respectively. 2. In all, eight sensitivity cases are compared to the base case in this analysis. Some of the sensitivity cases differ from the base case in only one area. For others, two or more factors have been varied. This section describes the salient features of each case, and then presents the results of the analysis. The assumptions in each case are presented in Table D-19 and the cost results are presented in Table D-20. a. This is the case described in the main text of this chapter and is presented here for purposes of comparison. b. Case 2 differs from the "Final Rule" (case 1) only in that it assumes the number of products in research and ready for commercialization grow at only one half the rate assumed in the rule. The results also show how the results would change if numbers of submissions were lower for reasons other than slow growth. Table D-19. Summary of Sensitivity Case Characteristics Table D-20. Year 5 Industry Costs for Sensitivity Cases c. Case 3 differs from the "Final Rule" only in that it assumes growth rates of twice those assumed in the base case. The results also show how the results would change if numbers of submissions were higher for reasons other than high growth, such as the addition of university reporting.
d. Case 4 differs from the "Final Rule" in that it assumes there will be one follow-on TERA for each initial TERA. e. Case 5 differs from the "Final Rule" in that it assumes there will be ten follow-on TERAs for each initial TERA. f.
Case 6 differs from the "Final Rule" only in that it assumes there will be 10 Significant New Use MCANs for each environmental application MCAN, while the "Final Rule" assumes there will be no Significant New Use MCANs. g. Case 7 differs from the "Final Rule" only in that it assumes that monitoring will be required for only 66 percent of TERAs and MCANs, rather than the 100 percent assumed in the rule. h. Case 8 differs from the "Final Rule" only in that it assumes that monitoring will be require for 20 percent of environmental applications at the general commercial use level, whereas the base case assumes no general commercial use monitoring. i. Case 9 recognizes that the real world could differ from the "Final Rule" in several ways, and that changes in several factors can compound one another. Case 9 combines several different assumptions to show how high the costs of the rule could be under an unfavorable constellation of circumstances. It differs from the "Final Rule" in that it assumes: growth at twice the expected rate for environmental applications,; ten follow-on TERAs per initial TERA, as in Case 5; and some general commercial use monitoring required as in case 8. 3. Table D-20 presents the results of the sensitivity analysis. For the Base Case and each of the 8 sensitivity cases, the table shows the number of submissions and the total cost to industry (in millions of dollars) projected for Year 5. Costs are shown both for the low and high unit cost assumptions. Sensitivity results are shown only for Year 5, as the sensitivity of the results is greater (particularly for changes in the growth rate) in the out-years. Changes in the growth rate would have pronounced effects on industry costs through their effects on the number of submissions. However it is important to note that the baseline costs would change with higher and lower growth rates and this mitigates the effect of the changes. Incremental costs under both the higher and lower growth rates assumed in Cases 2 and 3, therefore, are reduced. At generally lower growth rates, costs are reduced compared to the baseline because there are fewer additional environmental applications that have higher per product costs than under the baseline. At generally higher growth rates, costs are reduced because there are increased savings from closed system applications for which the rule has lower per product costs than for the baseline. Given that this industry is quite new and that its near-term prospects are uncertain, its true growth rate could easily turn out to be close to either end of the range considered. Uncertainty about costs stemming from uncertainty about the growth rate may not be a serious problem for policy makers, however. Changes in the growth rate change the total number of products subject to the rule, but will in themselves affect neither regulatory costs per product nor the incentives for developing products. Thus, uncertainty over how large the industry will be in the future does not increase the uncertainty about the relative burden the regulations will place on it. In addition, while the total costs rise with the size of the industry, the total benefits of the regulations can be expected to increase with the size of the industry. In contrast, uncertainty about the number of submissions that will be necessary for each product may have a greater effect. Higher numbers of submissions per final product raise costs per product, thereby increasing the relative burden imposed on the industry without necessarily increasing the benefits. The sensitivity analysis shows that increases in the number of submissions as a result of follow-on TERAs could have significant impacts on costs. A smaller number of TERA follow-ons (Case 4) would mean a drop in costs. If each initial TERA were associated with ten follow-on TERAs (Case 5), costs would also be higher by greater than 60 percent. Note that an increase in submissions for Significant New Use MCANs (Case 6) would not result in a change in incremental industry costs because these submissions would also occur under the baseline. Case 9, intended to show how much greater costs could be if several cost-increasing factors were combined, indicates that industry costs could be several times higher than estimated in the "Final Rule." Note that in order to provide the highest costs, only the growth rate for environmental applications (and not closed system applications) was increased for this case. |