******************************************************** NOTICE ******************************************************** This document was converted from WordPerfect or Word to ASCII Text format. Content from the original version of the document such as headers, footers, footnotes, endnotes, graphics, and page numbers will not show up in this text version. All text attributes such as bold, italic, underlining, etc. from the original document will not show up in this text version. Features of the original document layout such as columns, tables, line and letter spacing, pagination, and margins will not be preserved in the text version. If you need the complete document, download the WordPerfect version or Adobe Acrobat version, if available. ***************************************************************** Before the Federal Communications Commission Washington, D.C. 20554 In the Matter of ) ) Price Cap Performance Review ) CC Docket No. 94-1 for Local Exchange Carriers ) ) Access Charge Reform ) CC Docket No. 96-262 ) FURTHER NOTICE OF PROPOSED RULEMAKING Adopted: November 12, 1999 Released: November 15, 1999 Comment Date: December 30, 1999 Reply Comment Date: January 14, 2000 By the Commission: I. INTRODUCTION 1. In 1997, the Commission represcribed the amount by which it annually adjusts price caps for incumbent local exchange carriers subject to the price cap rules ("price cap LECs"). The revised price cap adjustment required price cap LECs to reduce inflation-adjusted prices for interstate access services by an "X-factor" of 6.5 percent annually. Pursuant to petitions for review of the Commission's order, the United States Court of Appeals for the District of Columbia Circuit reversed and remanded the Commission's decision. The court has stayed issuance of its mandate until April 1, 2000, to allow time for us to conduct this proceeding. 2. In this Notice we seek comment on how we should represcribe an X-factor. More specifically, we seek comment on prescribing two separate X-factors to address retroactively the period affected by the court remand (July 1, 1997 to June 30, 2000), and prospectively the period from July 1, 2000 forward, or a single X-factor to cover the combined period. Further, we seek comment on resetting, on a forward-looking basis, price cap LEC prices to a level that is consistent with any X-factor prescription in order to rebalance the sharing of benefits of price caps between LECs and their customers. This Notice is limited to issues surrounding the setting of the X-factor, and does not include any broader changes to our method of price cap regulation. 3. We seek comment on represcribing the current X-factor from the reasonable range determined in the 1997 Price Cap Review Order. In the alternative, we also seek comment on represcribing the X- factor based on the results of a staff study of the rate of growth in total factor productivity ("TFP") for price cap LECs. A third alternative is to prescribe an X-factor based on the results of another staff study which directly determines, from aggregate interstate expenses and revenues, the X-factor that would have produced a competitive level of capital compensation in the interstate jurisdiction during the period between performance reviews. We seek comment on these alternative methods for represcribing the X-factor. We also seek comment on whether we should rely on one of these alternatives to prescribe an X-factor on a going-forward basis, and, if so, whether we should use data from the same time period as for the period affected by the court's remand. In addition, we seek comment on studies submitted in the record that propose to quantify the consumer productivity dividend ("CPD") necessary to offset the elimination of sharing requirements from the price cap rules. 4. In a separate but related proceeding, the Commission is seeking comment on a proposal submitted by the Coalition for Affordable Local and Long Distance Services ("CALLS"). The CALLS proposal would purportedly eliminate the necessity of retrospectively adjusting the X-factor in response to the court's remand. Instead, it would keep the X-factor at 6.5 percent, but would target X-factor reductions to the traffic-sensitive price cap basket. Once local switching rates reached a certain level, all price cap indices would be frozen. Adoption of the CALLS proposal would also eliminate the need to prescribe an X-factor on a going-forward basis. We seek comment in this proceeding on the prescription of the X-factor because, in the event that the CALLS proposal is not adopted, or not all price cap LECs become signatories to the proposal, the Commission must be prepared to prescribe a new X-factor before April 1, 2000. II. BACKGROUND A. Prior Commission Decisions 5. The Commission has determined that competition should be the model for setting just and reasonable LEC rates because: Effective competition encourages firms to improve their productivity and introduce improved products and services, in order to increase their profits. With prices set by marketplace forces, the more efficient firms will earn above-average profits, while less efficient firms will earn lower profits, or cease operating. Over time, the benefits of competition flow to customers and to society, in the form of prices that reflect costs, maximize social welfare, and efficiently allocate resources. In 1990, the Commission determined that an incentive-based price cap system would more closely represent the results of a competitive market than did the prior regulatory method of rate-of-return regulation. Specifically, the price cap plan was "designed to mirror the efficiency incentives found in competitive markets . . . by encouraging LECs to move prices for interstate access services to economically efficient levels, to reduce costs, to invest efficiently in new plant and facilities, and to develop and deploy innovative service offerings." In order to promote consumer welfare and economic efficiency, ceilings were set on prices that were intended to allow carriers to cover their costs and earn a normal, competitive rate of return. If prices were set too high, consumers would fail to reap the benefits of the carriers' efficiency; if prices were set too low, the return on capital would be insufficient to attract investment into the industry. 6. Regulatory structures that base a firm's allowed rates directly on the reported costs of the individual firm create perverse incentives, because reimbursing the firm's costs removes the incentive to reduce costs and improve productive efficiency. The Commission's price cap plan for LECs avoids this problem in part by divorcing the annual rate adjustments from the performance of each individual LEC, and in part by adjusting the cap for experience only with a considerable lag. Individual companies retain an incentive to cut costs and to produce efficiently, because in the short run their behavior has no effect on the prices they are permitted to charge, and they will be able to keep any additional profits resulting from reduced costs. The introduction of LEC price cap regulation was expected to stimulate cost reduction and accelerate technological innovation because the regulated firms would be able to benefit from such behavior as they could not do under rate-of-return regulation. 7. To achieve these goals, the Commission's LEC price cap scheme allows prices to increase by a measure of inflation minus a productivity offset, or X-factor. The X-factor represents the amount by which LECs can be expected to outperform economy-wide productivity gains. The Commission has periodically adjusted the LEC price cap plan to ensure that it continues to provide strong incentives to incumbent LECs to provide a substantial benefit to customers, while not basing permitted prices explicitly on individual firms' costs. 8. The X-factor adopted in the LEC Price Cap Order initiating price cap regulation for the largest LECs included an additional 0.5 percent consumer productivity dividend ("CPD") to ensure "direct benefits to ratepayers." The CPD was included to account for an expectation that, because of the more efficient regulatory scheme being adopted, future productivity growth would be faster than measured past productivity growth. 9. Initially, price cap LECs were required to share a portion of their earnings in excess of specified rates of return with their access customers by temporarily reducing the price cap ceiling in a subsequent period. In 1990, the Commission prescribed two X-factors; a minimum 3.3 percent X-factor, or an optional 4.3 percent X-factor. Price cap LECs that opted to use the higher X-factor were allowed to retain larger shares of their earnings. The LEC Price Cap Order required that the Commission periodically review the performance of the price cap regime. The order in the first performance review was released in 1995, at which time the Commission increased the minimum X-factor from 3.3 percent to 4.0 percent, and provided two optional X-factors at 4.7 and 5.3 percent. The 1990 and 1995 prescriptions were derived from two 1990 staff studies that examined LECs' historical unit cost changes. These studies did not utilize a total factor productivity ("TFP") methodology. 10. In the next performance review order, released in 1997, the Commission further revised the price cap plan by eliminating all sharing requirements and prescribing a new X-factor of 6.5 percent. This X-factor prescription relied primarily on a staff study of the historical rate of growth in LEC TFP ("1997 staff TFP study"). 11. TFP measurement is a methodology commonly used to measure productivity and productivity growth in the economy as a whole. Productivity is measured as the ratio of an index of the outputs of a firm (or industry, or nation) to an index of its inputs. Productivity growth is measured by changes in this ratio over time. Beyond showing changes in the amount of input required per unit of output, TFP analysis generally does not shed light on the mechanisms by which productivity growth occurs. 12. In TFP models, output can be measured either in terms of physical units of the service produced, such as minutes or calls, or by dividing revenues by an index of output prices. Output indices are created to measure changes in the level of outputs over time. Indices for particular categories of output are weighted to create a single output index. 13. Inputs are usually classified into three categories: labor, materials, and capital services. Again, indices reflecting changes in the quantities of labor, materials, and capital services are weighted and aggregated into a single input index. The growth rate of the aggregate input index depends on the growth rates of the individual input indices and their relative weights. Capital services are assumed to be a fixed proportion of the capital stock (i.e., plant and equipment), so that changes in capital services can be measured through measurement of changes in the capital stock. 14. In a regulatory setting, if the TFP calculation sets the X-factor too low, and, consequently, sets prices too high, end users will purchase less of the services produced, and the quantity of output will be lower than if prices were set at a competitive level. The productivity of which the plant is capable will not be revealed. Since the marginal cost of additional output is believed to be very low in telecommunications, the effect on measured productivity may be large. 15. The 1997 staff TFP study calculated the historical difference in productivity growth between LECs and the economy nationwide for the period 1986 through 1995. Specifically, it calculated the difference between LEC TFP change and economy-wide TFP change. The study then calculated an input price differential reflecting the difference in the rate of the change of LEC input prices as compared with the economy as a whole. These two factors were then added together for each year. Several averages were created using these numbers, with the first average being for the entire period, the next average dropping the oldest year (1986), and each subsequent average dropping another year, with the final average including only the years 1991 to 1995. These averages created a "zone of reasonableness" of 5.2 percent to 6.1 percent. Placing some reliance on slightly higher X-factor estimates calculated by AT&T, the Commission increased the upper bound of the zone of reasonableness to 6.3 percent. The Commission prescribed an X- factor near the upper end of the zone of reasonableness, at 6.0 percent. 16. In addition to the 6.0 percent historical portion of the X-factor, the Commission retained a 0.5 percent CPD "to ensure that price cap LECs flow-through a reasonable portion of the benefits of productivity growth to ratepayers." The Commission stated that inclusion of a CPD was especially critical in achieving this goal because the sharing requirements were being eliminated from the price cap plan. B. Court Decision 17. Several entities filed petitions for review of the 1997 Price Cap Review Order. In its decision, the U.S. Court of Appeals for the D.C. Circuit generally denied the petitioners' challenges to the Commission's decision, but reversed and remanded for further explanation of several aspects of the analysis supporting the Commission's prescription of a 6.5 percent X-factor. 18. The court questioned the Commission's stated rationales for selecting 6.0 percent, from the high end of the 5.2-6.3 percent zone of reasonableness, as the historical component of the X-factor. Specifically, the court found that the Commission had not supported its conclusion that the two lowest TFP year averages, from 1986-1995 and 1991-1995, should be accorded less weight in the selection of the X- factor. The court also questioned the Commission's reliance on an upward trend in the X-factor from 1993, noting that the trend could be part of a larger cyclical pattern, in which case a downward turn in the X- factor could be expected. In addition, the court noted that there was no discernible trend in either of the two X-factor components, i.e., the differences between LEC and U.S. changes in TFP, and the differences between LEC and U.S. changes in input prices. Finally, the court rejected the Commission's reliance on AT&T's higher X-factor results, noting that the Commission had incorporated the portions of AT&T's study deemed reasonable in its staff study, and therefore should not have given any independent weight to the results of AT&T's study, or adjusted the range of reasonableness upward from 6.1 percent. 19. The court also sought an explanation of the inclusion of a 0.5 percent CPD in the X-factor. Although the petitioners did not dispute "that it is defensible to include a CPD corresponding to whatever productivity increase may be expected from the elimination of sharing", the court found that retention of the prior CPD amount of 0.5 percent required a comparison of the productivity effects of sharing elimination with the initial rationale for adopting a 0.5 percent CPD. III. REPRESCRIPTION OPTIONS 20. We seek comment on three alternative bases for prescribing the historical component of the X- factor. First, we could prescribe an X-factor based on the reasonable range determined by the 1997 staff TFP study. In relying on this basis for represcription, we would address only those issues remanded by the court. We discuss this approach in Section III.A. 21. Second, we could rely on a new staff study revising the 1997 staff TFP study ("1999 staff TFP study"). The 1999 staff TFP study substitutes an independent measure of capital price changes for the endogenously-determined capital price change used in the 1997 staff TFP study and recalculates the change in the compensation of the capital input. The revised study also makes adjustments to the 1997 staff TFP study to update information and to adjust for possible biases. We discuss this approach in more detail in Section III.B. The 1999 staff TFP study appears as Appendix B. 22. Third, we could take advantage of our accumulated experience with price cap regulation and directly determine, from aggregate interstate expenses and revenues, the X-factor that, if it had been prescribed from the inception of price caps, would leave capital compensation at the competitive level at the end of the study period. We discuss this methodology in detail in Section III.C. A staff study that applies this methodology appears as Appendix C ("staff Imputed X study"). 23. We also seek comment on other alternatives that would serve as a basis for prescribing the historical component of the X-factor for the remand period. A. Option 1: The 1997 Staff TFP Study 24. We seek comment on whether we should use only the results from the 1997 staff TFP study in setting the historical component of the X-factor for the remand period. We seek comment on whether, in addressing the court's remand, we are precluded from revising the X-factor using any other methodology, or from supplementing the data in the 1997 staff TFP study. 25. The court did not find fault with the 1997 staff TFP study, and did not ask us to revisit it. Instead, the court limited its critique of TFP to our selection of a value at the upper end of the reasonableness range, and with the upward adjustment to the reasonable range. 26. In their responses to a 1998 request to refresh the record in our Access Charge Reform proceeding, both USTA and AT&T used the methodology in the 1997 staff TFP study to extend the calculation of the X-factor through 1997. USTA has also calculated an X-factor for 1998. We seek comment on the legal and logical arguments supporting consideration of data that have become available after the close of the record for the remanded prescription. We note that USTA and AT&T did not agree with each other on the value of the historical component for 1996 and 1997. We seek comment on USTA's and AT&T's updates of the 1997 staff TFP study, and on their recommendations for prescribing an X- factor. 27. If we set the X-factor by using the 1997 staff TFP study, the court's remand requires that we justify our selection from within the reasonable range. Within the reasonable range, should we use some measure of central tendency, e.g., the mean or median, as the best estimator of productivity? Could and should we consider prescribing above the mean? If the reasonable range includes a statistically meaningful trend, should this inform our choice? What other justifications could be made for selecting above or below some measure of central tendency? Should these justifications affect our selection from the reasonable range, or are they more relevant to the selection of a CPD? B. Option 2: The 1999 Staff TFP Study 1. Methodology 28. In comments filed with the Commission late last year, several parties identified what they believe is a problem in the way in which the 1997 staff TFP study employed the TFP methodology commonly used in economic analysis to set an X-factor. The 1999 staff TFP study takes this potential problem as a point of departure and attempts to correct it. We seek comment on the 1999 staff TFP study, and on its premise that the 1997 staff TFP study methodology may fail to calculate an X-factor that is consistent with the objectives of our price cap plan. 29. The 1997 staff TFP study subtracts the cost of the labor and material inputs from revenues, and the residual revenue is assumed to be the cost of the capital input. The 1999 staff TFP study attempts to capture the gains in productivity that would have been revealed in a competitive marketplace by varying total capital compensation according to a measure of the competitive capital compensation rate. 30. We seek comment on the following method for adjusting the capital compensation in the 1997 staff TFP study. The first step is to identify a competitive price index series to use as a surrogate for the annual change for the cost of capital in a competitive market. The second step is to assume LEC capital compensation in 1991, the first full year of LEC price cap, was at a competitive level. The third step is to combine the competitive price index and the 1991 LEC capital compensation rate to create a competitive LEC capital compensation rate for the historical period. The fourth step is to increase or decrease LEC capital compensation based on this competitive LEC capital compensation rate. The fifth step is to adjust LEC revenues, making appropriate allowance for taxes, for the change in capital compensation. The final step is to recalculate LEC historical TFP using these revised capital compensation and revenue data. 31. In addition to updating the data for the period 1996-1998, the 1999 staff TFP study makes three other modifications to the 1997 staff TFP study. First, the 1999 staff TFP study uses the recently revised Bureau of Labor Statistics ("BLS") series on multifactor productivity in place of the antecedent series. Second, the 1999 staff TFP study uses the number of dial equipment minutes, rather than the number of calls, in calculating the local service output index. Third, the 1999 staff TFP study recalculates the labor input to adjust for the fact that all the costs, but only a fraction of the benefits, of the 1992-95 employee buyouts have been recognized on the accounting books. We seek comment on these modifications to the 1997 staff TFP study. 32. Several additional aspects of the 1997 staff TFP study may warrant highlighting and comment. The 1999 staff TFP study does not make these adjustments because they either are not easily quantified, or do not make a significant impact on the level of the X-factor. We seek comment on the decision of the 1999 staff TFP study to not make any of these adjustments. We also seek comment on whether there are any additional issues, not noted here or in Appendix A, that necessitate adjusting the X- factor, how any such adjustments would affect the X-factor, and how they should be made. 2. Selecting the Appropriate X-Factor 33. The court's remand requires that we justify our selection from within a reasonable range. We seek comment on how we should determine the reasonable range and how we should select from within this range. In our determination of the reasonable range in the 1997 Price Cap Review Order, we gave recent years more weight than more distant years. Should we continue to discount more distant years? Should the period under price cap regulation be given more weight than the period under rate-of-return regulation? Given that price cap regulation may have been anticipated by price cap LECs for some years before its introduction, what years should be included in the price cap period? 34. We also seek comment on whether additional years of data should be considered in the remand, or whether the X-factor we select should rely on the same years of data as used in the 1997 Price Cap Review Order. We seek comment on the legal and logical arguments supporting consideration of data that have become available after the close of the record for the remanded prescription. Would it be more responsive to the court's remand to prescribe an X-factor based on data available in 1997 or to consider the additional data that has become available in the interim in setting the X-factor on a going-forward basis? C. Option 3: The Staff Imputed X Study 1. Methodology 35. As an alternative to either of the TFP methodologies, the Bureau staff also has performed a study, the staff Imputed X study, designed to calculate the X factor that yields the aggregate revenues that would have been generated in a competitive market. While price caps provide incentives for cost reduction similar to those of competition, they do not guarantee that revenues will follow a similar path. In a competitive market, revenues on average will be equal to costs, including compensation of capital at a competitive market level. This method is intended to replicate the effects of a competitive market in apportioning the gains from successful operation between carriers and consumers. The approach used here differs from the TFP approach, inter alia, in that it measures productivity growth by looking at aggregate expense and revenue data rather than by weighting and aggregating categories of physical inputs and outputs. In contrast to both of the TFP approaches, this method appears to have modest data requirements and to be computationally simple and easily understandable. Nevertheless, this method should have the same incentive effects as the TFP approach or any other method of calculating an X-factor. 36. The staff Imputed X study calculates the change in 1998 revenue and operating income for each price cap LEC that would result from imposing a hypothetical X-factor from the inception of price caps in 1991 through 1998. The results for all price cap LECs are aggregated, and the X factor required to produce revenues equal to costs, including a competitive level of capital compensation in the aggregate for all LECs, is calculated. The calculation was also performed for 1991 through 1995 for comparison with the original TFP study. The calculation takes account of the increase in the demand for service that would have resulted from the lower price. Changes in the competitive cost of capital were accounted for by adjusting the capital compensation found reasonable by the Commission at the inception of price caps by an index of bond rates over the period. The index is the same one used for the 1999 staff TFP study to measure the price of capital. 37. The data used for these estimates differ from those used for the TFP calculations in that they are purely interstate in nature. The TFP calculations used total company data because of the difficulty of separating interstate and intrastate costs for the TFP calculations, despite interstate data being conceptually more appropriate for representing the services regulated by the Commission under price caps. The data for the staff Imputed X study also include all price cap carriers, whereas the TFP studies use data for the regional Bell operating companies ("RBOCs") only. The calculations assume that a decrease in price would result in an increase in the quantity of service purchased, while the TFP calculations necessarily reflect only experience under the prices that were actually in effect. Finally, the staff Imputed X study does not make an adjustment in expense data comparable to the adjustment made in the 1999 staff TFP study to compensate for the accounting treatment of employee buyouts. To provide a check on the revised TFP calculations, the X-factor calculations using the staff Imputed X study were repeated using data only for the RBOCs and assuming no demand growth in response to lower prices. These calculations were performed for both 1995 and 1998. 2. Previous Study 38. We note that the approach described here is similar to the Direct Model proposed by AT&T, which the Commission has referred to as the Historical Revenue Approach. The staff Imputed X study differs from the approach proposed by AT&T primarily in that the staff calculation includes an adjustment to take account of likely demand stimulation resulting from a lower price cap, and the calculation takes account of changes over time in competitive return to capital. Data sources and calculations also differ somewhat. In the 1995 Price Cap Review Order, the Commission noted that the Historical Revenue Approach has the advantage that it reflects performance in providing the interstate services that are subject to price caps, and includes input cost changes. In comments to the Price Cap Fourth Further Notice, GSA supported the Historical Revenue Approach and noted that it incorporates both TFP growth and the input price differential. 39. Most criticisms of AT&T's Historical Revenue Approach dealt with the data and methodology used by AT&T in its calculations. Commenters responding to AT&T's proposal pointed out that data reported under Commission accounting, separations, and other rules may not accurately track economic costs. NYNEX in its comments to the Price Cap Fourth Further Notice criticized use of the Historical Revenue Approach on the grounds that accounting-based rules are a poor measure of a firm's economic performance. We note that the Commission declined to adopt the Historical Revenue Approach in the 1997 Price Cap Review Order due to administrative concerns and incentive effects. 3. Request for Comment 4. We seek comment on the validity of the staff Imputed X study for estimating the appropriate level of the X-factor. Does the X-factor estimated using these data and assumptions accurately represent the productivity growth achievable by the price cap LECs over the period examined? We request comment on the theoretical appropriateness of this methodology. We also seek comment on the following questions: Is an interstate-only calculation conceptually proper, and do the data allow an accurate measure of interstate revenues, expenses, and investment? Calculations reported in Appendix C show that X-factors calculated on an annual basis appear to increase over time. Are there explanations for the trend we see other than increasing efficiency? Does this apparent trend suggest that an additional adjustment, such as the CPD, is necessary in addition to revising the calculation of the X-factor? Alternatively, is the CPD no longer necessary because the approach described here sufficiently passes the benefits of increased efficiency to ratepayers? What is the appropriate method for determining the competitive cost of capital? Is applying an index of bond rates to the rate of return used by the Commission to initialize rates at the inception of price caps a reasonable approach? Would taking account of the mix of debt and equity held by the LECs yield a more accurate estimate of the trend in the cost of capital? 5. We request comment on the data and calculations used in the staff Imputed X study. Are more appropriate data sources available, and can adjustments be made that would improve the accuracy of the calculations reported here? AT&T in its Historical Revenue Approach in 1994 used Price Cap Indices ("PCIs") from the Commission's Tariff Review Plan data to measure actual changes in allowed rates. This approach includes all changes that occurred in the price caps, including exogenous changes not related to the operation of the X factor. Is such an approach conceptually appropriate? Would use of PCIs rather than the X factor in effect more accurately reflect price performance for purposes of these calculations? 6. We also seek comment on whether, in responding to the remand, it is appropriate to use data for the period that was available to us at the time of the 1997 Price Cap Review Order, or whether we should make use of the best information available to us now, including data for subsequent years that have become available in the meantime. We seek comment on the legal and logical arguments supporting consideration of data that have become available after the close of the record for the remanded prescription. Would it be more responsive to the court's remand to prescribe an X-factor based on data contemporaneous with the prescription and to consider the additional data in setting the X-factor on a going-forward basis? In addition, the court's remand requires that we justify our selection from within a reasonable range. How should we determine a reasonable range for setting the X-factor using the staff Imputed X study, and how we should select from within that range? IV. CONSUMER PRODUCTIVITY DIVIDEND 7. In the LEC Price Cap Order, the Commission included a CPD of 0.5 percent in the X-factor offset to ensure that access customers received the first benefits of price caps in the form of reduced rates. This CPD was also included in the X-factor in subsequent price cap review orders, including the 1997 Price Cap Review Order, in which it was intended to offset the elimination of sharing requirements. These requirements had compelled price cap LECs to share a portion of their earnings above set percentages with access customers. The sharing requirements were intended to protect consumers against the possibility of an error in the establishment of the X-factor. Pursuant to the court's remand, the Commission seeks comment on whether to retain the CPD. 8. In remanding this issue to the Commission, the court specifically questioned the quantification of the CPD. When the Commission made its decision to include a CPD in the 1997 X-factor, the record included a study by Strategic Policy Research ("SPR") that addressed the effects of eliminating the sharing requirements. The SPR study found that the LEC price cap plan with sharing requirements produced less than 35 percent of the efficiency incentives of unregulated competition. Those incentives decreased to 18 percent for price cap LECs whose earnings were in the 50-50 sharing category for each year of the four- year review cycle. The results of the SPR study were challenged by the Ad Hoc Telecommunications Users Committee ("Ad Hoc"), but Ad Hoc's own results indicated that sharing substantially reduced efficiency incentives. Ad Hoc's more conservative calculations indicated that elimination of sharing would increase efficiency incentives by at least 17 percent for all LECs, and by 41 percent for LECs in the 50-50 sharing category. We seek comment on the CPD amount justified on the basis of these studies to ensure that the benefits of sharing elimination would be apportioned between LECs and ratepayers. We also seek comment on additional methods for quantifying a CPD designed to ensure that consumers get a reasonable portion of the benefits from the elimination of sharing. 9. We also seek comment on whether a CPD should be included to reduce rates and correct for prior years when the X-factor may have been set too low. As noted above in Section III, the calculations used to set prior year X-factors may have underestimated LEC productivity. This underestimation may have caused rates to be set at too high a level. A mistake in the X-factor may not be self-correcting, but instead may cause increasingly erroneous prices over time. To obtain efficient prices in the future, it may be necessary both to adjust the value of the X-factor and to reset prices. Therefore, we seek comment on whether we should include in the X-factor a CPD designed to reduce rates, either by a one-time adjustment, or over a multi-year period, if we conclude that the X-factor historically has been set too low. If the reduction occurs over a multi-year period, should we account for the time value of money, and, if so, how should we calculate the reduction? The following table sets forth what CPD might be appropriate to correct for productivity underestimations in prior year X-factors depending on the X-factor that we prescribe. RATE REDUCTION IN 1998 IMPLIED BY NEW X-FACTOR PRESCRIPTION ______________________________________________________________________________ Overstatement Annual Consumer X-Factor of Price Indices Benefits Lost (%) (%) ($ billion) ______________________________________________________________________________ 5.0 0.14 - 0.44 0.24 5.5 3.19 - 4.36 1.16 6.0 7.10 - 8.14 2.06 6.5 10.70 - 11.79 2.95 7.0 14.26 - 15.31 3.80 7.5 17.69 - 18.71 4.64 8.0 21.00 - 21.98 5.45 8.5 24.20 - 25.15 6.23 ______________________________________________________________________________ V. PRESCRIBING THE X-FACTOR ON A GOING-FORWARD BASIS 10. We seek comment on whether we should prescribe an X-factor that would apply as of July 1, 2000 that is different from the retrospective X-factor applicable to the period affected by the court's remand, or whether the X-factor that we prescribe for the period beginning July 1, 1997 should continue in place until the next price cap performance review. We also seek comment on whether to include a prospective CPD adjustment in future X-factors to correct for any significant divergences between historic LEC productivity and prior X-factors, and on whether any such adjustment should be made at once or be phased in over several years. 11. In this Notice we seek comment on prescribing a future X-factor based on the results of the 1999 staff TFP study. In the alternative, we could prescribe an X-factor based on the results of the staff Imputed X study. Finally, we invite parties to comment on other alternatives that could serve as a basis for a future X-factor. 12. We also seek comment on how the prescription of the X-factor would affect smaller price cap LECs differently from other price cap LECs, and whether there should be a separate X-factor calculated for smaller price cap LECs. 13. In addition, we seek comment on how the Commission's proposed adjustments to the price cap rate structure should affect the annual reductions required by our price cap rules. We proposed in our recent Pricing Flexibility Order to add a "q" factor to the formulae used to adjust annually the price cap indices ("PCIs") for the baskets that contain the charges for local switching and tandem switching. The q factor would reduce switching charges based on growth in demand. As proposed, the affected baskets would be reduced annually by both the X-factor and the q factor. The staff studies attached herein, however, may capture in their X-factor estimates some or all of the effect intended to be captured by the q factor. We seek comment on whether a q factor is necessary if an X-factor is adopted that captures its effect, and on how to remove any double counting that might result from the application of both factors. For example, if the X-factor reduction was $10, and the q factor reduction was $4, then we could directly apply $4 to the baskets containing local and tandem switching, and allocate the remaining $6 amongst all the baskets according to our price cap rules. 14. We also proposed to adjust on a prospective basis for the past absence of a q factor in the formulae that annually adjust the PCIs of the baskets containing charges for local and tandem switching. We seek comment on how any such adjustment should affect any proposed adjustment to the PCIs for all price cap baskets to offset the cumulative effect of past X-factors that may have been set below the rate of cost reduction actually achieved by LECs. Should we apply the logic suggested in the example of the previous paragraph? If so, should the shift of switching ports to common line increase the common line basket's share of any adjustment based on the past absence of a q factor? 15. In addition to proposing a q factor, we proposed to increase the "g" factor that applies to certain revenues in the common line basket from g/2 to a full g. We seek comment on whether any prospective adjustment to our X-factor prescription would be appropriate to account for this. 16. Finally, we proposed to replace the existing per-minute rate structure for local switching and tandem switching with capacity charges. We seek comment on whether replacing per-minute charges with capacity charges affects future growth in LEC productivity. We seek comment on whether any prospective adjustment to our X-factor is required and on how we would quantify this adjustment. VI. PROCEDURAL ISSUES A. Ex Parte Presentations 17. This proceeding shall be treated as a "permit-but-disclose" proceeding in accordance with Section 1.1206(b) of the Commission's rules, 47 C.F.R.  1.1206(b). Ex parte presentations are permissible if disclosed in accordance with Commission Rules, except during the Sunshine Agenda period when presentations, ex parte or otherwise, are generally prohibited. Persons making oral ex parte presentations are reminded that memoranda summarizing the presentations must contain summaries of the substance of the presentations and not merely a listing of the subjects discussed. More than a one or two sentence description of the views and arguments presented generally is required. See 47 C.F.R.  1.1206(b)(2). Additional rules pertaining to oral and written presentations are set forth in Section 1.1206(b). B. Initial Regulatory Flexibility Act Analysis 18. As required by the Regulatory Flexibility Act ("RFA"), the Commission has prepared this Initial Regulatory Flexibility Analysis ("IRFA") of the possible significant economic impact on small entities by the policies and rules proposed in this Notice. Written public comments are requested on this IRFA. Comments must be identified as responses to the IRFA and must be filed by the deadlines for comments on the Notice provided below in Section VI.C. The Office of Public Affairs will send a copy of the Notice, including this IRFA, to the Chief Counsel for Advocacy of the Small Business Administration. In addition, the Notice and IRFA (or summaries thereof) will be published in the Federal Register. 19. Need for and Objectives of the Proposed Rules. As stated in Section II.B above, the court has remanded to the Commission the selection of a 6.5 percent productivity offset, or X-factor, in the LEC price cap formula. In this Notice we seek comment on how we should represcribe an X-factor. In Section III we seek comment on prescribing one or more X-factors to address retroactively the period affected by the court remand (July 1, 1997 to June 30, 2000), and in Section V we seek comment on represcribing one or more X-factors from July 1, 2000 forward. Further, we seek comment on resetting, on a forward- looking basis, price cap LEC prices to a level that is consistent with any X-factor prescription in order to rebalance the sharing of benefits of price caps between LECs and their customers. 20. Legal Basis. The proposed action is supported by Sections 1, 4(i), 4(j), 201-205, and 303(r) of the Communications Act of 1934, as amended, 47 U.S.C.  151, 154(i), (j), 201-205, and 303(r). 21. Description and Estimate of the Number of Small Entities to Which the Proposed Rules Will Apply. The RFA directs agencies to provide a description of and, where feasible, an estimate of the number of small entities that may be affected by the proposed rules, if adopted. The RFA generally defines the term "small entity" as having the same meaning as the terms "small business," "small organization," and "small governmental jurisdiction." In addition, the term "small business" has the same meaning as the term "small business concern" under the Small Business Act. A small business concern is one which: (1) is independently owned and operated; (2) is not dominant in its field of operation; and (3) satisfies any additional criteria established by the Small Business Administration ("SBA"). The SBA has defined a small business for Standard Industrial Classification ("SIC") category 4813 (Telephone Communications, Except Radiotelephone) to be an entity that has no more than 1,500 employees. 22. We have included small incumbent LECs in this RFA analysis. As noted above, a "small business" under the RFA is one that, inter alia, meets the pertinent small business size standard (e.g., a telephone communications business having 1,500 or fewer employees), and "is not dominant in its field of operation." The SBA's Office of Advocacy contends that, for RFA purposes, small incumbent LECs are not dominant in their field of operation because any such dominance is not "national" in scope. We have therefore included small incumbent LECs in this RFA analysis, although we emphasize that this RFA action has no effect on Commission analyses and determinations in other, non-RFA contexts. 23. The proposals in the Notice apply only to price cap LECs. At the current time, there are 13 price cap LECs. Of these companies, 11 are listed in the Commission's most recent Statistics of Communications Common Carriers ("SOCC") report as having more than 1,500 employees. Consequently, we estimate that 2 or fewer providers of local exchange service are small price cap LECs that may be affected by these proposals. 24. Description of Projected Reporting, Recordkeeping and Other Compliance Requirements. We expect that, on balance, the proposals in this Notice will not change price cap LECs' administrative burdens or cause price cap LECs to incur any additional costs associated with proposed reporting and recordkeeping requirements. The studies discussed in Section III would establish new X-factors that price cap LECs would need to utilize in their price cap calculations, but otherwise should not affect their administrative burdens or costs. 25. Steps Taken to Minimize Significant Economic Impact on Small Entities, and Significant Alternatives Considered. The RFA requires agencies to describe any significant alternatives that it has considered in reaching its proposed approach, which may include the following four alternatives: (1) the establishment of differing compliance or reporting requirements or timetables that take into account the resources available to small entities; (2) the clarification, consolidation, or simplification of compliance or reporting requirements under the rule for small entities; (3) the use of performance rather than design standards; and (4) an exemption from coverage of the rule, or any part thereof, for small entities. In the instant proceeding we are seeking comment on the prescription of the productivity offset, or X-factor, portion of the price cap formula. Therefore, only the first and last possible alternatives listed in section 603(c) of the RFA would be applicable. In Section V of the Notice, we seek comment on how the prescription of the X-factor would affect smaller price cap LECs differently from other price cap LECs, and whether there should be a separate X-factor calculated for smaller price cap LECs. We also do not believe it would be appropriate to exempt small price cap LECs from the application of an X-factor. We seek comment on these issues and urge commenting parties to support their comments with specific evidence and analysis. 26. Federal Rules that May Duplicate, Overlap, or Conflict With the Proposed Rules. None. C. Filing of Comments and Reply Comments 27. Pursuant to Sections 1.415 and 1.419 of the Commission's rules, 47 C.F.R.  1.415, 1.419, interested parties may file comments on or before December 30, 1999 and reply comments on or before January 14, 2000. Comments may be filed using the Commission's Electronic Comment Filing System ("ECFS") or by filing paper copies. 28. Comments filed through the ECFS can be sent as an electronic file via the Internet to . In completing the transmittal screen, commenters should include their full name, Postal Service mailing address, and the applicable docket or rulemaking number. Parties may also submit an electronic comment by Internet e-mail. To get filing instructions for e-mail comments, commenters should send an e-mail to ecfs@fcc.gov, and should include the following words in the body of the message, "get form ." A sample form and directions will be sent in reply. Only one copy of electronically-filed comments must be submitted. 29. Parties who choose to file by paper must file an original and four copies of each filing. All filings must be sent to the Commission's Secretary, Magalie Roman Salas, Office of the Secretary, Federal Communications Commission, 445 12th Street, S.W., Room TW-B204, Washington, D.C. 20554. 30. Parties who choose to file by paper should also submit their comments on diskette. The diskette should be submitted to: Wanda Harris, Federal Communications Commission, Common Carrier Bureau, Competitive Pricing Division, 445 12th Street, S.W., Fifth Floor, Washington, D.C. 20554. The submission should be on a 3.5 inch diskette formatted in an IBM compatible format using WordPerfect 5.1 for Windows or compatible software. The diskette should be accompanied by a cover letter and should be submitted in "read only" mode. The diskette should be clearly labeled with the commenter's name, proceeding (including the docket number in this case), type of pleading (comments or reply comments), date of submission, and the name of the electronic file on the diskette. The label should also include the following phrase: "Disk Copy - Not an Original." Each diskette should contain only one party's pleadings, preferably in a single electronic file. In addition, commenters must send diskette copies to the Commission's copy contractor, International Transcription Service, Inc., 1231 20th Street, N.W., Washington, D.C. 20036. Comments and reply comments will be available for public inspection during regular business hours in the FCC Reference Center, 445 12th Street, S.W., Room CY-A257, Washington, D.C. 20554. .VII. ORDERING CLAUSES 31. Accordingly, IT IS ORDERED that, pursuant to the authority contained in Sections 1, 4(i), 4(j), 201-205, and 303(r) of the Communications Act of 1934, as amended, 47 U.S.C.  151, 154(i), (j), 201-205, and 303(r), NOTICE IS HEREBY GIVEN of the rulemaking described above and that COMMENT IS SOUGHT on those issues. 32. IT IS FURTHER ORDERED that the Commission's Office of Public Affairs, Reference Operations Division, SHALL SEND a copy of this Further Notice of Proposed Rulemaking, including the Initial Regulatory Flexibility Analysis, to the Chief Counsel for Advocacy of the Small Business Administration. FEDERAL COMMUNICATIONS COMMISSION Magalie Roman Salas Secretary Appendix A Sources of Potential Errors in the 1997 Staff TFP Study Noel D. Uri A review of the 1997 Staff TFP Study, upon which the Commission relied in its 1997 Price Cap Review Order, has identified a number of potential errors that appear to affect significantly the results of the study. Below we discuss a possible conceptual error and a method of correcting the problem. We then discuss three sources of potential bias that are attributable to the study's sources of data and how these potential biases can be corrected. Finally, we discuss a number of other sources of possible bias. These possible biases either are not easily quantified, or do not have a significant impact on the level of the X- factor. I. Competitively Determined Historical Cost of Capital The 1997 Staff TFP Study used actual imputed cost of capital when measuring the productivity of regulated companies. Several parties allege that this creates a problem, because of the use of the residual value approach for determining the cost of the capital input. With this approach, the Commission in the 1997 Price Cap Review Order first directly determined the price and quantity of every input except capital, i.e., labor and materials. That is, it directly calculated the realized return to every non-capital input. Next, based on an estimate of the capital stock, the assumption that all revenue is distributed to the inputs, and given the returns to labor and materials based on historical data, a cost of capital is imputed. Conceptually, the difference (residual) between revenue and the required returns to all non-capital inputs (which is just the nominal amount used to impute a cost of capital) consists of two parts. The first part is the required return to capital. The second part is the excess profit earned by the firm. Instead of attempting to separate this difference into two parts, the Commission in the 1997 Price Cap Review Order assumed that all of this residual was the required return to capital, i.e., that no excess profit was earned. This is a reasonable assumption for a competitive market, and even for non-competitive markets, as long as the goal of the study is simply to measure the productivity gains revealed by market forces. In a regulatory setting, however, the productivity gains are "revealed" by the application of the X-factor, not by market forces. By attributing all of the residual to the capital inputs, the residual value method tends automatically to define whatever profits or losses the LECs realized during the historical period as increases or decreases in the cost of capital inputs. Critics contend that if, for example, the Commission chose too low an X-factor during the historical period being considered, causing the LECs' profits to increase, the residual value method would indicate that the historical cost of LEC capital inputs rose more rapidly during this period than it actually did. The Commission then would predict an equally large growth in LEC capital cost in the future, and thus calculate an X-factor that was still too low. Consequently, critics contend that LECs' profits would continue to increase despite no increase in LEC productivity. On the other hand, suppose that the Commission chose too high an X-factor during the historical period, causing LECs' profits to decrease. Critics contend that the residual value method would indicate that the historical cost of LEC capital inputs fell more rapidly during this period than it actually did. The Commission then would predict an equally large price reduction for capital in the future, and thus calculate an X-factor that was still too high. Consequently, LECs' profits would continue to decrease despite no decrease in LEC productivity. In either scenario, critics claim, the LECs' productivity remains constant; only their rate of return is affected by the incorrect X-factor. Thus, they assert that the residual value method produces an X-factor that is biased in the same direction as whatever bias existed in the historical period. Accordingly, the critics' theory posits that, to avoid this problem, any TFP study of a regulated industry should augment or reduce the actual amounts of capital compensation in such a way as to duplicate the forces of a competitive market. USTA proposed an index similar to the one we are using. What USTA did not do is reduce the revenue as capital compensation went down. To capture the gains in productivity that would have been revealed in a competitive marketplace, total capital compensation must vary with the competitive capital compensation rate. In order to correct the alleged miscalculation of the LECs' cost of capital in the 1997 Staff TFP study, it is necessary to replace the TFP study's cost of capital with a competitive cost for the inputs during the historical years. Next, it is necessary to adopt a surrogate to emulate a competitive cost of capital for LECs because LECs have never operated in a competitive market. This study employs an independent price series to compute the annual change in the cost of capital for a competitive market. Specifically, Moody's Baa corporate bond rate reported in the 1999 Economic Report of the President (Table B-73) is used to calculate the adjustment. Combining the base year imputed cost of capital from the 1997 staff TFP study with the change in the competitive cost of capital gives an independent competitive cost of capital for LECs in each year of the historical period. This competitive cost of LECs' capital input is used in conjunction with an adjusted cost of labor (see below) and materials inputs from the 1997 staff TFP study, as well as the percent change in the U.S. nonfarm business sector input price growth, to compute the corrected input price differential portion of the historical X-factor. Recalculating the LECs' historical cost of capital changes the level of the LECs' revenues, taxes, and operating expenses for the historical years. If we take the difference between our estimate of the competitive cost of LECs' capital inputs and the imputed cost from the 1997 staff TFP study, we obtain an estimate of the excess profits realized by LECs for each year of the historical period. Reducing the LECs' total revenue by this amount results in a revenue stream that emulates what LECs would have realized in a competitive market. A reduction in revenue implies a reduction in taxes, which in turn reduces operating expenses. Reducing LEC revenues to what would have been realized in a competitive market also changes the LECs' TFP portion of the X-factor. The new revenue level changes the LECs' input growth rate by altering the relative share of revenue accounted for by each input: labor, materials, and capital. In other words, the LECs' input growth rate changes in response to the changes in the relative weight of each input factor. Combining the output growth rate from the 1997 staff TFP study with the revised LECs' input growth rate yields the corrected TFP growth rate for the LECs. II. Other Modifications to the 1997 Staff TFP Study In addition to updating the data for the period 1996-1998, the study concludes that a number of adjustments to the study reasonably could be performed to improve the 1997 Staff TFP Study's accuracy. First, we could substitute the recently revised BLS series on multifactor productivity for the series previously employed. In addition, our recalculation of the X-factor could adjust for potential biases in the local service output index and the price of labor. These adjustments are easily quantified and are likely to have a significant impact on the level of the recalculated X-factor. The potential bias in the LECs' local output index would affect the LEC's TFP. The local service output index from the 1997 staff TFP study was based on the number of local calls. Because the length of calls can vary significantly, however, basing the LECs' local output index on the number of dial equipment minutes (DEMs) more accurately measures the use of the LECs' local equipment. For much of the period beginning in 1985, the average length of a local call measured by the number of local DEMs divided by the number of local calls was approximately constant. Beginning in 1992, however, there is a pronounced upward trend in local call length. This trend is attributable in large part to the increase in the use of the Internet by local subscribers. This increase has important implications for measuring local output and, in turn, LECs' TFP. Using the number of local calls as the output measure, output increased between 1992 and 1997 at a 3.5 percent annual rate. On the other hand, using the number of DEMs as the local output measure, output increased at a 6.5 percent annual rate. Because the data are not yet available, it is necessary to forecast two 1998 values used in computing the X-factor - the number of local dial equipment minutes and the multifactor productivity. Local dial equipment minutes for 1998 are forecast using an extrapolation of the average rate of growth in dial equipment minutes beginning in 1992. The multifactor productivity value for 1998 was forecast using a structural model relating the natural logarithm of the multifactor productivity to the natural logarithm of the output of all persons employed. This second series is reported by the Bureau of Labor Statistics, and its value for 1998 is available. The parameter estimates of the structural model in conjunction with the 1998 value for output per hour of all persons employed gives the forecast of 1998 multifactor productivity. Another adjustment we could make concerns the price of labor. This problem has further implications because labor's share of total factor payments will be inflated relative to the other factors of production. Between 1985 and 1998, the price of labor, defined as the average compensation per employee, increased in real terms at a 3.6 percent annual rate. In terms of total compensation, the amount paid for labor was approximately the same in 1985 as it was in 1998 even though the number of workers in the aggregate had been reduced by over 30 percent. The reason for this, at least in part, is that, coincident with the adoption of price cap plans, labor force reductions were accomplished by offering employees monetary incentives to leave the company (i.e., buyouts). There is some variability in the applicable accounting rules, but these payments were generally accrued as one-time charges against current earnings. This will tend to inflate costs in the early years after the implementation of price caps and, in turn, bias the measure of TFPLEC growth downward for these periods. The resulting recalculation of the X-factor, in which we use a competitively determined cost of capital, update the multifactor productivity series, substitute a better measure than the number of calls for local output, and adjust the price of labor to reduce the effects of buyouts, is presented in Table B-12 in Appendix B. While we have identified other potential sources of bias from the 1997 Price Cap Order in our proposed recalculation of the X-factor, the 1999 staff TFP study does not adjust for them because they either are not easily quantified, or do not have a significant impact on the level of the X-factor. These other possible sources of bias are discussed below. III. Other Possible Sources of Bias The productivity offset or X-factor is given as X(H) = (%_ TFPLEC(H) - %_ TFPUS(H)) + (%_ IPUS(H) - %_ IPLEC(H)) where X(H) denotes the historical X-factor, %_ TFPLEC(H) denotes the percent change in the historical total factor productivity of local exchange carriers between period t-1 and t, %_ TFPUS(H) denotes the percent change in the historical total factor productivity for the entire U.S. economy between period t-1 and t, %_ IPUS denotes the percent change in the historical input prices of all goods and services used to produce output of goods and services in the United States between period t-1 and t, and %_ IPLEC denotes the percent change in the historical input prices of local exchange carriers used to produce local service, intrastate toll/access service, and interstate service between period t-1 and t. It is important to note that there is no debate about whether there is appropriately an X-factor in the price cap equation nor is there a question about whether an input price differential should properly be included in the specification. The issue at hand is whether the components of the X-factor are accurately measured based on the data actually used in the computation. An assessment of this is the focus of what follows. A. Biases in the Measurement of TFPLEC There are a number of biases inherent in the measurement of TFPLEC. There are general biases and data-specific biases which will tend to overestimate or underestimate the true value of TFPLEC. i. General Biases First, TFPLEC will be biased since it is not compatible with incentives to be economically efficient because the disaggregated input and output data are under the control of the regulated LECs and, thus, are subject to manipulation. It has been noted that price cap regulation has the potential to introduce pure waste, inefficient factor utilization, excessive research and development expenses, and over-investment in demand increasing expenditures. This will bias the measurement of TFPLEC downward. The extent of this downward bias will be a function of the quantitative magnitude of any excessive expenditures and inefficient factor utilization. Another general bias introduced in the measurement of TFPLEC is associated with the use of LEC performance data immediately post-divestiture. This will give a downward bias to the measure. It is generally conceded that under rate-of-return (ROR) regulation, regulated firms employed too much capital and too much labor. Transition from the regulated environment is not instantaneous. The optimal level of factors of production will be in disequilibrium for some period of time following the demise of ROR regulation. Hence, the measured TFPLEC as reported in the 1997 Price Cap Review Order will contain some disequilibrium periods. That is, the measurement of the productivity of LECs is necessarily calculated, at least partially, in a ROR environment which will bias downward the TFP measure for at least a portion of the period. A final general bias introduced in the measurement of TFPLEC is associated with the demand for interstate access to local LECs' local loops. Interstate access service is the focus of price cap regulation. It has grown much more rapidly on average than demand for local service and intrastate access service. The data on this are clear. Thus, in the presence of economies of density, this leads to the conclusion that TFPLEC in interstate services has grown faster than company-wide (regulated) TFPLEC. More specifically, the price cap index (PCI) being considered here applies to interstate access service but TFPLEC is computed based on all LEC services. There is every reason to expect that productivity enhancements experienced historically in the interstate access market would be substantially greater than the overall rate of productivity growth experienced by LECs in supplying all services. First, most of the productivity growth experienced in the telecommunications industry is related to reductions in switching costs and to savings in transmissions costs which occur as a result of using electronics to expand the carrying capacity of transmissions facilities. In contrast productivity growth in supplying loop services has been relatively lower. As a result of this, the average measure of TFPLEC used in setting X and which should properly reflect productivity growth in the interstate access market is biased downward. ii. Data-Specific Biases a. Output Biases We turn now to specific issues involving the measurement of TFPLEC. It is generally recognized that output growth is a key determinant of the rate of growth of TFP. Thus, it is especially important to measure it as precisely as possible. Characterizing output is not a trivial undertaking. Output of telephone service includes both access and use. Access and use, however, are complicated by access and use externalities. The presence of these externalities means that, contrary to what is typically presumed, preferences are interdependent across subscribers. That is, as more users gain access to telephone, the utility of customers in the aggregate rises. These externalities are exceedingly difficult to quantify and are not measured here. This will bias downward the TFPLEC measure. Additionally, output comes in a variety of forms including type (station, person, collect, etc.), time-of-day, day of week, distance, and duration. In the specification used, total company output consists of three identifiable components: local service, intrastate toll/access service, and interstate service. Revenue shares are used to weight each component's contribution to total output. Each of the components in total company output is measured differently. This was dictated by the available data. It introduces a potential source of bias. The nature and extent of this bias, however, is indeterminate. That is, it is not clear whether there will be a positive or negative bias. This is an empirical issue. Since comparable measures are not available for each of the output components, it is not possible to measure it empirically. The local service output measure is based on the number of local calls. Intrastate service output index is based just on the number of dial equipment minutes (DEMs) while the interstate service output index is a function the number of access lines, the number of switched access minutes, and the number of special access lines. No measure of special access minutes is used. It is interesting to note that the interstate output index endeavors to capture both the access and use components of interstate output. Output measures for local service and intrastate service do not endeavor to do so. The component of the output measure that is probably the most questionable in the 1997 Price Cap Review Order is the local service output index. This, however, has been corrected in the updated study. b. Input Biases We next consider the input (factors of production) side of measuring TFPLEC. Before doing so, however, it is important to note that the cost of inputs (i.e., the prices paid for inputs) in the TFPLEC measurement also appear in the measurement of %_ IPLEC. Hence, observations made concerning measurement errors and their impact on the computed value of TFPLEC are equally applicable to the measurement of the percentage change in input prices of LECs. Furthermore, the nature of the relationship given by equation (1) means that most measurement errors associated with the prices of the inputs will tend to cancel out so that the impact on the productivity offset will, in general, be minimal. The most problematic input measure is that for the quantity of labor. Measurement of each input in a total factor productivity study typically requires subaggregation of various types of the factor. In the case of labor this would include several different occupational groups and skill levels and would be calibrated as the number of hours worked with appropriate adjustments for overtime and part-time workers. The labor quantity measure used in the 1997 Price Cap Review Order consists solely of the total number of full-time employees of RBOCs. Thus, inherent productivity differentials between employees based on such things as skill levels is not reflected. If the LECs were in equilibrium such that the relative proportion of employees in the different occupational groups with defined skill levels characteristic of LECs remained unchanged as did the average number of hours each employee worked, then there would be little concern over how the labor variable is measured. This is not the case, however. Between 1985 and 1998, there was been a significant change in the number of employees. The LECs labor force declined by 2.4 percent annually over this period. Moreover, the reduction has taken place so that the number of managers has been reduced disproportionately more than the number of technical employees. Thus, there is a disparity between the change in observed labor versus effective labor. The latter is what should be included in the measurement of TFPLEC while the former is what has been included. Correct measurement of the labor variable would reduce aggregate labor expenses and hence increase TFPLEC. Hence, as constructed, TFPLEC is biased downward due to the incorrect measurement of the number of units of labor. Materials quantity and materials expense are computed as residuals. Materials quantity is equal to materials expense divided by materials price. Materials expense is defined to equal total adjusted operating expense minus the sum of labor compensation, depreciation, and amortization expense. The increase in materials expense account for the major portion of the increase in total factor payment over the period 1985 to 1998. It is unclear whether there is a superior approach for measuring materials quantities and materials expense. The nature of any TFPLEC measurement bias is uncertain. The measurement of capital stock and the cost of capital was of greatest concern in the 1997 Price Cap Review Order. Accurate measurement of capital stock is exceedingly difficult. Different types of capital with different vintages and technology comprise the stock in any given period. This stock, moreover, is subject to different utilization rates. With these difficulties come two significant problems in the development of the measures of capital and the cost of capital that potentially serve to bias downward the reported measure of TFPLEC. The first involves the changing quality of capital over the historical period that is not adequately reflected in the computations. The second involves changes in capacity utilization that are not captured. The quantity of capital is measured for each depreciable asset by the perpetual inventory method from data on investment. Many types of capital inputs used by LECs have changed including computer chips, digital electronics, fiber optics, and digital switching equipment. This change in the quality of capital is not effectively reflected in the perpetual inventory approach adopted because the output associated with one unit of new capital brought on line is not necessarily equivalent to the output of one unit already in operation. If they are different, there has been a change in quality. The typical approach to reflecting this quality change is to adjust the cost of capital downward to avoid understating the change in the effective level of real capital stocks. Additionally, to the extent that succeeding generations of capital equipment are more productive, an appropriate adjustment to reflect this increases the computed level of capital stock, increases the flow of capital services, and, holding output constant, decreases the measured TFPLEC. Since there was no quality adjustment in the computations reported in the 1997 Price Cap Review Order, TFPLEC will be biased upward. The second significant omission is the failure to adjust for changes in capacity utilization. Over time as the existing capital stock is used more intensively, output is higher without a corresponding increase in capacity. Consequently, TFPLEC is greater. There is, however, no adjustment for a change in capacity utilization in the report. TFPLEC will, consequently, be biased downward. B. Biases in the Measurement of TFPUS The measure of TFPUS used is the Bureau of Labor Statistics' estimate of Nonfarm Business Sector Multifactor Productivity. There is concern that this measure underestimates the growth in TFPUS. There is, however, no basis for determining the magnitude of the underestimation although the consensus is that it is relatively small. This being the case, while the productivity offset will be overestimated, its size is small but indeterminate. The major issues related to the BLS measure of productivity that potentially give rise to overestimation of TFPUS include: (1) Mismeasurement of productivity growth in the services portion of the economy. This portion of the economy is defined to include communications. The problem centers on how to define output in the service-producing industries. In the case of the services provided by LECs, this can be thought of as measuring output in terms of access and use. There is an enormous literature on the mechanics of defining output in these service-producing industries although there is no consensus on how this should be done. Inadequate definitions of output underlie the published data for many service activities. (2) Biases in the Consumer Price Index (CPI) on productivity data. Components of the CPI are used to construct approximately 57 percent of the business sector output measure used for BLS productivity statistics. Thus, any biases in constructing the CPI are directly reflected in TFPUS. Biases in the CPI are well documented although their precise magnitude is subject to debate. These biases include incorrectly reflecting the substitution bias where consumers' tendencies to substitute among items or categories of items as relative prices change over time, failure to capture new-outlet bias where consumers respond to relative price changes by shopping at different retail outlets, and inadequately measuring quality changes and new products reflecting changes in technology and consumer demand. (3) Whether the BLS productivity data fully reflect changes in the quality of goods and services. Questions persist as to whether the total factor productivity measure for the United States economy reflect the extent that succeeding generations of capital equipment are more productive and hence whether an appropriate adjustment to reflect has been incorporated. Note that this will increase the computed level of capital stock and increase the flow of capital services. (4) Whether the best techniques are used to introduce new, advanced products into the data series. (5) Whether the BLS methods capture the full impact of new information technology on economic performance. It has been suggested, for example, that failure to adequately reflect information technology understates TFPUS. These issues are discussed fully in a series of articles in the February 1998 issue of the Monthly Labor Review. One other source of bias not previously noted is also present. This involves composition changes in inputs. In the case of labor composition, changes include increased educational attainment over time which increases TFPUS and an increased share of teenagers and females in the labor force which reduces TFPUS. In the case of capital, there has been continuous substitution of equipment for structures and of short-lived equipment like computers for long-lived equipment like furniture. Failure to reflect this substitution underestimates the capital stock component of the TFPUS measure. C. Biases in the Measurement of IPUS The general input price measure used is the Bureau of Labor Statistics Nonfarm Business Sector Price Index. The sources of the data used in constructing this measure are the same as those for the BLS measure of TFPUS. Consequently, the major issues related to the BLS measure of TFPUS are applicable to the measurement of IPUS. These include the question of how to define output in service-producing industries, the biases present in constructing the Consumer Price Index, the issue of the introduction of new, improved factors of production, and the effect of quality changes in goods and services used in production. While there are biases in the measure of IPUS due to biases in its components, considered together with any bias in the measure of TFPUS attributable to input prices would most likely have an insignificant net effect on the productivity offset. For example, an increase in the price of one of the factor inputs will lower TFPUS but it will increase the input price differential resulting in little net change in X. D. Biases in the Measurement of IPLEC The measure of IPLEC is a composite index using the labor price index, a materials price index, and a cost of capital index. The biases introduced were previously noted in the context of measuring TFPLEC including changes in the quality of capital and capacity utilization variations. The effective price of labor is probably overestimated while the cost of capital even after adjustment to reflect competitive conditions might or might not be correct. An adjustment in the cost of capital to reflect changes in quality would increase the price while an adjustment to reflect changes in capacity utilization would reduce the price. As noted previously, there are obvious upward biases in the measure of IPLEC due to biases in its components. This, however, considered in conjunction with the bias in the measure of TFPLEC attributable to these input prices results in a minimal net effect on the productivity offset (X-factor). That is, increasing (decreasing) the price of one of the factor inputs will lower (raise) TFPLEC but it will reduce (raise) the input price differential resulting in little net change in X. IV. Conclusion Based on the above analysis, the X-factor calculated in the 1997 Staff TFP Study is a significantly downward biased estimator of the actual rate of cost reductions achieved by the price cap LECs. The residual value method employed in this study inappropriately incorporated the steadily rising profits of LECs into increases in the cost of the capital input. Correction of the LECs' historical cost of capital plus the change in the local output measure results in a somewhat larger absolute value of the X-factor than the Commission computed in the 1997 staff TFP study. Over the 1986-1995 period, for example, we compute the average X-factor to be 5.82 versus an average of 5.23 from the 1997 staff TFP study. Between 1991 and 1995, when price caps were in effect, our corrected average X-factor is 6.14 versus an average of 5.22 from the 1997 staff TFP study. Also, there is somewhat more year-to-year variation in our corrected X-factor than there is in the 1997 staff TFP study. Using data covering the period 1986 to 1998, the updated X-factor is 6.02 while for the 1991 to 1998 period, the X-factor is 6.33. Appendix B The 1999 Staff TFP Study Noel D. Uri Introduction Under price cap regulation, the weighted average of the prices for the services in a given price cap, or the actual price index (API), must be less than or equal to the price cap index. An incumbent LEC's Price Cap Index (PCI) is adjusted annually in accordance with the PCI relationship defined in the Code of Federal Regulations. The PCI relationship consists of a measure of inflation, in this case the Gross Domestic Product Price Index (GDP-PI), minus the X-factor, plus or minus any permitted exogenous cost changes. The X-factor is nominally referred to as the productivity offset. The X-factor is defined to equal the sum of the change in LECs' productivity less the change in productivity of the aggregate economy plus the change in input prices for the aggregate economy less the change in LECs' input prices. In the 1997 Price Cap Review Order, a total factor productivity (TFP) approach was adopted for estimating LECs change in productivity. Considerable thought has gone into and extensive comments were made concerning the selection of a technique to calibrate the X-factor. This decision process will not be revisited. Rather, the immediate concern is with measuring the X-factor based on the most recent, consistent, and comprehensive data generally and publicly available. Background The methodology used by the FCC's staff to estimate LEC Total Factor Productivity (TFP) and the input prices, and to calculate the LEC TFP and input price differentials used in the FCC's LEC X-factor for interstate access is discussed here. TFP is calculated based on the LECs regulated books of account excluding miscellaneous services. Thus, the measure of total factor productivity is based on the productivity of all LEC activities including local service, intrastate toll/access service, and interstate service. The calculations are for the period 1985 through 1998. The calculations are based on the Fisher Ideal Index, a standard and economically justifiable measure. The TFP estimates embody what is believed to be the best approach available for measuring the X-factor and emulates the approach used in 1997 Price Cap Review Order. The study is based on data publicly available from the Federal Communications Commission, the Bureau of Economic Analysis of the U.S. Department of Commerce, and the Bureau of Labor Statistics of the U.S. Department of Labor. Consistent with the 1997 Price Cap Review Order, the data are for just the Regional Bell Operating Companies (RBOCs). Methodological Issues A productivity index is generally defined as an output index divided by an input index. An important question is which functional form for the index number should be used. Here, known functional forms for index numbers are related to functional forms for the underlying aggregator function. Two important examples of a quantity index are the Laspeyres and Paasche index. The Laspeyres quantity index, QL, is defined as (1) QL (p0, p1, x0, x1) = p0 x1 / p0 x0 where pt > 0 is a vector of prices in period t = 0, 1 and x > 0 is the corresponding vector of quantities. The Paasche quantity index, QP, is defined as (2) QP (p0, p1, x0, x1) = p1 x1 / p1 x0. Fisher suggested that the Ideal Index is given as the geometric mean of QL and QP (3) QF (p0, p1, x0, x1) = ((p0 x1 / p0 x0) (p1 x1 / p1 x0))1/2. Let f(x) define an aggregator function and suppose that xt > 0 is the solution to (4) maxx {f(x): pt x £ pt xt } t=0, 1 where f(x) = (x'Ax)1/2 and A = [aij] is a symmetric matrix of coefficients. Konus then shows that (5) f(x1) / f(x0) = QF (p0, p1, x0, x1). If a quantity index QF (p0, p1, x0, x1) and a functional form aggregator function f satisfy (5), then the quantity index QF is said to be exact for the aggregator function. Konus shows that the Laspeyres quantity index is exact for a fixed aggregator function while the Paasche quantity index is exact for a linear aggregator function. Afriat, Diewert, Pollack, and Samuelson and Swamy provide other examples of exact index numbers. Among exact index numbers, however, the Fisher quantity index corresponds to a functional form for f that can provide a second order approximation to an arbitrary twice-differentiable linearly homogeneous function. Thus, the use of the Fisher Ideal Index is economically justified. The input and output quantity indices are constructed using the Fisher Ideal Index. As noted, this index is the geometric average of the Laspeyres Index and the Paasche Index. This index is desirable because it is calculated using the weights of adjacent years. During periods of relatively substantial and significant changes in prices, a significant bias can appear in fixed weight measures, even during periods close to the base period. Expanding beyond just two periods, a chained Fisher Ideal Quantity Index can be constructed between periods 0 and t. It is the product of each of the Fisher Ideal Quantity Indexes between periods 0 and t. All input and output quantity indexes are chained Fisher Ideal Quantity Indexes. The chained Fisher Ideal Quantity Index addresses one of the most fundamental problems in measuring output - the choice of the base period with which all other periods are compared. Since changes in the Fisher Ideal Quantity Index are calculated using weights of adjacent years, the chaining of the annual changes allows for the effect of changes in relative prices. Thus, the Fisher Ideal Chained Quantity Index calculates an index that is appropriate for each period and avoids having to update a fixed-weight index. It also negates the substitution bias that is inherent in a fixed-weight index. Finally, the chain-type index provides a more accurate measure of current period output during periods of significant price changes. Indexes of input prices are constructed in a fashion analogous to that for the input and output quantity indexes. That is, by calculating a Fisher Ideal Price Index. The Laspeyres price index, PL, is defined as (6) PL (p0, p1, x0, x1) = p1 x0 / p0 x0 where pt > 0 is a vector of prices in period t = 0,1 and x > 0 is the corresponding vector of quantities. The Paasche price index, PP, is defined as (7) PP (p0, p1, x0, x1) = p1 x1 / p0 x1. The Fisher Ideal Index for prices is given as the geometric mean of PL and PP (8) PF (p0, p1, x0, x1) = ((p1 x0 / p0 x0) (p1 x1 / p0 x1))1/2 . A chained Fisher Ideal Price Index can be constructed between periods 0 and t. It is the product of each of the Fisher Ideal Price Indexes between periods 0 and t. All input price indexes are chained Fisher Ideal Price Indexes. As is the case for output and input indexes, a chained price index is superior to a fixed-weight index. Aggregating across different types of outputs to construct the Fisher Ideal Quantity Index for output in a given period is accomplished by using the relevant commodity share of total revenue corresponding the type of output. Aggregating across the different factors of production to construct the input quantity and input price indexes in a given period is accomplished by using the relevant shares of total payments to the corresponding factors. Calculation of Output Quantity Indexes (a) Data Sources The output index is based on actual quantity measures available from the Commission. Basic local service revenue, end user revenue, switched access revenue, special access revenue, state access revenue, and total long distance network revenue are taken from the Commission's Statistics of Communications Common Carriers for 1985 through 1998. Also taken from this source are data on the number of special access lines, the number of business access lines, residential access lines, and public access lines. Local access volume is measured by the number of local dial equipment minutes as are state toll and intrastate access volumes by dial equipment minutes. These data are taken from the FCC Monitoring Reports. They are directly accessible via the FCC web site. Interstate switched access minutes are from the same Monitoring Reports. The component data for interstate revenue are given in are given in Table B-1. Total revenue data for each type of service are given in Table B-2. (b) Output Category Quantity Indices and Revenue Shares The initial step in the process is to construct an interstate quantity index to measure growth of interstate services. This index is constructed using the physical measures of three services including the number of access lines, the number of interstate switched access minutes, and the number interstate special access lines. The number of access lines is measured by the sum of the number of business, public, and residential access lines. A service's share of total interstate revenue is used to weigh each service in the construction of the measure of interstate output. That is, the number of access lines is weighted by the End User Common Line revenue share of total interstate revenues. The number of switched access minutes is weighted by the switched access revenue share, and the number of special access lines is weighted by the special access revenue share. A Fisher Ideal Quantity Index is then constructed. The composite Fisher Ideal Interstate Quantity (Output) Index is derived by chaining the Fisher Interstate Ideal Output Index. The resulting Fisher Ideal Chained Output Index is given in Table B-3. A comparable procedure is used to construct revenue shares and quantity indices for total local service and state toll/access service. State toll/access revenues are total toll service revenues plus intrastate access revenues. The physical units associated with total local service are the number of dial equipment minutes for local calls while for state toll/access service, the physical units are intrastate dial equipment minutes. These data are taken from the FCC Monitoring Reports. (c) Total Output Index The total LEC output index is computed using the quantity indexes for local service, intrastate toll/access service, and interstate service and their respective revenue shares. The interstate share of total revenue is calculated using the sum of end user revenue, switched access revenue (formerly called "carrier's carrier facilities revenues"), and interstate special access revenue. A Fisher Ideal Quantity (Output) Index is constructed by aggregating these measures for each year. The composite Fisher Ideal Output Index is derived by chaining the Fisher Ideal Output Index. The resulting Fisher Ideal Chained Output Index for total LEC output is given in Table B-4. Note that the percentage changes (growth rate) are calculated as natural logarithmic (loge) changes. Calculation of Input Quantity Indexes (a) Labor The measure of the quantity labor is based on annual accounting data for the number of employees from the Commission's Statistics of Communications Common Carriers. These data are reported in Table B-5. Since there is no objective way to account for the contribution of part-time versus full-time employees, just the total number of employees is used as the labor input measure. This, however, does not introduce a substantial bias in the labor quantity measure since part-time employees accounted for less that 0.7 percent of the workforce in 1998. (b) Capital (i) Perpetual Inventory Method The starting point for construction of a quantity index of capital input is the measurement of the capital stock. A perpetual inventory method is employed to estimate the level of capital stock. In discrete time, the method is (9) Kt = It + (1 - d) Kt-1 where Kt is the end-of-period capital stock, It is the quantity of investment occurring in the period, and d is the rate of replacement (depreciation). The data requirements for implementation of this perpetual inventory method are investment in constant (inflation-adjusted) prices, a capital benchmark, and a rate of replacement. The perpetual inventory method is used to remove embedded inflation that would distort the measurement of capital. Consistent with the 1997 Price Cap Review Order, only one asset class is considered because the record shows that the number of asset classes does not significantly impact the measured change in TFP. The application of the perpetual inventory model relies on Commission depreciation rates. The book value of plant is used as the basis for calculating the benchmark (i.e. initial level) capital stock. In order to calculate constant dollar investment, chained Fisher asset prices from the Bureau of Economic Analysis of the U.S. Department of Commerce are used deflate capital additions. The benchmark capital stock is based on the end of year 1985 book value. Because of the 1988 capital/expense shift, it is necessary to adjust both end of year 1985 total plant in service less accumulated depreciation and 1985-1987 capital additions. The capital/expense shift factor used in the 1997 Price Cap Review Order is employed here to reduce capital additions for 1985 through 1987. Adjusted capital additions for 1985, 1986, and 1987 are just the product of the unadjusted capital additions and the adjustment factor of 0.888. Unadjusted capital addition data were obtained from FCC Form M. (ii) Asset Price With just a single asset class, a single composite asset price index is needed. First, Bureau of Economic Analysis asset prices were obtained from the National Income and Product Accounts. These included prices for three asset categories including Communications Equipment (NIPA Table 7.8: Chained-Type Price indexes for Private Purchases of Producers' Durable Equipment by Type, Line 7), Telecommunication Structures (NIPA Table 7.7: Chained-Type Price Indexes for Private Purchases of Structure by Type, Line 12), and a composite asset price for Producer Durables (NIPA Table 7.1, Line 39). Capital additions data are then grouped into categories corresponding to the NIPA asset categories, and each category's share calculated. (The capital/expense shift adjustment factor noted previously has no effect on the shares because it is multiplicative and applies equally to all asset categories.) From these data, a Fisher Ideal Price Index is computed and used to form a chained Fisher Ideal Price Index for the single asset. This price index is used to deflate adjusted capital additions in the perpetual inventory model. (iii) Benchmark Capital Stock The benchmark capital stock is derived using the FCC accounting relationship (10) TPIS.BOYt + CAt - Retirest = TPIS.EOYt where TPIS.BOYt denotes beginning of the year total plant-in-service for period t, CAt denotes capital additions in period t, Retirest denotes plant retirements in period t, and TPIS.EOYt denotes total plant-in- service at the end of the year for period t. All of these data are taken from the FCC Form M. Adjusted capital additions are incorporated which results in revised values for TPIS.EOYt for 1985, 1986, and 1987. The benchmark capital stock is then obtained by subtracting accumulated depreciation from adjusted TPIS.BOY1985. As is standard practice in most TFP studies, land is not included as a capital stock component when forming the benchmark capital stock. (iv) Depreciation Rate A time-invariant depreciation rate for the single asset class is computed as an arithmetic average of annual depreciation rates. The depreciation rate for period t is given as (11) dt = (DEPR.ACRLSt) / ((TPIS.BOYt + TPIS.EOYt)/2). where DEPR.ACRLSt denotes depreciation accruals. The data are taken from FCC Form M. The value of d is the average (mean) value of dt over the period 1985-1998. A summary of capital adjustments and average depreciation is given in Table B-6. (v) Capital Input Quantity Index The capital input quantity index series is computed by dividing current period capital stock by the base period capital stock. These data are reported in Table B-7. (c) Materials - Part I The materials quantity is computed as materials expense divided by a materials price index. Materials expense is a residual. It is the difference between total operating expense and the sum of labor compensation and depreciation and amortization expense. Until now computation of the various components of the X-factor have been straightforward and mechanistic. Things are somewhat more complicated, however, from this point forward. To explain the complications and what is necessary to compute materials quantity, a digression is in order. (d) Digression TFP models in general measure productivity as the ratio of an index of outputs of a firm to an index of its inputs over a given time period (e.g., a year). The growth in productivity is the amount by which this ratio changes over time. In making these calculations, it is critical that real changes in productivity be separated from the effects of changes in input prices. It is typically the case that three inputs are considered in TFP-type studies. These inputs consist of labor, capital, and materials. Sometimes a fourth input energy is considered. This is not done here. When just three inputs are used in the analysis, materials is considered as a residual. That is, it accounts for whatever input expenditures are not attributable to labor and capital. The growth rate of the index of inputs is determined by the growth rates of the labor, capital, and materials input indices weighted by their relative contribution to total input cost. The growth rate of capital is a complex procedure, requiring a determination of capital stock and the flow of capital services from capital stock. A number of very explicit assumptions underlie TFP analyses. In particular, the conventional method of measuring TFP of firms (which is used in the 1997 Price Cap Review Order) is to assume that there are constant returns to scale and that observed inputs and outputs have been generated by firms in competitive, long run equilibrium. With prices of output and inputs fixed, the firm chooses input levels so as to maximize profit. The measure of TFP growth obtained by conventional means is not, however, appropriate whenever firms are not in a long run cost minimizing equilibrium. A firm is not in long run equilibrium whenever the firm's input-output bundle is other than corresponding to a point on the long run unit cost curve. If the firm is not in long run equilibrium, then profit is not zero. Before exploring the implications of this observation for the results reported in the 1997 Price Cap Review Order, it is necessary to examine what price caps are and to assess the mechanics of estimating an historical productivity offset. (i) Price Caps in Theory Overview Assume that there are two periods, t= 1,2. Suppose that in each period the firm produces units of a single output, qt, using a single input which will be called labor and denoted by Lt. Let wt denote the price of labor in period t and pt denote the price of output in period t. An outcome for the market in period t is defined to be a vector of these four variables, (pt, qt, wt, Lt). If the output price of the firm is regulated so that its revenues equal its costs, then the outcome in each period will satisfy (12) pt = wt Lt /qt . Define the variable zt to be the productivity of the firm in period t measured in units of output per units of input. Thus, (13) zt = qt /wt . Substitution of relationship (13) into relationship (12) yields (14) pt = wt /zt . That is, in a regulated market where the firm is constrained to earn zero economic profit, the output price will be equal to the input price divided by productivity. Use the following notation to denote the percentage change in each variable between periods 1 and 2: (15) %D p = (p2 - p1)/p1 (16) %D q = (q2 - q1)/q1 (17) %D pw = (w2 - w1)/w1 (18) %D L = (L2 - L1)/L1 (19) %D z = (z2 - z1)/z1 It follows immediately from equation (14) that when outcomes in both periods are regulated so that the firm earns zero economic profit, then (20) %D p = %D w - %D z. That is, if the firm is constrained to earn zero economic profit, then the percentage change in output price will just equal the percentage change in input price minus the percentage change in productivity. Price Caps With an Exogenous Input Price Index The well-understood problem with cost-based regulation is that the regulated firm has very little incentive to be efficient because any efficiency gains are immediately captured by the regulator. Price caps are meant to ameliorate this problem. The basic idea is very simple. Suppose that period 1 has just ended and the regulator is considering how to set price in period 2. The regulator knows that if he or she uses cost-based regulation to set price in period 2 that the resulting price will satisfy relationship (9). The essential idea of price caps is to replace the term %D z with an estimate of %D z made in period 1. Let E denote the estimate of productivity change. At the end of period 1, it is announced that the price in period 2 will be set according to the formula (21) %D p = %D w - E. That is, the regulator will wait until the end of period 1 and observe the actual change in input prices and allow output prices to increase by this percentage. Rather than wait to observe actual changes in productivity, however, the output prices are adjusted by a fixed amount regardless of the firm's realized productivity change. The firm will then earn a normal profit (and zero economic profit) if the regulator is able to correctly guess the actual (realized) productivity change. The firm will have a strong incentive, however, to improve productivity because it will keep any positive or negative profits that result if its actual productivity improvement is different than the predicted productivity change. One could also imagine the regulator replacing %D w with an exogenous, pre-specified estimate. Then the regulator would simply tell the regulated firm that its output prices would change by a certain percent regardless of the level of actual input price inflation in period 2. This practice is not usually followed, however, because the regulated firm typically purchases inputs in relatively competitive markets so that the regulated firm has very little influence over the price it pays for inputs. Therefore, exogenously pre-specifying a target percentage change for input prices would have no desirable incentive effects. It would increase the risk, however, that the regulated firm is required to incur to the extent that future input prices vary unpredictably. Therefore, %D w is not typically replaced by a pre-specified estimate. Price Caps With An Exogenous National Output Price Index Price cap schemes generally have the form that output price is allowed to increase in a given year by the amount that some price index actually increases minus some pre-specified amount. The foregoing section suggests that a very simple, transparent, and straightforward method for implementing a price cap system would be to base the price index on INPUT prices for the industry being regulated. This practice, however, is not typically followed. Rather, instead of basing the price index on INPUT prices for the industry being regulated, regulators generally use a composite price index for OUTPUT prices for the economy as a whole. In particular, this is the practice followed by the Federal Communications Commission (hereafter Commission) to regulate interstate access prices. When a national output index is substituted for an industry input index, a correction has to be made for the fact that input prices in the particular industry being regulated may be systematically growing at a different rate than output prices in the economy as a whole. For example, because of rapid technological advances in telecommunications, input prices for telecommunications firms are not changing at the same rate as output prices in the economy. Consequently, the regulator must choose an adjustment factor to correct for this difference. Let D denote the regulator's estimate of the rate at which output prices in the economy as a whole change differently than input prices in the regulated industry. Let RPI denote the percentage change in the national output price index. The price cap formula becomes (22) %D p = RPI - D - E . This can be rewritten as (23) %D p = RPI - X where (24) X = D + E. This is the familiar form of a price cap plan. That is, the X-factor is composed of two separate components. It is the sum of the predicted productivity growth of the regulated firm, E, plus the expected amount that national output prices grow faster than industry input prices, D. Choosing the X-Factor The efficiency incentives created by a price cap scheme are not affected by the size of the X-factor that the regulator chooses. The simple fact that the X-factor is fixed and independent of the actual costs incurred creates an incentive for the firm to be efficient. The regulator is still concerned with the level of the X-factor, however, because this affects the distribution of the economic surplus between the firm and consumers. It can also potentially affect whether or not the regulated firm chooses to exit the industry. If the X-factor is chosen too high, the regulated firm may well choose to exit the industry. On the other hand, if it is chosen too low, the regulated firm will earn positive economic profit and consumers will pay higher prices than necessary. Therefore, the regulator has a considerable interest in choosing an X-factor so that it will be above but as close as possible to the sum of the expected value of the firm's productivity change and the difference between the industry's input price change and the economy's output price change. One relevant piece of information is the behavior of these variables in the recent past. Let EH denote the historical rate of productivity growth for the regulated firm over the recent past. Let DH denote the historical difference between the rate of change in output prices for the economy as a whole and the rate of change of input prices for the regulated industry. Let XH denote the sum of these two variables. That is, (25) XH = DH + EH. The variable XH will be referred to as the historically justified X-factor. The typical practice of regulators is to calculate the historically justified X-factor and then to set the X-factor equal to this unless factors can be identified which will cause the future values to depart systematically from the historical values. If so, some attempt must be made to adjust exogenously for such factors. These identifiable factors are customarily most important when price caps are being first introduced to replace rate-of-return regulation. In such a circumstance, it is generally argued that the increased efficiency incentives created by the switch to price caps will result in higher future rates of productivity growth than were historically observed. Therefore, it is argued that future productivity growth is likely to be higher than past productivity growth for some period of time. Therefore, when price caps are first introduced, regulators often set the X-factor somewhat higher than the historically justified rate in anticipation of increased production efficiency. The Commission followed this practice when it first introduced price caps for LECs in 1991. The actual X-factor was set 0.5 percent above the historically justified level and this difference was nominally referred to as the consumer productivity dividend. Note, however, that increased efficiency incentives created by the switch to price caps will dissipate when price caps have been in effect for some time and the historical period being evaluated is one where price caps were in place. In this case, the historical data are already reflecting the increased incentives for efficiency created by price caps. In theory it is possible to imagine all sorts of reasons why future rates of productivity growth may diverge from past rates, even if price caps have been in place for some time. For example, one of the main sources of measured productivity growth is simply growth in demand, since production technologies of regulated firms frequently exhibit economies of scale. Therefore, if the estimated rate of demand growth in the future is different than the historical rate of demand growth, this would create a basis for predicting a difference between historical and future rates of productivity growth. In practice, however, regulators' information is often much too unreliable to attempt to make such corrections. Consequently, in many regulated industries it may be that the only course of action on the part of the regulator than can withstand a critical challenge is for the regulator to set the future X-factor equal to the historical rate. There is some evidence that the Commission is in fact in this position. In the 1997 Price Cap Review Order, the Commission calculated that the historically justified X-factor was approximately 5.2 percent but then attempted to argue that the future X-factor should be set at 6.0 percent. The U.S. Court of Appeals for the D.C. Circuit recently remanded this decision to the Commission, stating that the Commission had not adequately justified the additional amount to the historically justified rate. (i) Price Caps in Theory and Practice The Commission's description of its methodology for calculating the historically justified X-factor in the 1997 Price Cap Review Order appears to be different than the procedure for estimating the DH and EH described above. This difference, however, is only superficial. The procedure used by the Commission to estimate XH is, in fact, precisely as described above. This is explained in what follows. In the description of its procedure in the 1997 Price Cap Review Order, the Commission states that it calculates the historically justified X-factor according to the following formula: (26) X(H) = (%D TFPLEC(H) - %D TFPUS(H)) + (%D IPUS(H) - %D IPLEC(H)) where X(H) denotes the historical X-factor, %D TFPLEC(H) denotes the percent change in the historical total factor productivity of local exchange carriers between period t-1 and t, %D TFPUS(H) denotes the percent change in the historical total factor productivity for the entire U.S. economy between period t-1 and t, %D IPUS denotes the percent change in the historical input prices of all goods and services used to produce output of goods and services in the United States between period t-1 and t, and %D IPLEC denotes the percent change in the historical input prices of local exchange carriers used to produce local service, intrastate toll/access service, and interstate service between period t-1 and t. On the surface, relationship (26) does not appear to be the same as equation (25) which the previous analysis suggests is what is used to calculate the historically justified X-factor. Equation (26) can be rewritten, however, as (27) X(H) = {%D TFPLEC(H)} + {- %D TFPUS(H)) + %D IPUS(H) - %D IPLEC(H)} . The term in the first set of brackets in equation (27) corresponds to the variable EH in equation (25). Therefore, equations (27) and (25) are in fact equivalent if it can be shown that the terms in the second set of bracket of relationship (27) correspond to DH. This equivalence is demonstrated in what follows. Let RPI(H) denote the rate of change in output prices in the economy as a whole over the historical period. Then, it is always true that (28) RPI(H) = - %D TFPUS(H)) + %D IPUS(H). That is, because the economy in aggregate is nearly competitive, the percentage change in output prices is just equal to the percentage change in input prices minus the percentage change in productivity. Substitute relationship (28) into the second bracketed term of relationship (27). Rearranging terms gives (29) RPI(H) - %D IPLEC(H) This is the historical difference between the rate of change of prices for aggregate output and industry input, which is, by definition, DH. (iii) The Use the Residual Value Method for Determining Capital Prices The above discussion shows that the Commission in the 1997 Price Cap Review Order implemented price caps and calculated an historically justified X-factor equal to the sum of DH and EH where the formulas for estimating DH and EH are given by (30) DH = RPI(H) - %D IPLEC(H) (31) EH = %D TFPLEC(H). This being the case, it is possible to explain the error that the Commission made in the 1997 Price Cap Review Order in the calculation of the historically justified X-factor. The 1997 Staff TFP Study associated with the 1997 Price Cap Review Order made a conceptual error in using actual imputed cost of capital when measuring the productivity of regulated companies. The problem arises because of the use of the residual value approach for determining the cost of the capital input. With this approach, the Commission in the 1997 Price Cap Review Order first directly determined the price and quantity used of every input (i.e., labor and materials) except capital. That is, it directly calculated the realized return to every non-capital input. Next, based on an estimate of the capital stock, the assumption that all revenue is distributed to the inputs, and given the returns to labor and materials based on historical data, a cost of capital is imputed. Conceptually, the difference (residual) between revenue and the required returns to all non-capital inputs (which is just the nominal amount used to impute a cost of capital) consists of two parts. The first part is the required return to capital. The second part is the excess profit earned by the firm. Instead of attempting to separate this difference into two parts, however, the Commission in the 1997 Price Cap Review Order simply assumed that all of this residual was the required return to capital, i.e., that no excess profit was earned. This is a reasonable assumption for a competitive market, and even for non-competitive markets, as long as the goal of the study is simply to measure the productivity gains revealed by market forces. In a regulatory setting, however, the productivity gains are "revealed" by the application of the X- factor, not by market forces. By attributing all of the residual to the capital inputs, the residual value method tends automatically to define whatever profits or losses the LECs realized during the historical period as increases or decreases in the cost of capital inputs. Suppose, for example, that the Commission chose too low an X-factor during the historical period being considered, causing the LECs' profits to increase. Under the residual value method, the Commission would conclude that the historical cost of LEC capital inputs rose more rapidly during this period than it actually did. The Commission then would predict an equally large growth in LEC capital cost in the future, and thus calculate an X-factor that was still too low. Consequently, LECs' profits would continue to increase despite no increase in LEC productivity. On the other hand, suppose that the Commission chose too high an X-factor during the historical period, causing LECs' profits to decrease. Under the residual value method, the Commission would conclude that the historical cost of LEC capital inputs fell more rapidly during this period than it actually did. The Commission then would predict an equally large price reduction for capital in the future, and thus calculate an X-factor that was still too high. Consequently, LECs' profits would continue to decrease despite no decrease in LEC productivity. In either scenario, the LECs' productivity remains constant. Only their rate of return is affected by the incorrect X-factor. Thus, the residual value method produces an X-factor that is biased in the same direction as whatever bias existed in the historical period. To be useful to regulators, then, any X-factor study of a regulated industry must augment or reduce the actual amounts of capital compensation in such a way as to duplicate the forces of a competitive market. Merely replacing the residual capital value method of determining the capital compensation rate with an independent price index would not be sufficient. It would merely cause counterbalancing shifts between the input price differential and the productivity growth rate differential. To capture the gains in productivity that would have been revealed in a competitive market, total capital compensation must vary with the cost of capital one would observe in a competitive market. In order to correct the miscalculation of the LECs' cost of capital in the 1997 Staff TFP study, it is necessary to replace the TFP study's cost of capital with a competitive cost for the inputs during the historical years. Next, it is necessary to adopt a surrogate to emulate a competitive cost of capital for LECs because LECs have never operated in a competitive market. An independent price series is employed to compute the annual change in the cost of capital for a competitive market. Specifically, Moody's Baa corporate bond rate reported in the 1999 Economic Report of the President (Table B-73) is used to calculate the adjustment. These data are reproduced in Table B-8. Combining the base year imputed cost of capital from the 1997 staff TFP study with the change in the competitive cost of capital gives an independent competitive cost of capital for LECs in each year of the historical period. This competitive cost of LECs' capital input is used in conjunction with an adjusted cost of labor (see below) and materials inputs from the 1997 staff TFP study, as well as the percent change in the U.S. nonfarm business sector input price growth, to compute the corrected input price differential portion of the historical X-factor. Recalculating the LECs' historical cost of capital changes the level of the LECs' revenues, taxes, and operating expenses for the historical years. If the difference between the estimate of the competitive cost of LECs' capital inputs and the imputed cost from the 1997 staff TFP study is computed, an estimate of the excess profits realized by LECs for each year of the historical period is obtained. Reducing the LECs' total revenue by this amount results in a revenue stream that emulates what LECs would have realized in a competitive market. A reduction in revenue implies a reduction in taxes, which in turn reduces total operating expense. It is this reduced total operating expense that is used in the computation of the materials input quantity. Reducing LEC revenues to what would have been realized in a competitive market also changes the LECs' TFP portion of the X-factor. The new revenue level changes the LECs' input growth rate by altering the relative share of revenue accounted for by each input - labor, materials, and capital. In other words, the LECs' input growth rate changes in response to the changes in the relative weight of each input factor. Combining the output growth rate from the 1997 staff TFP study with the revised LECs' input growth rate yields the corrected TFP growth rate for the LECs. (e) Materials - Part II As noted, an adjustment to the cost of capital changes the effective total operating expense due to a change in taxes. The adjusted total operating expense less labor compensation less depreciation and amortization divided by the materials price index gives materials input quantity. The materials price index is the same as that used in the 1997 Price Cap Review Order. The materials input quantity series is found in Table B-9. (f) Total Input Quantity Index Having constructed input quantity indexes for each of the factors of production, a composite input index can be computed. Using relative shares of payments to each of the factors as weights, an aggregate input value for each year is calculated. These values are used to compute a Fisher Ideal Index for input quantity. To compute the relative shares, payments to each of the factors is needed. The payment to labor is adjusted total compensation, the payment to materials is adjusted materials expense, and the payment to capital is adjusted property income with depreciation. Adjusted property income is capital stock quantity in period t-1 times the imputed competitive cost of capital for period t. Adjustments to material payments and property income (capital compensation) have been previously discussed. Adjustments to labor compensation are deferred to a subsequent section. Payments to the factors of production together with factor shares are given in Table B-10. The composite Fisher Ideal Input Quantity Index is derived by chaining the Fisher Ideal Input Quantity Index. The resulting Fisher Ideal Chained Input Quantity Index is given in Table B-11. Note that the percentage changes (growth rate) are calculated as natural logarithmic (loge) changes. Total Factor Productivity for LECs and the TFP Differential The percentage change in total factor productivity is the measure for the productivity component of the X-factor. This is calculated as the percentage change in total output (Table B-4) less the percentage change in input quantity (Table B-11). For both output and input quantity, the Fisher Ideal Chained Index is used. The percentage changes (growth rate) are calculated as natural logarithmic (loge) changes. Annual changes together with the historical averages for various periods is given in Table B-10. The productivity component of the X-factor as specified in relationship (26) consists of the difference between LECs TFP growth rate and the TFP growth rate of the aggregate economy. This differential is presented in Table B-12. The Bureau of Labor Statistics of the U.S. Department of Labor estimate of Private Nonfarm Business Sector Multifactor Productivity is used to compute the TFP growth rate for the aggregate economy., The 1998 value of this multifactor productivity series is not currently available. It is forecast by estimating a structural relationship between output per hours of all persons employed and private nonfarm business multifactor productivity using a double logarithmic (loge) transformation. Data for output per hour of all persons employed for 1998 is available from the Bureau of Labor Statistics and is used to generate the forecast. The results indicate that total factor productivity of LECs grew about 4.2 percent faster than productivity in the nonfarm private business sector of the economy over the 1986 to 1998 period. Calculation of Input Price Indexes The second component of the X-factor is the input price differential. This is just the value in the second set of parentheses in equation (26), %D IPUS(H) - %D IPLEC(H). Focus for now on the changes in LEC input prices. The price of each factor of production will be considered separately. (a) Labor Total labor compensation divided by the number of employees gives a simple way of computing the average annual price of labor. The resulting series, however, is not homogeneous over the 1985-1998 period and hence it introduces measurement error into the analysis. Between 1985 and 1998, the price of labor, defined as the average compensation per employee, increased in real terms at a 3.6 percent annual rate. In terms of total compensation, the amount paid for labor was approximately the same in 1985 as it was in 1998 even though the number of workers in the aggregate had been reduced by over 30 percent. The reason for this, at least in part, is that, coincident with the adoption of price cap plans, labor force reductions were accomplished by offering employees monetary incentives to leave the company (i.e., buyouts). There is some variability in the applicable accounting rules, but these payments were generally accrued as one-time charges against current earnings., To have a labor price series meaningful for TFP analysis, it is necessary to adjust for the impact of the exogenous changes in labor compensation and accounting rules. This is accomplished by adjusting the labor compensation series to net out one-time charges for such things as buyouts and accounting rule changes. Sufficiently detailed information to allow for such an adjustment to the total compensation series is not available from Statistics of Communication Common Carriers, the source of the total labor compensation and number of employees data used in the 1997 Price Cap Review Order. Alternate data must be relied upon. These data are discussed below. From the FCCs' ARMIS Report 43-02, total compensation for the year is defined to equal payroll, including salaries, wages, and payroll related benefits. Separate items from the same report include salaries and wages and benefits. Salaries and wages are defined to equal salaries, wages, commissions, bonuses, incentive awards, and termination payments. Benefits include pensions, saving plan contributions, worker's compensation, life and health insurance, social security and other payroll taxes. These data series are reported in Table B-5. Benefits are equal to approximately 20 percent of the sum of salaries plus wages plus benefits for the years 1988-1990. Subsequently, they increase annually, reaching a maximum of 30 percent in 1994 and then decline to approximately 21 percent in 1998. The increase in the proportion of benefits above 20 percent is assumed to be the amount attributed to buyouts, accounting rule changes, and so on. This amount is nominally referred to as excess benefits. This is the amount that must be netted out to give a labor compensation series that can be used in the TFP analysis. Using this rule, excess benefits are computed with the resulting amount subtracted from total compensation to give the adjusted labor compensation series in Table B-5. The values in this series divided by the number of workers gives an adjusted labor price. The adjusted labor price index is then computed (Table B-5). (b) Capital As noted previously, in order to correct the miscalculation of the LECs' cost of capital in the 1997 Staff TFP study, it is necessary to replace the TFP study's cost of capital with a competitive cost for the inputs during the historical years. Next, it is necessary to adopt a surrogate to emulate a competitive cost of capital for LECs because LECs have never operated in a competitive market. An independent price series is employed to compute the annual change in the cost of capital for a competitive market. Specifically, Moody's Baa corporate bond rate reported in the 1999 Economic Report of the President (Table B-73) is used to calculate the adjustment. Combining the base year imputed cost of capital from the 1997 staff TFP study with the change in the competitive cost of capital gives an independent competitive cost of capital for LECs in each year of the historical period. (c) Materials The materials price index is the same as that used in the 1997 Price Cap Review Order. The price index is based on those categories of expenditures from the National Input-Output Tables compiled by the Bureau of Economic Analysis of the U.S. Department of Commerce that focus on materials purchases by communications industries. The materials price index is a Tornquist index. The materials price index is given in Table B-9. (d) Total Input Price Index Having constructed input price indexes for each of the factors of production, a composite input index can be computed. Using relative shares of payments to each of the factors as weights, an aggregate input value for each year is calculated. These values are used to compute a Fisher Ideal Index for input price. To compute the relative shares, payments to each of the factors is needed. The payment to labor is adjusted total compensation, the payment to materials is adjusted materials expense, and the payment to capital is adjusted property income with depreciation. Adjusted property income is capital stock quantity in period t-1 times the imputed competitive cost of capital for period t. Adjustments to material payments and property income (capital compensation) have been previously discussed. Adjustments to labor compensation are deferred to a subsequent section. Payments to the factors of production together with factor shares are given in Table B-10. The composite Fisher Ideal Input Price Index is derived by chaining the Fisher Ideal Input Price Index. The resulting Fisher Ideal Chained Input Price Index is given in Table B-13. Note that the percentage changes (growth rate) are calculated as natural logarithmic (loge) changes. Input Price Differential The input price differential, the second component of the X-factor, is computed as the percentage change in input prices for the aggregate economy less the percentage change in LECs' input prices. The measure of aggregate input price change used is the Nonfarm Business Sector Input Price Index compiled by the Bureau of Labor Statistics of the U.S. Department of Labor. The input price differential is presented in Table B-12. The results suggest that input prices in the nonfarm private business sector of the economy grew about 1.8 percent faster than input prices for LECs' over the 1986 to 1998 period. The X-Factor The computed X-factor defined by equation (26) and consisting of a productivity component plus an input price differential component is presented in Table B-12. Averages for different time period intervals are presented. The average is in the range of 5.8 to 6.3 percent. It is interesting to note that the averages are approximately equal for these different time periods. Table B-1. LEC Interstate Revenue ($) - 1985-1998 ______________________________________________________________________________ Year End User Interstate Switched Special Access Total Interstate Revenue Access Revenue Revenue Revenue ______________________________________________________________________________ 1985 1499413893 10906203190 1960688644 14366305727 1986 2400475814 10484265170 2574800716 15459541700 1987 3090639929 9611996187 2657677439 15360313555 1988 3604221000 9662529000 2539698000 15806448000 1989 4398692000 9092575000 2253922000 15745189000 1990 4679142000 8595750000 2209064000 15483956000 1991 4828177000 8514130000 2119037000 15461344000 1992 4963262000 8650880000 2153565000 15767707000 1993 5244094000 8999065000 2097997000 16341156000 1994 5589662000 9293783000 2217125000 17100570000 1995 5770285000 9332869000 2529667000 17632821000 1996 5930960000 9409639000 3070598000 18411197000 1997 6268026000 8763815000 3851028000 18882869000 1998 7928205000 7447289000 4894584000 20270078000 ______________________________________________________________________________ Source: Federal Communications Commission, Statistics of Communication Common Carriers [various years] Table B-2. LEC Revenue ($) by Type of Service1 - 1985-1998 ______________________________________________________________________________ Year Local Service Intrastate Toll Interstate Total Revenue and Intrastate Service Revenue Access Revenue Service Revenue ______________________________________________________________________________ 1985 26960554164 13047095682 14366305727 54373955573 1986 28626174049 13538946795 15459541700 57624662544 1987 29150842991 14166723124 15360313555 58677879670 1988 29226988000 14994975000 15806448000 60028411000 1989 29973157000 14868219000 15745189000 60586565000 1990 30699085000 15014729000 15483956000 61197770000 1991 32059008000 14522276000 15461344000 62042628000 1992 33359990000 14225181000 15767707000 63352878000 1993 34598957000 14496831000 16341156000 65436944000 1994 35758637000 14355983000 17100570000 67215190000 1995 37684860000 13123225000 17632821000 68440906000 1996 40523387000 12987476000 18411197000 71922060000 1997 42460592000 12308613000 18882869000 73652074000 1998 45643024000 12236469000 20270078000 78149571000 ______________________________________________________________________________ 1 This excludes miscellaneous services. Source: Federal Communications Commission, Statistics of Communication Common Carriers [various years] Table B-3. Interstate Output Index - 1985-1998 ________________________________________________________________________________________________________________________________________________________________________________ Year End User Interstate Special Number of Number of Number of Laspeyres Paasche Fisher Fisher Revenue Switched Access Access Switched Special Output Output Ideal Ideal Share Access Revenue Lines Access Access Index Index Output Chained Revenue Share Minutes Lines Index Output Share Index ________________________________________________________________________________________________________________________________________________________________________________ 1985 0.10437 0.75915 0.13647 92671959 156853820000 1230590 1 1 1 1 1986 0.15527 0.67817 0.16655 95333884 157302701000 1664101 1.05324 1.05225 1.05275 1.05275 1987 0.20120 0.62576 0.17302 98228585 173154171000 1764445 1.08309 1.07881 1.08095 1.13797 1988 0.22802 0.61130 0.16067 98270787 187663836000 2701817 1.14444 1.11496 1.12960 1.28546 1989 0.27936 0.57748 0.14314 101190050 210406134000 2448090 1.06576 1.05892 1.06233 1.36559 1990 0.30219 0.55513 0.14266 103857988 231960296000 3518005 1.12908 1.11449 1.12176 1.53188 1991 0.31227 0.55067 0.13705 107383807 246710182000 5151699 1.11181 1.09485 1.10330 1.69012 1992 0.31477 0.54864 0.13658 108938065 262187655000 6033139 1.06251 1.06025 1.06138 1.79387 1993 0.32091 0.55069 0.12838 112196681 278173161000 10153615 1.13614 1.10261 1.11925 2.00781 1994 0.32686 0.54347 0.12965 115264861 298342017323 13824365 1.09511 1.08680 1.09095 2.19042 1995 0.32724 0.52928 0.14346 119887506 334981582000 16107677 1.10126 1.09992 1.10059 2.41077 1996 0.32213 0.51108 0.16677 125333996 363445050000 20775150 1.10141 1.10070 1.10105 2.65440 1997 0.33194 0.46411 0.20394 131618657 387587696669 28051449 1.10851 1.10823 1.10837 2.94207 1998 0.39112 0.36740 0.24146 138481147 407903661000 34142101 1.08591 1.08785 1.08688 3.19768 ________________________________________________________________________________________________________________________________________________________________________________ Source: Federal Communications Commission, Statistics of Communication Common Carriers [various years] and Federal Communications Commission, Monitoring Reports [various years] Table B-4. Total LEC Output Index - 1985-1998 ______________________________________________________________________________________________________________________________________________________________ ___ Year Revenue Revenue Revenue Local Intrastate Interstate Laspeyres Paasche Fisher Fisher Growth Share - Share - Share - DEMs DEMs Fisher Output Output Ideal Ideal Rate (%) Local Intrastate Interstate (000) (000) Ideal Index Index Output Chained Toll Chained Index Output Output Index Index ______________________________________________________________________________________________________________________________________________________________ ___ 1985 0.4958 0.2400 0.2642 1380145900 164191177 1 1 1 1 1 1986 0.4968 0.2350 0.2683 1396014000 173173536 1.05275 1.03276 1.03218 1.03247 1.03247 3.19590 1987 0.4968 0.2414 0.2618 1404776000 183597411 1.13797 1.03908 1.03778 1.03843 1.07215 3.77145 1988 0.4869 0.2498 0.2633 1469781200 191904837 1.28546 1.06784 1.06673 1.06728 1.14430 6.51202 1989 0.4947 0.2454 0.2599 1496826800 207298177 1.36559 1.04541 1.04429 1.04485 1.19562 4.38736 1990 0.5016 0.2453 0.2530 1514588700 217913904 1.53188 1.05008 1.04755 1.04881 1.25399 4.76646 1991 0.5167 0.2341 0.2492 1512946987 219713721 1.69012 1.02751 1.02531 1.02641 1.28711 2.60724 1992 0.5266 0.2245 0.2489 1558762543 224278538 1.79387 1.03580 1.03567 1.03573 1.33311 3.51158 1993 0.5287 0.2215 0.2497 1640600472 227540869 2.00781 1.06059 1.05960 1.06010 1.41324 5.83650 1994 0.5320 0.2136 0.2544 1719329169 235362364 2.19042 1.05559 1.05559 1.05559 1.49181 5.41052 1995 0.5506 0.1917 0.2576 1802545593 246926539 2.41077 1.06183 1.06161 1.06172 1.58389 5.98968 1996 0.5634 0.1806 0.2560 1955027929 263719641 2.65440 1.08554 1.08570 1.08562 1.71951 8.21567 1997 0.5765 0.1671 0.2564 2179309093 273526580 2.94207 1.09909 1.09937 1.09923 1.89015 9.46127 1998 0.5841 0.1566 0.2594 2275450746 286005821 3.19768 1.05533 1.05512 1.05523 1.99454 5.37567 ______________________________________________________________________________________________________________________________________________________________ ___ Source: Federal Communications Commission, Statistics of Communication Common Carriers [various years] and Federal Communications Commission, Monitoring Reports [various years] Table B-5. Price of Labor - 1985-1998 ____________________________________________________________________________________________________________________________________________________________________________________ Year Labor Number of ARMIS ARMIS Ratio Excess Adjusted Labor Labor Labor Labor Labor Labor Compensa- Employees Salaries+ Benefits Benefits Labor Price Price Price Price Price - Price - Tion ($) Wages ($000) ($) Compensa- (original) (adjusted) Index Index % Change % Change ($000) tion ($) ($) ($) (original) (adjusted) (original) (adjusted) A B B/(A+B) ____________________________________________________________________________________________________________________________________________________________________________________ 1985 16991572326 504113 16991572326 33705.88008 33705.88008 1.00000 1.00000 1986 16728435454 482698 16728435454 34656.11097 34656.11097 1.02819 1.02819 2.78018 2.78018 1987 16978905847 477714 16978905847 35541.98924 35541.98924 1.05447 1.05447 2.52407 2.52407 1988 17030359791 466827 15033849 3636033 0.19475 0 17030359791 36481.09426 36481.09426 1.08234 1.08234 2.60794 2.60794 1989 16910850694 461149 14977589 3669768 0.19680 0 16910850694 36671.12082 36671.12082 1.08797 1.08797 0.51954 0.51954 1990 17586868921 443105 15230268 3768099 0.19834 0 17586868921 39690.07102 39690.07102 1.17754 1.17754 7.91115 7.91115 1991 17186211200 414457 15038534 4537703 0.23180 622455600 16563755600 41466.81369 39964.95559 1.23025 1.18570 4.37924 0.69019 1992 17160988000 411167 14976159 4920448 0.24730 941126600 16219861400 41737.26977 39448.35408 1.23828 1.17037 0.65011 -1.30106 1993 17956438000 395639 15479969 5918883 0.27660 1639112600 16317325400 45385.91494 41242.96493 1.34653 1.22361 8.38073 4.44882 1994 17154284000 367196 15085400 6539928 0.30242 2214862400 14939421600 46716.96859 40685.14254 1.38602 1.20706 2.89056 -1.36176 1995 16203522000 346843 15088974 5677574 0.27340 1524264400 14679257600 46717.16598 42322.48481 1.38602 1.25564 0.00042 3.94555 1996 18457448000 338040 15337179 5140712 0.25104 1045133800 17412314200 54601.37262 51509.62667 1.61994 1.52821 15.59473 19.64502 1997 17451673000 338177 15358125 4395933 0.22253 445121400 17006551600 51605.14464 50288.90670 1.53104 1.49199 -5.64377 -2.39842 1998 18262990000 345317 15302883 4263993 0.21792 350617800 17912372200 52887.60762 51872.25709 1.56909 1.53897 2.45477 3.09996 ____________________________________________________________________________________________________________________________________________________________________________________ Source: Federal Communications Commission, Statistics of Communication Common Carriers [various years] and ARMIS Reports 43-02, Table 1.B. Table B-6. Capital Stock Adjustments and the Average Rate of Depreciation - 1985-1998 (Dollar amounts shown in 000) ______________________________________________________________________________________________________________________________________________________________ ___ Year TPIS.BOY Capital TPIS.EOY Capital Adjustment Adjusted Adjusted Depreciation Adjusted Additions Retires Factor Capital TPIS.EOY Accruals Depreciation Additions Rate A B C D=A+B-C E F=B*E G+A+F-D H I=H/((A+G)/2) ______________________________________________________________________________________________________________________________________________________________ ___ 1985 138879365 15001998 149061793 4819570 0.888 13321774 147381569 10241376 7.15527 1986 149061793 14842725 159010189 4894329 0.888 13180340 157347804 11826961 7.71971 1987 159010189 14138370 168505114 4643445 0.888 12554873 166921617 13311655 8.16837 1988 168505114 14284742 175860216 6929640 1 14284742 175860216 13134992 7.62852 1989 175860216 13283569 182978381 6165404 1 13283569 182978381 13420810 7.48014 1990 182978381 14476334 187168695 10286020 1 14476334 187168695 13439933 7.26194 1991 187168695 14527049 192034545 9661199 1 14527049 192034545 13200593 6.96228 1992 192034545 14611866 196411915 10234496 1 14611866 196411915 13337581 6.86714 1993 196411915 14860116 203082418 8189613 1 14860116 203082418 14032782 7.02527 1994 203082418 14717999 209325562 8474855 1 14717999 209325562 14863196 7.20801 1995 209325562 15374568 217430207 7269923 1 15374568 217430207 15358553 7.19782 1996 217430207 18026150 227317120 8139237 1 18026150 227317120 16252281 7.30855 1997 227317120 18253199 236896179 8674140 1 18253199 236896179 16667034 7.18077 1998 236896179 18553791 248970288 6479682 1 18553791 248970288 17154619 7.06145 avg1(85-98) 7.30180 var2(85-98) 0.11142 ______________________________________________________________________________________________________________________________________________________________ ___ 1 avg denotes the arithmetic mean of the series. 2 var denotes the variance of the series. Source: FCC Form M Table B-7. Quantity of Capital for 1985-1998 and the Imputed Cost of Capital for 1991 (Dollar amounts shown in 000) _____________________________________________________________________________________________________________________ Year Benchmark Adjusted BEA Capital Capital Property Imputed Capital Capital Composite Stock Stock Income Cost of Stock Additions Asset Price Quantity Quantity /w Depreciation Capital Index Index _____________________________________________________________________________________________________________________ 1984 103903095 1985 109602959 13321774 1 109602959 1 23445593794 1986 13180340 1.01048 114643584 1.04599 26792578943 1987 12554873 1.02734 118493306 1.08111 27701751800 1988 14284742 1.03047 123703569 1.12865 26866209000 1989 13283569 1.07018 127083465 1.15949 25845853000 1990 14476334 1.08973 131088425 1.19603 25584541000 1991 14527049 1.10222 134696416 1.22895 24641357000 0.18798 1992 14611866 1.10830 138045138 1.2595 26477135000 1993 14860116 1.11231 141325020 1.28943 26914823000 1994 14717999 1.11766 144174285 1.31542 26366385000 1995 15374568 1.11481 147438176 1.3452 27166096000 1996 18026150 1.11862 152787122 1.39401 30414808000 1997 18253199 1.11764 157962763 1.44123 30679731000 1998 18553791 1.11769 163028757 1.48745 33830949286 _____________________________________________________________________________________________________________________ Source: Table B-6 and National the Income and Product Accounts compiled by the Bureau of Economic Analysis of the U.S. Department of Commerce Table B-8. Cost of Capital - 1985-1998 _____________________________________________________________ Year Moody's Imputed Competitive Baa Corporate Competitive Cost of Bond Rate (%) Cost of Capital Capital ($) Index ______________________________________________________________ 1985 12.72 0.21717 1.00000 1986 10.39 0.19387 0.89271 1987 10.58 0.19577 0.90146 1988 10.83 0.19827 0.91297 1989 10.18 0.19177 0.88304 1990 10.36 0.19357 0.89133 1991 9.8 0.18797 0.86554 1992 8.98 0.17977 0.82778 1993 7.93 0.16927 0.77944 1994 8.62 0.17617 0.81121 1995 8.2 0.17197 0.79187 1996 8.05 0.17047 0.78496 1997 7.86 0.16857 0.77621 1998 7.22 0.16217 0.74674 _____________________________________________________________ Source: Moody's Baa Corporate Bond Rate is from Table B-73 of the Economic Report of the President- 1999, U.S. Government Printing Office, Washington, DC, 1999. Table B-9. Materials Input Quantity - 1985-1998 _________________________________________________________________________________________________________ Year Materials Adjusted Depreciation Adjusted Materials Materials Materials Price Total and Employee Expense Quantity Quantity Index Operating Amortization Compensation ($) Index Expense ($) Expense ($) ($) A B C D E=B-C-D F=E/A _________________________________________________________________________________________________________ 1985 1.00000 40609705224 10024710656 16991572326 13593422242 13593422242 1 1986 1.03135 40262200069 11592001248 16728435454 11941763367 11578768961 0.85179 1987 1.05353 42242744320 13316999560 16978905847 11946838913 11339818432 0.83421 1988 1.08639 45494083588 13646937000 17030359791 14816786797 13638527159 1.00332 1989 1.12623 47773003404 13860101000 16910850694 17002051710 15096429424 1.11057 1990 1.17203 49160848820 13931515000 17586868921 17642464899 15052912382 1.10737 1991 1.20494 50278593400 13499778000 16563755600 20215059800 16776818597 1.23419 1992 1.23480 48875289820 13822882000 16219861400 18832546420 15251495319 1.12198 1993 1.25535 49744107201 13244514000 16317325400 20182267801 16077004661 1.18270 1994 1.29144 53129310801 15068058000 14939421600 23121831201 17903914391 1.31710 1995 1.32167 54381863101 15556284000 14679257600 24146321501 18269554050 1.34400 1996 1.36140 54780054849 16377242000 17412314200 20990498649 15418318385 1.13425 1997 1.39550 57364316191 16758832000 17006551600 23598932591 16910736361 1.24404 1998 1.43074 58408447446 17646242000 17912372200 22849833246 15970695654 1.17488 ________________________________________________________________________________________________________ Source: Materials price index comes from the Input/Output Tables compiled by the Bureau of Economic Analysis of the U.S. Department of Commerce, depreciation and amortization expense data come from the Statistics of Communication Common Carriers, and the other values are derived as detailed in the text. Table B-10. Factor of Production Shares of Total Payments - 1985-1998 _____________________________________________________________________________________________________________ Year Adjusted Property Adjusted Adjusted Labor Capital Materials Labor Income Materials Total Share Share Share Compensa- /w Deprecia- Payment ($) Factor tion ($) tion ($) Payments ($) _____________________________________________________________________________________________________________ 1985 16991572326 22565165047 13593422242 53150159615 0.31969 0.42455 0.25576 1986 16728435454 21249284636 11941763367 49919483458 0.33511 0.42567 0.23922 1987 16978905847 22444359210 11946838913 51370103970 0.33052 0.43691 0.23256 1988 17030359791 23494272047 14816786797 55341418635 0.30773 0.42453 0.26773 1989 16910850694 23723264293 17002051710 57636166697 0.29341 0.41160 0.29499 1990 17586868921 24600194384 17642464899 59829528203 0.29395 0.41117 0.29488 1991 16563755600 24641359753 20215059800 61420175153 0.26968 0.40119 0.32913 1992 16219861400 24215061717 18832546420 59267469536 0.27367 0.40857 0.31776 1993 16317325400 23367604541 20182267801 59867197742 0.27256 0.39032 0.33712 1994 14939421600 24897949618 23121831201 62959202419 0.23729 0.39546 0.36725 1995 14679257600 24794387029 24146321501 63619966130 0.23073 0.38973 0.37954 1996 17412314200 25134537868 20990498649 63537350717 0.27405 0.39559 0.33036 1997 17006551600 25756104311 23598932591 66361588503 0.25627 0.38812 0.35561 1998 17912372200 25617626841 22849833246 66379832287 0.26985 0.38592 0.34423 _____________________________________________________________________________________________________________ Source: Federal Communications Commission, Statistics of Communication Common Carriers [various years] with adjustments as described in the text. Table B-11. Total LEC Input Quantity Index - 1985-1998 __________________________________________________________________________________________________________________________________________________ Year Labor Capital Materials Labor Capital Materials Laspeyres Paasche Fisher Fisher Growth Share Share Share Quantity Quantity Quantity Input Input Ideal Ideal Rate (%) Index Index Quantity Quantity Input Chained Index Index Quantity Input Index Quantity Index __________________________________________________________________________________________________________________________________________________ 1985 0.31969 0.42455 0.25576 504113 1 1 1 1 1 1 1986 0.33511 0.42567 0.23922 482698 1.04599 0.85179 0.96804 0.96360 0.96582 0.96582 -3.47804 1987 0.33052 0.43691 0.23256 477714 1.08111 0.83421 1.00590 1.00588 1.00589 0.97150 0.58715 1988 0.30773 0.42453 0.26773 466827 1.12865 1.00332 1.05882 1.05913 1.05898 1.02880 5.73029 1989 0.29341 0.41160 0.29499 461149 1.15949 1.11057 1.03648 1.03715 1.03681 1.06668 3.61531 1990 0.29395 0.41117 0.29488 443105 1.19603 1.10737 1.00064 0.99974 1.00019 1.06688 0.01899 1991 0.26968 0.40119 0.32913 414457 1.22895 1.23419 1.02608 1.02662 1.02635 1.09499 2.60077 1992 0.27367 0.40857 0.31776 411167 1.25950 1.12198 0.97791 0.97651 0.97721 1.07003 -2.30554 1993 0.27256 0.39032 0.33712 395639 1.28943 1.18270 1.01657 1.01592 1.01625 1.08742 1.61153 1994 0.23729 0.39546 0.36725 367196 1.31542 1.31710 1.02658 1.02765 1.02712 1.11690 2.67569 1995 0.23073 0.38973 0.37954 346843 1.34520 1.34400 1.00330 1.00269 1.00300 1.12025 0.29912 1996 0.27405 0.39559 0.33036 338040 1.39401 1.13425 0.94905 0.94842 0.94874 1.06282 -5.26234 1997 0.25627 0.38812 0.35561 338177 1.44123 1.24404 1.04549 1.04625 1.04587 1.11157 4.48479 1998 0.26985 0.38592 0.34423 345317 1.48745 1.17488 0.99809 0.99732 0.99770 1.10902 -0.22988 __________________________________________________________________________________________________________________________________________________ Source: Table B-10, Federal Communications Commission, Statistics of Communication Common Carriers [various years], Table B-7, and Table B-9. Table B-12. Summary of the Components of LECs' Price Cap X-Factor (excluding the Consumer Productivity Dividend) - 1985-1998 ______________________________________________________________________________________________________________________________________________________________ ___ Year U.S. LECs' LECs' LECs' TFP U.S. LECs' Input X-factor Previous Nonfarm Output Input TFP Differential Nonfarm Input Price (%) X-factor1 Business Growth Growth Growth (%) Business Price Differential (%) Sector Rate (%) Rate (%) Rate (%) Sector Growth (%) TFP Input Rate (%) Growth Price Rate (%) Growth Rate (%) A B C D = B-C E = D-A F G H = F-G I = E+H J ______________________________________________________________________________________________________________________________________________________________ ___ 1986 1.10166 3.19590 -3.47804 6.67394 5.57228 2.80830 -3.15211 5.96041 11.53269 -0.5 1987 -0.39920 3.77146 0.58715 3.18431 3.58351 2.53178 1.92376 0.60802 4.19153 5.0 1988 0.29955 6.51202 5.73029 0.78173 0.48218 3.72958 2.39282 1.33676 1.81894 5.0 1989 0.19920 4.38736 3.61531 0.77206 0.57285 3.03629 -1.52894 4.56522 5.13808 7.9 1990 -0.69895 4.76646 0.01899 4.74748 5.44643 3.30913 3.88344 -0.57432 4.87211 8.8 1991 -1.41274 2.60725 2.60077 0.00647 1.41921 2.05824 -0.13437 2.19261 3.61182 5.8 1992 1.61294 3.51159 -2.30554 5.81713 4.20419 2.88104 -1.36727 4.24830 8.45250 3.4 1993 0.09995 5.83651 1.61153 4.22497 4.12502 3.71664 -0.64768 4.36432 8.48934 4.7 1994 0.39880 5.41052 2.67569 2.73483 2.33603 3.50341 2.22171 1.28171 3.61774 5.4 1995 0.29806 5.98969 0.29912 5.69056 5.39250 1.96268 0.84015 1.12253 6.51503 6.8 1996 1.47713 8.21568 -5.26234 13.47802 12.00089 1.38258 5.65415 -4.27157 7.72932 1997 0.39024 9.46127 4.48479 4.97648 4.58623 1.89887 -0.22680 2.125670 6.71190 1998 0.59259 5.37568 -0.22988 5.60556 5.01297 0.71810 0.18976 0.52834 5.54131 avg2(86-98) 4.21033 1.80677 6.01710 var3(86-98) 8.08404 6.48203 6.22591 avg(91-98) 4.88463 1.44899 6.33362 var(91-98) 8.81863 6.38164 3.33409 avg(86-95) 3.31342 2.51056 5.82398 5.23 var(86-95) 3.55698 4.06310 7.69230 5.93 avg(91-95) 3.49539 2.64189 6.13729 5.22 var(91-95) 2.03050 1.98155 4.75223 1.29 ______________________________________________________________________________________________________________________________________________________________ ___ 1 X-factor reported in the 1997 Price Cap Review Order. 2 avg denotes the arithmetic mean of the series. 3 var denotes the variance of the series. Source: Bureau of Labor Statistics' Multifactor Productivity Table 2: Private Nonfarm Business: Productivity and Related Indexes (annual and quarterly tables), Table B-4, Table B-11, and Table B-13. Table B-13. Total LEC Input Price Index - 1985-1998 __________________________________________________________________________________________________________________________________________________ Year Labor Capital Materials Labor Capital Materials Laspeyres Paasche Fisher Fisher Growth Share Share Share Price Price Price Input Input Ideal Ideal Rate (%) Index Index Index Price Price Input Chained Index Index Price Input Index Price Index __________________________________________________________________________________________________________________________________________________ 1985 0.31969 0.42455 0.25576 1 1 1 1 1 1 1 1986 0.33511 0.42567 0.23922 1.02819 0.89271 1.03135 0.97148 0.96647 0.96897 0.96897 -3.15211 1987 0.33052 0.43691 0.23256 1.05447 0.90146 1.05353 1.01788 1.01768 1.01778 0.98620 1.76258 1988 0.30773 0.42453 0.26773 1.08234 0.91297 1.08639 1.02157 1.02184 1.02170 1.00760 2.14711 1989 0.29341 0.41160 0.29499 1.08797 0.88304 1.12623 0.99750 0.99801 0.99776 1.00534 -0.22468 1990 0.29395 0.41117 0.29488 1.17754 0.89133 1.17203 1.04001 1.03918 1.03960 1.04515 3.88344 1991 0.26968 0.40119 0.32913 1.18570 0.86555 1.20494 0.99842 0.99889 0.99866 1.04375 -0.13437 1992 0.27367 0.40857 0.31776 1.17037 0.82779 1.23480 0.98717 0.98567 0.98642 1.02958 -1.36727 1993 0.27256 0.39032 0.33712 1.22361 0.77944 1.25535 0.99388 0.99321 0.99354 1.02293 -0.64768 1994 0.23729 0.39546 0.36725 1.20706 0.81121 1.29144 1.02192 1.02302 1.02247 1.04591 2.22171 1995 0.23073 0.38973 0.37954 1.25564 0.79187 1.32167 1.00872 1.00816 1.00844 1.05473 0.84015 1996 0.27405 0.39559 0.33036 1.52821 0.78497 1.36140 1.05810 1.05824 1.05817 1.11609 5.65415 1997 0.25627 0.38812 0.35561 1.49199 0.77622 1.39550 0.99737 0.99810 0.99773 1.11356 -0.22680 1998 0.26985 0.38592 0.34423 1.53897 0.74675 1.43074 1.00231 1.00149 1.00190 1.11568 0.18976 __________________________________________________________________________________________________________________________________________________ Source: Table B-10, Table B-5, Table B-8, and Table B-9. Appendix C The Staff Imputed X Study Florence Setzer This study estimates an X-factor for LEC price caps by solving for the past X-factor that would have allowed price cap LECs on average to achieve the return they would have had if they had been subject to competitive market forces. The study is based on the reported interstate operating revenues, operating expenses, and average net investment of price cap LECs and on an estimate of the elasticity of demand for interstate telephone services. The calculations show the effects on the revenues and operating income of price cap LECs, as well as on the welfare of consumers of telecommunications services, of setting the X-factor at a level producing a competitive return. These calculations assume that the hypothetical X-factor remained in effect from the inception of price caps in 1991 through the end of the period. An adjustment is made for the change in the quantity of output sold as a response to changes in price. In order to examine trends over time in the rate of productivity growth, the study also calculates for individual years the X-factor that would have been necessary to prevent the rate of return from rising above that of the previous year. For comparison with the 1997 and 1999 TFP Studies, calculations were also made assuming no demand response and using data for the RBOCs only, rather than for all price cap LECs. Calculations were performed for 1995, for comparison with the 1997 TFP study, and for 1998, to make use of the most recent data available. The effects of a hypothetical X-factor are estimated by first calculating the change in price, output, and revenue that would have occurred in each year if that X-factor had been in effect. The calculation of these effects for an X-factor of 6.5 percent, the X-factor (including CPD) chosen in the 1997 Price Cap Review Order, is shown in Table C-1. Separate calculations are made for some carriers because not all carriers chose the same X-factor during the period of sharing, so that the effects of moving to a new X-factor differ among carriers. The second column of Table C-1 shows the actual X-factor, taken from Tariff Review Plan ("TRP") data submitted by the carriers. A minimum X-factor of 4.0 percent is used because that level was imposed retroactively by the Commission on carriers that chose lower X- factors. The difference between the actual and hypothetical X-factors is shown in the third column. The fourth column is an index of the cumulative price change over time resulting from the change in the X- factor. The fifth column shows the percentage change in price in each year relative to the actual price in that year resulting from the cumulative effect of the changed X-factor. If the 6.5 percent X-factor had been in effect for the entire price cap period, prices would have been between 10.7 percent and 11.79 percent lower than they actually were in 1998. The last two columns of Table C-1 show the change in output and revenue resulting from a given X-factor change. Because X-factors change on July 1, the beginning of the tariff year, an average of the prices, both actual and hypothetical, in adjacent tariff years is used in the calculation. The percentage change in output was calculated by assuming that a 1 percent reduction in price results in an increase in output purchased of 0.2 percent. This assumes that all reductions in access prices are passed on to interstate end-user customers. The percentage change in revenue is calculated taking account of both the price change and the output change. The effects of changes in the X-factor are calculated separately for each carrier using data on operating revenue, operating expense, and average net investment reported by the carriers on FCC Form 492-A. These data are shown in Table C-2 for 1995 and 1998. Aggregate results are shown in Table C-3. For each carrier, adjusted operating revenue under the hypothetical X-factor is calculated by adjusting actual revenue for the change in price and level of output, as reflected in the final column of Table C-1. Operating expense is adjusted for the change in federal and state taxes, which together are assumed to be 39 percent of revenues. We have no evidence of the effects of an increase in output on costs, but short-run marginal costs are generally believed to be very low. Consequently we assume that costs other than taxes are unchanged by the increase in output, and make no further adjustments in operating expense. Adjusted operating expense is subtracted from adjusted operating revenue to give after-tax profits, which are reported as adjusted operating income. Adjusted operating income and average net investment, which is assumed to be unchanged by the increase in output, are summed for all companies. An average adjusted rate of return is calculated for the industry as the ratio of adjusted operating income to average net investment for all firms in the industry. As shown in the first line of Table C-3, the average rate of return in 1998 if the X-factor of 6.5 percent had been in effect since 1991 would have been 11.88 percent. Thus an X-factor of 6.5 percent for the entire period of price caps would still have allowed carriers on average to earn returns above the target rate in 1998. The benefits to consumers of the rate reduction resulting from the given X-factor are shown by the change in consumer surplus (the final column in Table C-3). The calculation of consumer surplus assumes that IXCs pass on the entire reduction in access prices to end-user customers. The change in consumer surplus consists of two parts: the benefit accruing from the lower price of the amount of output originally purchased, and the benefit accruing from the additional output purchased because of the reduction in price, which would have been forgone had the price remained at the original level. Because of the increase in output resulting from the reduction in price, the increase in consumer welfare is larger than the reduction in LECs' revenues. If the 6.5 percent X-factor had been in effect for the whole period beginning in 1991, benefits to consumers from lower prices in 1998 would have been $2.95 billion. The second and third lines of Table C-3 show the calculation of the X-factor required to produce a competitive rate of return in 1995 and 1998. The background calculations of percentage changes in price, output, and revenue have not been presented because these calculations can easily be replicated. The required rate of return for each year was found by adjusting the rate of return approved by the Commission at the inception of price caps by the basis-point change in bond rates over the period to reflect the action of competitive capital markets. The precise level of the X-factor required to reach the target rate of return in each case was found by trial and error. The estimated X-factor using data through 1995 was 7.10 percent; using data through 1998 the required X-factor was 7.71 percent. For comparison with the TFP studies, the calculations were repeated assuming no demand stimulation and using data only for the RBOCs. The results are shown in the last two lines of Table C-3. These calculations show required X-factors of 6.61 percent in 1995 and 6.97 percent in 1998. Note that these numbers are lower than those using data for all price cap LECs and including a demand adjustment, but somewhat higher than those for the same time periods in the 1999 staff TFP study. The data on which they are based have not been adjusted for accounting biases in the data as have the TFP data. On the other hand, they are based on interstate services only, so they more closely reflect price cap services. This result suggests that more accurate data and assumptions would increase the TFP estimates. Table C-4 shows the X-factor required in each year to maintain the average rate of return at the level of the previous year. This single-year X-factor was calculated for each year from 1992, the first full year of price caps, through 1998. To make this calculation, operating revenues in each year were adjusted to account for the effects of sharing and low-end adjustments. This calculation resulted in an estimate of the revenues that would have resulted from the action of a given X-factor in the absence of sharing and low-end adjustments. Operating expenses were correspondingly adjusted for the change in taxes. In addition, for years through 1994, the actual X-factor in effect for each carrier at the time, rather than the retroactive X-factor imposed later, was used. The X-factor calculation was performed as described above, except that changes in output and revenue were based on the difference between the actual and hypothetical X-factors in a given year rather than on the cumulative price level change from the beginning of the period. As Table C-4 shows, the X-factor required to maintain the rate of return of the previous year trended upward, though not monotonically, from a low of 5.50 percent in 1992 to a high of 8.51 percent in 1998. This appears to reflect an increase in the rate of productivity growth over the period of price caps, and suggests that an X-factor based on an average over the period is likely to underestimate the rate of productivity growth. Table C-1 HISTORIC PRICE, OUTPUT, AND REVENUE CHANGES RESULTING FROM HYPOTHETICAL X-FACTOR Hypothetical X-factor 6.50% End user price elasticity -0.5 Access price elasticity* -0.2 CALENDA R YEAR Actual X-Factor** X-Factor Change Cumulative Price Index Price Change Output Change Revenue Change Ameritech, Bell Atlantic, NYNEX, SBC, GTE, others*** 1991 4.00% 2.50% 0.975 -2.50% 0.25% -1.13% 1992 4.00% 2.50% 0.951 -4.94% 0.74% -3.00% 1993 4.00% 2.50% 0.927 -7.31% 1.23% -4.98% 1994 4.00% 2.50% 0.904 -9.63% 1.69% -6.92% 1995 5.30% 1.20% 0.893 -10.72% 2.03% -8.35% 1996 5.30% 1.20% 0.882 -11.79% 2.25% -9.25% 1997 6.50% 0.00% 0.882 -11.79% 2.36% -9.71% 1998 6.50% 0.00% 0.882 -11.79% 2.36% -9.71% BellSouth 1991 4.00% 2.50% 0.975 -2.50% 0.25% -1.13% 1992 4.30% 2.20% 0.954 -4.65% 0.71% -2.88% 1993 4.00% 2.50% 0.930 -7.03% 1.17% -4.74% 1994 4.00% 2.50% 0.906 -9.35% 1.64% -6.69% 1995 5.30% 1.20% 0.896 -10.44% 1.98% -8.11% 1996 5.30% 1.20% 0.885 -11.52% 2.20% -9.02% 1997 6.50% 0.00% 0.885 -11.52% 2.30% -9.48% 1998 6.50% 0.00% 0.885 -11.52% 2.30% -9.48% Pacific Telesis 1991 4.30% 2.20% 0.978 -2.20% 0.22% -0.99% 1992 4.30% 2.20% 0.956 -4.35% 0.66% -2.64% 1993 4.00% 2.50% 0.933 -6.74% 1.11% -4.50% 1994 4.00% 2.50% 0.909 -9.07% 1.58% -6.45% 1995 5.30% 1.20% 0.898 -10.17% 1.92% -7.88% 1996 5.30% 1.20% 0.888 -11.24% 2.14% -8.79% 1997 6.50% 0.00% 0.888 -11.24% 2.25% -9.25% 1998 6.50% 0.00% 0.888 -11.24% 2.25% -9.25% U S West 1991 4.30% 2.20% 0.978 -2.20% 0.22% -0.99% 1992 4.30% 2.20% 0.956 -4.35% 0.66% -2.64% 1993 4.30% 2.20% 0.935 -6.46% 1.08% -4.38% 1994 4.30% 2.20% 0.915 -8.51% 1.50% -6.10% 1995 5.30% 1.20% 0.904 -9.61% 1.81% -7.41% 1996 5.30% 1.20% 0.893 -10.70% 2.03% -8.33% 1997 6.50% 0.00% 0.893 -10.70% 2.14% -8.79% 1998 6.50% 0.00% 0.893 -10.70% 2.14% -8.79% Sprint 1991 4.00% 2.50% 0.975 -2.50% 0.25% -1.13% 1992 4.00% 2.50% 0.951 -4.94% 0.74% -3.00% 1993 4.00% 2.50% 0.927 -7.31% 1.23% -4.98% 1994 4.07% 2.43% 0.904 -9.56% 1.69% -6.89% 1995 5.30% 1.20% 0.894 -10.65% 2.02% -8.29% 1996 5.30% 1.20% 0.883 -11.72% 2.24% -9.20% 1997 6.50% 0.00% 0.883 -11.72% 2.34% -9.65% 1998 6.50% 0.00% 0.883 -11.72% 2.34% -9.65% Source: See text * Assumes access = 40% of IXC costs, and all price reductions passed on to end user customers. ** In effect 7/1 of each year *** Assumes "others" chose lowest X-factor 1991-1994 Table C-2 FINANCIAL DATA FOR PRICE CAP LECS Price Cap Company Operating Revenue* (000) Operating Expense (000) Average Net Investment (000) 1995 Ameritech $2,314,807 $1,795,638 $3,093,308 Bell Atlantic $6,177,664 $5,121,678 $8,122,916 BellSouth $3,341,690 $2,613,050 $4,618,137 Pacific Telesis $1,742,343 $1,346,448 $2,505,561 SBC $2,091,805 $1,646,644 $3,327,268 U S West $2,416,832 $1,954,164 $3,855,836 GTE $2,729,239 $2,195,426 $4,422,624 Sprint $1,028,598 $778,826 $1,335,617 Others $498,379 $403,046 $728,670 All $22,341,357 $17,854,920 $32,009,937 1998 Ameritech $2,553,594 $1,918,674 $2,794,765 Bell Atlantic $6,453,096 $5,378,333 $8,380,851 BellSouth $3,794,553 $2,842,101 $4,578,390 Pacific Telesis $2,027,231 $1,639,515 $2,645,273 SBC $2,359,902 $2,022,258 $3,407,300 U S West $2,670,048 $2,089,034 $3,513,985 GTE $3,222,880 $2,354,224 $4,432,509 Sprint $1,130,092 $857,222 $1,400,433 Others $939,899 $739,759 $1,241,895 All $25,151,295 $19,841,120 $32,395,401 Source: FCC Form 492-A * Interstate Revenue Table C-3 X-FACTOR REQUIRED FOR COMPETITIVE AGGREGATE RETURN Year X-Factor Since 1991 Actual Operating Revenue* (000) Adjusted Operating Revenue (000) Actual Operating Expense (000) Adjusted Operating Expense** (000) Actual Operating Income (000) Adjusted Operating Income (000) Average Net Investment (000) Actual Rate of Return Competitive Rate of Return Change in Consumer Surplus*** (000) All Price Cap LECs, Demand Elasticity = -0.2 1998 with 1997 X**** 6.50% $25,151,295 $22,753,012 $19,841,120 $18,905,790 $5,310,176 $3,847,222 $32,395,401 16.39% 11.88% $2,947,187 1995 7.10% $22,341,357 $20,051,518 $17,854,920 $16,961,883 $4,486,437 $3,089,635 $32,009,937 14.02% 9.65% $2,810,519 1998 7.71% $25,151,295 $21,053,989 $19,841,120 $18,243,171 $5,310,176 $2,810,819 $32,395,401 16.39% 8.68% $4,979,309 RBOCs Only, No Demand Response 1995 6.61% $18,085,141 $16,208,672 $14,477,622 $13,745,799 $3,607,519 $2,462,873 $25,523,026 14.13% 9.65% $1,876,469 1998 6.97% $19,858,424 $16,948,374 $15,889,915 $14,754,996 $3,968,510 $2,193,379 $25,320,564 15.67% 8.66% $2,910,050 Source: See text. * Interstate revenue. ** Assumes federal + state tax rate = 39 percent. *** Assumes all price reductions passed on to end user customers. **** X-factor chosen in 1997 Price Cap Review Order. Table C-4 X-FACTOR REQUIRED TO MAINTAIN UNCHANGED RATE OF RETURN FROM PREVIOUS YEAR Year X-Factor (%) 1992 5.50 1993 5.94 1994 5.51 1995 6.83 1996 7.90 1997 6.57 1998 8.51 Source: See text.