We use cookies to give you the best experience possible. By continuing we’ll assume you’re on board with our cookie policy

Scale Economies on Wireless Telecommunications Industry Essay

Telecommunications Policy 33 (2009) 29–40 Contents lists available at ScienceDirect Telecommunications Policy URL: www. elsevierbusinessandmanagement. com/locate/telpol Estimating scale economies of the wireless telecommunications industry using EVA data$ Changi Nam a, Youngsun Kwon a,A, Seongcheol Kim b, Hyeongjik Lee c a b c

School of IT Business, Information and Communications University, 119, Munjiro, Yuseong-gu, Daejon 305-732, Republic of Korea Associate Professor, School of Journalism and Mass Communication, Korea University, 5-1, Anam-dong, Seongbuk-gu, Seoul, 136-701, Republic of Korea Full-time instructor, Department of Management Science, Republic of Korea Naval Academy, 88-1 Angok-dong, Jinhae, Kyungnam, 645-797, Republic of Korea a r t i c l e in fo abstract This paper proposes a new estimation method of total cost and average cost curves and applies it to the telecommunications industry.

The method is more ? exible and entails less hassle for data collection than traditional methods. The results show that the longrun average cost (LRAC) curve is downward sloping, revealing the presence of economies of scale in production. The two largest Korean mobile network operators are realizing full economies of scale, while the smallest operator is not. Finally, the paper recommends three policy alternatives that the Ministry of Information and Communication of Korea can draw on to increase ef? ciency in the Korean mobile telecommunications market. & 2008 Elsevier Ltd. All rights reserved.

We will write a custom essay sample on Scale Economies on Wireless Telecommunications Industry specifically for you
for only $16.38 $13.9/page

Order now

Keywords: Scale economy estimation Economic value-added (EVA) Telecommunications Asymmetric regulation 1. Introduction Empirically estimating the long-run average cost (LRAC) curve and the minimum ef? cient scale (MES) of production has been an important research topic in the ? eld of regulatory economics because data relating to both are critical for measuring production ef? ciency in industries, especially in public utility variants. One major problem in estimating the LRAC curve and the MES has been obtaining the appropriate data, especially data related to production factors other than capital and labor.

Nowadays, ? rms are capitalizing more than ever before on the bene? ts of intangible assets, such as computer and management software, brand, and intellectual property rights. 1 The traditional estimation method for the LRAC curve and the MES focusing on traditional production factors in manufacturing industries has been of decreasing validity and usefulness in the information age. According to Brynjolfsson and Hitt (2000, 2003), information and communication technology increases the productivity of ? rms by complementing organizational capital and streamlining business processes.

Because of the increasing importance of intangible assets, researchers who pay attention only to traditional production factors are likely to omit a proportion of input cost items. Therefore, this paper develops a new method of estimating the LRAC curve, free from changes in the composition of production factors. The purpose of this paper is to propose a new model to estimate the LRAC curve of ? rms, and to apply it to Korean mobile network operators (MNOs). This paper estimates the LRAC curve of production using annual sales data, estimated economic value-added (EVA) data, and annual This research was supported by the Ministry of Information and Communication (MIC), Republic of Korea, under the Information Technology Research Center (ITRC) support program supervised by the Institute of Information Technology Assessment (IITA) ‘‘(IITA-2006-C1090-0603-0041)’’ A Corresponding author. Quello Center, Michigan State University, MI, USA. Tel. : +1 517 803 0497; fax: +1 517432 8065. E-mail addresses: [email protected] ac. kr (C. Nam), [email protected] ac. kr (Y. Kwon), [email protected] ac. kr (S. Kim), [email protected] ac. kr (H. Lee). 1 Refer to Whitwell, Lukas, and Hill (2007) on the growing importance of intangible assets. 308-5961/$ – see front matter & 2008 Elsevier Ltd. All rights reserved. doi:10. 1016/j. telpol. 2008. 10. 005 ARTICLE IN PRESS 30 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 production. In other words, this paper is not utilizing traditional structural models derived from the cost and production functions, but devises a new model. When this model was applied to a Korean MNO, SK Telecom, it was found that the LRAC curve is downward sloping as expected and asymmetric regulation is a source of the pro? t to the dominant player in the market. One limitation in using the model is that it can only be applied to ? ms listed on the stock market. This paper is organized as follows. The Section 2 brie? y reviews traditional approaches to estimating LRAC curves used in previous research and discusses their limitations. In Section 3, a new model is proposed and compared with traditional models to identify the strengths and weaknesses of the model. Section 4 introduces data and output measurement issues. Section 5 presents the results of the estimation and Section 6 discusses the policy implications of the results for asymmetric regulation that have been used in the Korean telecommunications industry since 2000. Section 7 concludes the paper. . Traditional approaches of estimating the LRAC curve The main reason for estimating the LRAC curve is to ? nd out the magnitude of scale economies and the MES. This information is necessary and useful for regulators and business managers. Using this information, regulators can deduce the appropriate market structure of an industry and managers can calculate the optimal size of their business in terms of production ef? ciency. Unsurprisingly, managers can also use the information for such strategic purposes as deterring market entry by competitors and consolidating pro? ts. As Huettner (1973, pp. 27–429) summarized, previous studies have used one of two methods: estimating the LRAC curve or the short-run average cost curve or directly predicting the level of scale economies by estimating the production function. These two approaches are intrinsically the same because the model used in the cost function approach is also derived from the production function and budget constraints. 2 In other words, if the production technology has an increasing-returns-to-scale property, the estimated average cost function should be downward sloping either locally or globally when factor prices are stable or constant.

The cost function approach, estimating the LRAC and its curvature or estimating the LRAC and long-run marginal cost (LRMC) to calculate the level of scale economies, has been widely used in ? research. 3 Two recent studies drawing on the cost function approach are Asai (2006) and Lorincz (2006). 4 Huettner (1973) pointed out that one critical problem in measuring scale economies is obtaining appropriate data on cost variables and on the quantities of output and factor inputs. In addition, existing empirical papers suffer from measurement problems.

Some papers used only variable costs, ignoring capital costs, and others ignored the age of capital facilities. Recently, as information technology advances and competition becomes intense worldwide because of globalization, ? rms have tended to increase their investment in intangible assets such as software, brand, product design, and marketing in order to reduce production costs. This trend seems to be eroding the robustness of traditional approaches of estimating scale economies. Not taking into account the costs of invisible assets is likely to distort the estimation results of scale economies even though the direction of distortion, i. . , under- or overestimation of scale economies, is not yet well known. 3. The model This paper proposes a new method of deriving the LRAC, which is calculated by subtracting economic pro? t (EP), whose equivalent term is EVA in management, from total revenue (total sales). By de? nition, EP is equal to total revenue less total cost. Therefore, if EP is equivalent to EVA, and EVA can be somehow estimated, total cost can also be estimated. Then, the LRAC can be estimated by dividing the estimated total cost by quantity sold in each year.

In short, this paper adopts a reversed approach to deriving the LRAC curve compared with the traditional approaches that ? rst obtain the costs of major factor inputs and secondly sum them to obtain total average costs. The reversed approach reduces the hassles of obtaining data and avoids the issue of missing cost variables. Total revenue (TR) is easily observable, but total cost (TC) is not. The model starts from the idea that TC can be estimated as long as EVA is estimable as shown in the following equation: EVA ? TR A TC (1) Regarding Eq. (1), two points are worth mentioning. Eq. 1) is based on the assumption that EVA is the same as EP, or at least the assumption that EVA is an unbiased estimate of EP. EP is the residual remaining after the opportunity costs of all inputs are subtracted from TR. By the same token, EVA is the residual a ? rm retains after compensating for all explicit and implicit factor costs. In other words, the two are conceptually equivalent. The second point of discussion is whether the estimated EVA is comparable to the estimated EP. Conceptually, they are the same, but EP is TR less TC, where TC is the cost of producing the optimal quantity of production, which in turn is determined by a ? m in an effort to minimize production costs given that the prices of the ? nal product and inputs are ? xed. The point is that TC, used in estimating EP, is the 2 3 4 Refer to Fare and Logan (1983) for duality between cost and production functions. See Considine (1999) for how to use average cost and marginal cost concepts to understand the level of scale economies. Two more recent empirical papers are Sung and Gort (2000) and Katrishen and Scordis (1998). ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 31 minimum cost of production evaluated at the optimal factor proportion of inputs.

In other words, if the factor proportion resulting in maximum EP is different from that for maximum EVA, even though EVA and EP are conceptually the same, their estimated values are likely to be different. However, if the assumption that a ? rm is minimizing costs in business holds, EVA should be equal to EP. It would be fair to say that EP is the lower bound of EVA. In brief, using EVA as a surrogate of EP requires justi? cation. As explained later, estimating EVA in the model requires that ? rms be listed on the stock market. If ? rms are so listed, it is not an unreasonable assumption that they are pursuing maximization of ? m value, which in turn implies that they are trying to maximize their pro? t. Pro? t maximization requires cost minimization. Therefore, the EVA of a listed ? rm can be seen as a fair approximation of its pro? t. Finally, EVA can be considered an unbiased estimate of EP and be used as a proxy of EP. Because the equivalence of EVA and EP are proven at least intuitively, the next step is to estimate EVA. EVA ? NOPAT A WACC A IC (2) EVA is usually measured in the literature as the difference between a ? rm’s net operating pro? t after taxes (NOPAT) and its total cost of capital, which is the ? m’s invested capital (IC) times the weighted-average cost of capital (WACC), as shown in Eq. (2). 5 However, it is dif? cult to estimate EVA directly in practice because various accounting adjustments are required for calculating the ? rm’s NOPAT and IC. Lovata and Costigan (2002) point out that more than 100 possible adjustments are needed to estimate EVA directly. Therefore, estimating EVA with publicly available accounting data does not seem to be an ef? cient and objective way to estimate EVA. This paper, instead, attempts to estimate EVA indirectly, using the functional relationship between the EVA and market value-added (MVA) of a ? m. MVA ? EVA WACC (3) Conceptually, MVA is the present value of the stream of future EVAs of a ? rm. Therefore, although previous empirical studies such as Kramer and Pushner (1997) did not ? nd a strong relationship between EVA and MVA, the two measures of business performance should be closely correlated with each other. As Young (1997) mentioned, if a ? rm is expected to achieve the same magnitude of EVA every year, the ? rm’s MVA becomes the present value of its expected future EVAs as shown in Eq. (3). In addition, MVA is equal to the market value (MV) of a ? rm less the value of IC, as shown below MVA ?

MV A IC Then, from Eqs. (3) and (4), Eq. (5) is derived. This paper draws on Eq. (5) to estimate EVA. EVA ? WACC A ? MV A IC? (5) (4) If a ? rm is listed on the stock market, MV can be estimated by multiplying the stock price of a ? rm and the number of stocks outstanding, and then adding it to the book value of its debt. 6 IC also can be obtained from the current balance sheet of the company, although some asset values are based on historical data. WACC can also be calculated using both the stock price and ? nancial statements data. This paper calculates the data for MV, IC, and WACC, and then estimates EVA and TC, respectively.

Compared with the traditional methods, the model has several strengths as well as limitations. First, it subsumes all the cost items, tangible or intangible. As Weaver (2001) discussed, invisible assets such as R&D and advertising are not expensed but capitalized in the process of calculating EVA. Secondly, the estimation method in this paper might be more appropriate than traditional methods in estimating a ? rm’s total cost and the LRAC curve because the EVA estimated by the new method not only emphasizes a historical accounting performance, but also re? ects the future cash ? ow of the ? rm in the long run. Because a ? m’s total cost, obtained by subtracting the estimated EVA from its total revenue, represents the present value of future expected costs, the estimated LRAC curve is expected to overcome the shortcoming of the traditional approach that input prices and technology are assumed to be ? xed in the long run. 7 Thirdly, the reliability of the estimated EVA is also expected to be improved because the proposed method does not need to estimate the ? rm’s NOPAT with any adjustments. One limitation of the model is that EVA is derived from Eq. (3), based on the rather strong assumption that the current EVA continues into the future.

Another limitation is that the model is applicable only to ? rms whose stocks are traded publicly in the market because data for MV cannot otherwise be obtained easily. 5 6 7 Refer to Kramer and Pushner (1997) and Weaver (2001) for a detailed discussion of EVA concept. The book value of debt is added to the value of the ? rm’s equity, based on the assumption that the ? rm holds debts until maturity. See Huettner (1973) for a discussion on the shortcomings of the traditional methods. ARTICLE IN PRESS 32 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 4. Data and output measure 4. 1.

Data The key variables for the estimation of EVA are MV, IC, and WACC, as shown in Eq. (5). The calculations of these three variables and data sources are summarized in Appendix A. Financial statements of the Korean MNOs are obtained from the Korean Financial Supervisory Service,8 stock price data were taken from the Korea Exchange,9 and the annual number of subscribers is from the Korean Ministry of Information and Communication. 10 The average stock price for December is used in calculating the equity value of a year, and then the equity value data of 2 consecutive years are used for the calculation of the average equity value of a year.

The average stock price of December is used instead of the end of year stock price to use a more representative stock price of the year. The values of ? rms’ debts, IC, and WACC are calculated following the standard methods used in ? nance textbooks. Finally, the TC variable derived from Eq. (1) is de? ated by the gross domestic product (GDP) de? ator of the telecommunications industry in order to re? ect the decreasing equipment costs of the telecommunications sector. The GDP de? ator of the telecommunications sector is obtained from the Bank of Korea website. 4. 2.

Output measure Obtaining an appropriate output measure is a prerequisite for calculating the LRAC, especially when the industry concerned is producing multiple heterogeneous products. As Carlton and Perloff (2000) point out in their textbook, if a ? rm is producing multiple products such as oranges and apples, then deriving the average cost of production becomes problematic. In previous work, such as Nemoto and Asai (2002), the quantity produced is derived from the TR variable. TR, by de? nition, is simply the product prices times the quantities sold in the market.

Therefore, the quantity can be derived by dividing TR by the appropriate price index. However, this approach is subject to a few critical problems. First, if heterogeneous products are produced by a ? rm, it is dif? cult to interpret the quantity variable derived from TR because the unit of the quantity variable cannot be determined. In addition, the quantity variable derived from TR has problems if the composition of the product mix changes from year to year. Secondly, ? nding an appropriate price index for the quantity measure of multiple products is very dif? ult, especially when there are a large number of products. This paper uses a direct measure of quantity supplied, a publicly available output measure in the telecommunications industry—the number of subscribers. It is well known that traf? c is a major cost driver in the telecommunications industry. Therefore, using traf? c volumes as an output measure seems appropriate. Nowadays, however, almost all telecommunications ? rms deliver voice as well as data services in the market and different metrics are used for voice and data traf? c: call minutes for the former and bytes of data transferred for the latter. 1 It is obvious that conceptually two metrics of different units cannot be merged into one. In the telecommunications industry, the number of subscribers has been used as an output measure in addition to traf? c volume. The number of subscribers can de? nitely be considered a major cost driver because they are generating voice and data traf? c. Therefore, this paper uses the number of subscribers, N, rather than traf? c as the output measure. TC is a function of voice and data traf? c, as presented in the following equation: TC ? f ? t v ; t d ; w; T? (6) where tv and td are voice and data traf? respectively; w is the input price vector; and T is the technology characteristic vector. Because tv and td are increasing functions of the subscribers, Eq. (6) can be rewritten as TC ? f ? t v ? N? ; t d ? N? ; w; T? ? g? N; w; T? LRAC ? TC? N; w; T? N (7) (8) This paper ? nally calculates LRAC using Eq. (8). This way of measuring the LRAC is especially appropriate for present purposes because time series data spanning 16 years are used, for which output composition has changed from solely voice services to voice and data services in the telecommunications industry. 5.

Empirical results: an application to the Korean telecommunications industry The model is applied to the Korean wireless communication market, where three MNOs—SK Telecom (hereafter SKT), KTF and LG Telecom (hereafter LGT)—are competing. At the end of 2006, the three MNOs’ market shares were 50. 4%, 32. 2%, 8 9 http://www. fss. or. kr http://www. krx. co. kr http://www. mic. go. kr 11 See Kelly and Woodall (2000) for trends in traf? c data. 10 ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 33 12000 10000 TR TR (billion Won) 8000 6000 Subscribers 5 20 Subscribers (million) 15 10 4000 2000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 5 0 Year Fig. 1. SKT’s TR (left Y-axis) and subscribers (right Y-axis). and 17. 4%, respectively. The LRAC and LRMC curves are derived for only SKT, the largest MNO in Korea, primarily because of data availability. SKT’s stocks began to be traded on the Korean Exchange from 1989, whereas KTF and LGT were publicly listed on the stock market in 1999 and 2000, respectively. Therefore, there are scant available data for EVA estimation for KTF and LGT.

The model, however, is also applied to them in order to compare the TC and LRAC of the three MNOs. 5. 1. Estimation of the scale economies This section estimates the TC of SKT, whose ? nancial data are available since 1990 and derives LRAC and LRMC curves from the estimated TC function. Since 1990, SKT’s TR has grown continuously, with the number of subscribers increasing from 88,000 in 1990 to 19. 5 million in 2005, as shown in Fig. 1. Following the process shown in Appendix A, this paper estimates the MV, IC, WACC, and EVA of SKT from 1990 to 2005, as shown in Table 1.

For the whole estimation period, the EVA of SKT has been positive and growing in size up to a maximum of about 1. 5 trillion Korean Won in 2001. The estimated EVA of SKT has decreased since 2001 because of increased IC, mainly caused by new investments for network upgrades to accommodate wireless Internet services, and market saturation, but returned to nearly 1 trillion Won in 2005. The WACC of SKT fell continuously until 1997 and since then has slowly been returning to its 1990s level. By de? nition, WACC is the weighted average of the cost of equity and cost of debt, so the trend of SKT’s WACC re? cts the fall in the risk-free interest rates of the early1990s and the rise in the cost of equity and debt in the latter part of the decade. The TC of SKT, estimated by subtracting the estimated EVA from TR, is de? ated by the GDP de? ator of the telecommunications industry. Fig. 2 presents the trend of TC in nominal and real value terms. According to Fig. 2, TC has increased continuously, keeping pace with the number of subscribers, except for 1999 and 2001 when SKT recorded exceptionally high values of EVA. TC is unlikely to decrease in the real business world especially when a ? m grows, and the idiosyncratic events of 1999 and 2001 stem from the TC estimation method of this paper. EVA re? ects changes in the stock price, which in turn re? ect investors’ expectations on SKT’s future pro? ts. Therefore, when SKT commanded exceptionally good business prospects as it did in 1999 and 2001, TC would be biased downward temporarily because of an overestimated expected EVA. From the 1990s to the present, the GDP de? ator of the telecommunications industry has fallen consistently, so the real TC curve rotates counter-clockwise compared with the nominal TC curve.

Fig. 3 presents the real TC curve, which shows the relationship between the real total costs and the subscriber variable (the quantity produced by the ? rm). The smoothly curved TC curve, ? tted using the ordinary least-squares method to estimated total costs, is shown in Eq. (9) and also presented in Fig. 3. TC ? 2:8147N3 A 57:262N2 ? 653:32N (9) The ? tted TC curve looks like the typical TC curve, found in microeconomics textbooks, i. e. , TC initially rises at a decreasing rate until it reaches the in? exion point and, after that, at an increasing rate.

In deriving the TC curve, the intercept is set to zero because TC is a long-run concept in this paper. ARTICLE IN PRESS 34 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 Table 1 Trends of major variables (billion Korea Won). Year 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 MV (A) 238 467 859 1558 3293 4705 4667 5137 6512 28,430 26,433 26,823 27,783 23,416 23,253 22,064 IC (B) 53 83 92 166 294 631 1242 1751 2374 2999 3399 4099 5898 8231 9526 9868 MVA (C ? AAB) 185 383 768 1392 2999 4074 3425 3386 4139 25,431 23,034 22,724 21,885 15,184 13,726 12,196 WACC (D) (%) 10. 4 8. 49 7. 75 7. 15 5. 74 5. 46 5. 32 3. 28 4. 63 5. 47 4. 05 6. 51 5. 32 5. 22 4. 64 8. 08 EVA (E ? C A D) 19 33 59 99 172 222 182 111 191 1392 932 1479 1164 793 637 985 14000 12000 10000 8000 6000 4000 2000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 TC (billion Won) TC (real value) TC (current value) Year Fig. 2. Total production cost of SKT. The LRAC curve is obtained by dividing the TC equation by N and LRMC curve by taking the derivative of the TC equation with respect to N. The LRAC and LRMC curves derived from the TC curve are drawn in Fig. 4.

From Fig. 4, it is easy to identify the MES, the size of service production that minimizes LRAC. The MES of SKT turns out to be about 10 million subscribers, half its current size. The MES derived from the nominal TC is about 13 million subscribers, which is greater than that derived from real values. 12 The scale economies index, (LRAC–LRMC)/LRAC, is calculated and presented in Fig. 5. 13 The scale economies index can be greater than, less than, or equal to zero, when returns to scale increase, decrease, or are constant. The index is maximized when the subscriber number is about 6 million.

From Figs. 4 and 5, it can be easily noted that the gap between LRAC and LRMC begins to widen rapidly after the number of subscribers exceeds 10 million. However, based on this result it cannot be said that decreasing returns to scale exist and intensify as the number of subscribers rise above 10 million. This is because the rising portion of the LRMC curve above the minimum LRAC is likely to be caused by the large investment in network upgrade needed for new data services, while the number of subscribers is stagnant and the revenue from the data service is not robust. 4 Considering this, it is fair to say that the recent LRAC and LRMC of SKT are overestimated. The equation of the nominal TC curve is not presented here but is available upon request from the corresponding author. Willig (1979) used a different but equivalent measure of scale economies. The share of revenue from wireless Internet services in monthly average revenue per user was only 1% in January 2000, but it increased to 27% in December 2005, which indirectly reveals the investment in the wireless Internet service made by SKT. 13 14 12 ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 35 4000 12000 10000 TC (billion Won) 8000 TC (real value) 6000 4000 2000 0 0 2 4 6 8 10 12 14 16 18 20 Subscribers (million) Fig. 3. Fitted TC curve of SKT. TC = 2. 8147N3 – 57. 262N2 + 653. 32N R2 = 0. 987 2000 1800 1600 AC/MC (thousand Won) 1400 1200 1000 800 600 400 200 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Subscribers (million) LRAC LRMC Fig. 4. AC and MC curve of SKT. 5. 2. Estimation of cost curves of other MNOs in Korea The same estimation method of TC is applied to KTF and LGT, and Fig. 6 compares their TCs with that of SKT. Even though data from KTF and LGT are scant, some ? dings are worth attention. First, when the number of subscribers is around ? ve million, the estimated total costs of the three MNOs are roughly the same; as the number of subscribers rises, the total cost curves begin to diverge. Considering that SKT’s estimated TC was exceptionally low because of a high expected EVA when it reached 10 million subscribers, KTF’s TC curve could be viewed as SKT’s hypothetical TC curve without any overexpectation on SKT’s business performance on the stock market. Secondly, LGT and KTF use a 1. 8 GHz spectrum, while SKT uses an 800 MHz spectrum.

It is frequently argued that the difference in spectrum bands results in a difference in costs, (mainly in investment costs), which is in turn caused by the difference in the cell coverage of wireless communication. However, it is also argued that in metropolitan areas, where the majority of subscribers are living, the cost variation owing to spectrum band differences is almost zero and, as a result, the overall cost difference between SKT and the other two ? rms should be insigni? cant. Fig. 6 seems to support the latter view, i. e. , the proposition that the cost difference among MNOs is not signi? ant, controlling for the number of subscribers. Three points regarding this are important. First, there is no reason for LGT, using the adjacent spectrum with KTF, to have a different cost function, given the assumption that the two ? rms adopt ef? cient technologies and management methods. Secondly, because SKT has invested in its network upgrades for wireless Internet service during the past few years, the ARTICLE IN PRESS 36 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 0. 5 0. 0 1 Scale Economies Index -0. 5 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 -1. 0 -1. 5 -2. Subscribers (million) Fig. 5. Scale economy index. 14000 12000 10000 TC (billion Won) 8000 6000 4000 2000 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Subscribers (million) KTF LGT SKT Fig. 6. Estimated TCs of Korean MNOs. other two ? rms followed the same path even though they are underdogs in the market. 15 This can be observed in Fig. 6. In recent years, the three MNOs’ TC curves exhibit the same pattern, digressing from the previous trend. Thirdly, LGT’s TC curve appears to be an upward sloping straight line with a very steep slope, which implies that LGT does not bene? from scale economies. This TC curve suggests that LGT invested or was forced to invest, because of competition with other MNOs, in building a wireless data network before realizing economies of scale in the voice communications market. Fourthly, subscribers to KTF have increased from 5. 3 million in 2000 to 12. 3 million in 2005; they have exceeded 10 million, the MES derived from SKT’s LRAC curve, since 2002. Therefore, even though there are only six sample data points for estimation of the TC curve, the ? tted TC curve is derived and appears in Eq. (10) with R2 ? 0. 96. TC ? :2718N3 A 106:3N 2 ? 830:1N (10) The basic shape of KTF’s TC curve is quite similar to SKT’s. From Eq. (10), the LRAC curve of KTF is drawn and juxtaposed with that of SKT in Fig. 7. The MES, calculated from KTF’s LRAC curve, is about 7. 3 million, smaller than the MES derived from SKT’s LRAC curve. However, considering that KTF has invested in network upgrades for its data service since 2001, it is highly likely that the MES of KTF is underestimated. LGT’s subscriber number was about 6. 5 million at the end of 2005, which is smaller than either of the MESs and does not appear to be an ef? ient scale. 15 According to our estimation, the ICs of SKT, KTF, and LGT increased by 174%, 152%, and 72%, respectively, between 2000 and 2005. ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 37 1800 1600 LRAC_KTF 1400 AC (thousand Won) 1200 1000 800 600 400 200 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Subscribers (million) LRAC_SKT Fig. 7. AC curves of SKT and KTF. According to Fig. 7, SKT looks more ef? cient than KTF over the whole range of domain denoted by the number of subscribers.

Above all, the available data for the estimation of KTF’s TC curve is too small to derive a general conclusion. However, the differences in spectrum ef? ciency, economies of scale, and economies of scope can be designated as potential contributors that result in differences in the LRAC curves, even though the reasons this occurs cannot be determined exactly. 16 The estimated cost curves will change as the customer base grows and new services such as mobile banking, digital media broadcasting, and wireless Internet connection begin to earn pro? s, so additional time and data are needed to draw a more de? nite conclusion. 6. Policy implications 6. 1. Asymmetric regulation in Korean telecommunication market In the Korean telecommunications market, wireless telephony service had been provided solely by SKT since 1984. Competition was introduced in the mid 1990s after four MNOs using cellular and personal communications service (PCS) technologies launched their mobile telephony services in 1995 and in 1996, respectively. However, the competition among ? e MNOs existed only for 4 years because two MNOs merged with two other MNOs (SKT and KTF) in 2000. The mergers of 2000 brought forth current market structure and since then the market shares of three MNOs have rarely changed as shown in Fig. 8. The Ministry of Information and Communication (MIC) of Korea has implemented an asymmetric regulation policy since 2000 to help late market entrants such as KTF and LGT compete with the incumbent, SKT, in the market, based on the belief that economies of scale and the network externality effect hinder fair competition. 7 Examples of asymmetric regulation policy implemented in Korea are regulating interconnection charges based on the individual ? rm’s network cost rather than on industry average network cost, regulating only the rate of SKT voice service, and applying number portability policy to MNOs sequentially, not concurrently, with 6 months time lags among them from LGT to SKT. Asymmetric regulation prohibited price competition among MNOs in Korea’s mobile phone service market and instead resulted in non-price competition through handset subsidies and various membership card promotions. 18 6. 2.

Implications for policy makers The MIC is arguing that asymmetric regulation is necessary for promoting effective competition in the Korean mobile telephone service market despite the criticism that it has done more harm than good, especially in terms of subscriber welfare. SKT has been making considerable positive (excess) economic pro? ts for more than a decade, which indicates that it has plenty of room for maneuver. The MIC continues to worry that price competition would drive LGT out of the market, which would eventually harm consumers by making it easier for the remaining two operators to collude in price setting. 6 In the Korean metropolitan area, the coverage area of a cell of an 800 MHz spectrum is 6. 36 km2 while that of a 1. 8 GHz spectrum is 2 km2. The number of base stations of KTF (7051) is higher than that of SKT (5243) according to Kim (2006). 17 Refer to Perrucci and Cimatoribus (1997) for general discussions on asymmetric regulation and Peitz (2005a, b) for recent theoretical studies on asymmetric regulation. 18 See Park, Kwon, and Park (2005) for a comparison of price reduction and membership card promotion in Korea. ARTICLE IN PRESS 38 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 60. 0% 2001. 2 50. 0% 40. 0% 30. 0% 20. 0% 10. 0% 0. 0% SKT KTF LGT Fig. 8. Changes in market shares of Korean MNOs. 2007. 7 The authors’ calculation shows that LGT is currently making a small amount of economic pro? t and the ? nancial data from an accounting ? rm reveal that LGT incurred economic losses between 2002 and 2003. That is, LGT is surviving on the margin; thus price competition could drive it out of the market as suggested by the MIC. From an alternative perspective, however, prohibited price competition can be seen as playing the role of guaranteeing positive pro? ts for SKT and KTF and the survival of LGT. 9 Focusing on the number of players in the mobile telephone service market, the MIC has been ignoring consumer welfare for more than six years. Asymmetric regulations have not produced effective competition in the Korean mobile telecommunication industry so far as illustrated by Fig. 8. Based on the analysis of this paper, three policy recommendations regarding asymmetric regulation in Korea are addressed below. The ? rst policy option is to abolish asymmetric regulation and simply let the market work. However, abolishing asymmetric regulation does not mean removing rate regulation altogether. The MIC can egulate the rate of the dominant player in the market by adopting price cap regulation. 20 This option could result in two players in the Korean mobile telephone service market after a period of ? erce price competition as discussed above. The removal of asymmetric regulation might not produce spontaneous price competition if SKT and KTF want to make economic pro? ts. If this is the case, the MIC can make the regulated rate fall gradually until the EVA of the dominant player falls to an appropriate level for the sake of consumer welfare. The second option is to reshuf? e the current market structure by intensifying asymmetric regulation.

Such regulation does not automatically allow LGT to expand its market share and enjoy economies of scale effects. Therefore, the success of this policy suggestion will depend on how effectively the MIC reshuf? es the market structure by using policy measures. According to the present analysis, the MES in the Korean mobile telephone service industry is located at between 10 million and 13 million subscribers, and derived from nominal values of TC. At the end of 2006, the total number of subscribers to mobile telephone services in Korea was about 40 million, which is large enough for three MNOs to attain the MES.

A balanced market share among three MNOs in Korea would result in a win–win outcome for producers and consumers even though the dominant ? rm obviously would not be happy with the outcome. The key problem is how to implement it in a free market economy, considering that notwithstanding asymmetric regulation by the MIC since 2000, the market share among the three did not change signi? cantly until recently. The policy that forces the dominant MNO to resell wireless services (airtime) to competitors (especially to LGT) at wholesale rates seems to be a feasible way to attain a more balanced market structure. 1 This means that LGT becomes a reseller of mobile airtime compared with being a traditional facilitybased network service provider. The third option is to induce MNOs that cannot attain the MES to turn from facility to service-based competitors, i. e. , becoming mobile virtual network operators (MVNOs). This policy would increase ef? ciency in production as well as in consumption by inducing those that cannot attain the MES to drop off from facility-based competition and boost competition in retail market.

In addition, the MIC needs to boost price competition in Korean mobile telecommunications market to motivate LGT to get out of facility-based competition and to enter into service competition. 7. Conclusion This paper proposes a new method of estimating scale economies by deriving TC from TR using estimated EVA. This method overcomes the problem of using historical accounting data and uses readily available ? rm-level data. One major 19 20 21 One reviewer also mentioned that the survival of LGT could stem from the asymmetric regulation.

The UK is using this method of price regulation (Ofcom, 2003). KT, the dominant wired network operator in Korea, has been reselling airtime at retail prices by purchasing it at wholesale rates from KTF. ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 39 drawback is that this method is only applicable to ? rms whose stocks are traded on the stock exchange. The new method is applied to the Korean MNOs. The empirical results show that the MES lies between 10 million and 13 million subscribers, although the result cannot be generalized because of a lack of data.

The estimated MES implies that production ef? ciency and consumer welfare might not fall without asymmetric regulation even though only two MNOs exist in the Korean mobile telephone service market. Considering the size of the estimated MES, three MNOs can coexist without lowering production ef? ciency and consumer welfare if the market share is rearranged in a more balanced way. Available policy options to the MIC would either abolish asymmetric regulation, which has not been effective in changing market structure in the past, or reinforce it to increase LGT’s market share.

While this paper presents some meaningful implications, it is not without limitations. The estimation method of predicting scale economies is based on several strong assumptions as discussed above, even though it is believed that relaxing these assumptions would not signi? cantly change the results of the analysis. These assumptions should be relaxed to improve the generality of the analysis. Secondly, the results of the authors’ estimation and the implications drawn from the empirical results might be limited in applicability mainly because of the scarcity of samples.

Therefore, a useful area of future research would be to extend the empirical analysis to other parts of the telecommunications industry or to an international context. Appendix A. The estimation process of EVA for the Korean MNOs in year X See Table A1. Table A1 Items EVAX MVX EVX PX CSX ATDX AICX ICX TAX NIB_CLX APX OAPX AEX OAFCX IMMX URX UIX FDX OCLX NIB_LLX LURX LOAPX LUIX OLPX OLLX NOAX LAPX LFIX ISX ISSRX DDCTX BAX AX WACCX Meanings Economic value-added (economic pro? ) Market value Market value of equity The daily average closing price during Decembera The number of issued common stocks Average total debtsb Average invested capital Invested capital The total assets Non-interest-bearing current liabilities Accounts payable Other accounts payable (including accrued dividends and tax payable) Accrued expenses Other advance from customers Import margin money Unearned revenue Unearned income Financial derivatives Other current liabilities Non-interest-bearing long-term liabilities Long-term unearned revenue Long-term other accounts payable Long-term unearned income OTHER liability provisions Other long-term liabilities Non-operating assets Loans to af? liated companies Long-term ? ancial instruments Investment securities Investment securities of person with a special relationship Differences of deferred corporate taxes Building account Allowanced Cost of capital Calculation (source) (MVXAAICX) A WACCX EVX ? ATDX PX A CSX (Korea Exchange) (Korean Financial Supervisory Service) TDX ? TDXA1 2 ICX ? ICXA1 2 TAX A NIB_CLX A NIB_LLX A NOAX (Korean Financial Supervisory Service) APX ? OAPX ? AEX ? OAFCX ? IMMX ? URX ? UIX ? FDX ? OCLX (Korean Financial Supervisory Service) (A. 2) (A. 1) LURX ? LOAPX ? LUIX ? OLPX ? OLLX (Korean Financial Supervisory Service) LAPX ? LFIX ? ISX ? ISSRX ? DDCTX ? BAX ? AX (Korean Financial Supervisory Service) AIBLX ATEX A CODX ? 1 A TX ? ? A COEX ATAX ATAX (A. 3) ARTICLE IN PRESS 40 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40

Table A1 (continued ) Items ATAX ATEX AIBLX IBLX CODX IEX COEX RFX Meanings Average total assets Average total equity Average total interest-bearing liabilities Total interest-bearing liabilities Cost of debt Interest expenses Cost of equity Annual rate of national housing bond type 1 Beta Risk premium Annual average market return for the last ten years Corporate tax rate Calculation (source) TAX ? TAXA1 2 TEX ? TEXA1 2 IBLX ? IBLXA1 2 TAXANIB_CLXANIB_LLXc IEX AIBLX (Korean Financial Supervisory Service) RFX ? bX A RPX (Bank of Korea) (Korea Exchange)d RFX A MRX (Korea Exchange) (Korean Financial Supervisory Service)e bX RPX MRX TX a The last closing price is ideal but inappropriate for estimating the market value because of its volatility.

We also attempted to estimate the market value using the last closing price and found that there is almost no difference compared with the results when using the daily average closing price during December. b Because the market value of the debt is not publicly available, the book value is used instead. c It is easily calculated by estimating IC. d This is obtained from the regression between a market return and a rate of return of the MNO, from the listed day to the last day of year X. e The corporate tax rate is constant at 20. 7% in Korea. References Asai, S. (2006). Scale economies and scope economies in the Japanese broadcasting market. Information Economics and Policy, 18(3), 321–331. Brynjolfsson, E. , & Hitt, L. (2000). Beyond computation: Information technology, organizational transformation and business performance.

Journal of Economic Perspectives, 14(4), 23–48. Brynjolfsson, E. , & Hitt, L. (2003). Computing productivity: Firm-level evidence. Review of Economics and Statistics, 85(4), 793–808. Carlton, D. W. , & Perloff, J. M. (2000). Modern industrial organization (3rd ed). New York: Addison Wesley Longman. Considine, T. J. (1999). Economies of scale and asset values in power production. Electricity Journal, 12(10), 37–42. Fare, R. , & Logan, J. (1983). The rate-of-return regulated ? rm: Cost and production duality. Bell Journal of Economics, 14(2), 405–414. Huettner, D. A. (1973). Shifts of long run average cost curves: Theoretical and managerial implications.

Omega, 1(4), 421–450. Katrishen, F. , & Scordis, N. A. (1998). Economies of scale in services: A study of multinational insurers. Journal of International Business Studies, 29(2), 305–323. Kelly, T. , & Woodall, M. (2000). Telecom traf? c indicators. Telecommunications Policy, 24(2), 155–159. Kim, Y. (2006). The impact of cellular-spectrum expansion on the structure and performance of mobile telecom market. Korea Telecommunications Policy Review, 13(1), 1–25 (in Korean). Kramer, J. K. , & Pushner, G. (1997). An empirical analysis of economic value added as a proxy for market value added. Financial Practice and Education, 7, 41–49. Lovata, L. M. & Costigan, M. L. (2002). Empirical analysis of adopters of economic value added. Management Accounting Research, 13, 215–558. ? Lorincz, S. (2006). Cost structure and complementarity in US Telecommunications, 1989–1999. Information Economics and Policy, 18(3), 285–302. Nemoto, J. , & Asai, S. (2002). Scale economies, technical change and productive growth in Japanese local telecommunications services. Japan and the World Economy, 14(3), 305–320. Ofcom. (2003). Fixed narrowband retail services market: Identi? cation and analysis of markets, determination of market power and setting SMP conditions. Explanatory Statement and Noti? cation. Park, J. , Kwon, Y. & Park, M. (2005). Membership card promotion vs. price reduction in the mobile telephony market. In 2005 Korea management information system international conference, Seoul, Korea (in Korean). Peitz, M. (2005a). Asymmetric access price regulation in telecommunications markets. European Economic Review, 49, 341–358. Peitz, M. (2005b). Asymmetric regulation of access and price discrimination in telecommunications. Journal of Regulatory Economics, 28(3), 327–343. Perrucci, A. , & Cimatoribus, M. (1997). Competition, convergence and asymmetry in telecommunications regulation. Telecommunications Policy, 21(6), 493–512. Sung, N. , & Gort, M. (2000).

Economies of scale and natural monopoly in the US local telephone industry. Review of Economics and Statistics, 82(4), 694–697. Weaver, S. C. (2001). Measuring economic value added: A survey of the practices of EVA proponents. Journal of Applied Finance, 11, 7–17. Whitwell, G. J. , Lukas, B. A. , & Hill, P. (2007). Stock analysts’ assessments of the shareholder value of intangible assets. Journal of Business Research, 60(1), 84–90. Willig, R. D. (1979). Multiproduct technology and market structure. American Economic Review, 69(2), 346–351. Young, D. (1997). Economic value added: A primer for European managers. European Management Journal, 15(4), 335–343.

Scale Economies on Wireless Telecommunications Industry

Averagecost,

Cost,

Costcurve,

Costs,

Marginalcost,

Microeconomics,

Totalcost,

VariablecostARTICLE IN PRESS Telecommunications Policy 33 (2009) 29–40 Contents lists available at ScienceDirect Telecommunications Policy URL: www. elsevierbusinessandmanagement. com/locate/telpol Estimating scale economies of the wireless telecommunications industry using EVA data$ Changi Nam a, Youngsun Kwon a,A, Seongcheol Kim b, Hyeongjik Lee c a b c

School of IT Business, Information and Communications University, 119, Munjiro, Yuseong-gu, Daejon 305-732, Republic of Korea Associate Professor, School of Journalism and Mass Communication, Korea University, 5-1, Anam-dong, Seongbuk-gu, Seoul, 136-701, Republic of Korea Full-time instructor, Department of Management Science, Republic of Korea Naval Academy, 88-1 Angok-dong, Jinhae, Kyungnam, 645-797, Republic of Korea a r t i c l e in fo abstract This paper proposes a new estimation method of total cost and average cost curves and applies it to the telecommunications industry.

The method is more ? exible and entails less hassle for data collection than traditional methods. The results show that the longrun average cost (LRAC) curve is downward sloping, revealing the presence of economies of scale in production. The two largest Korean mobile network operators are realizing full economies of scale, while the smallest operator is not. Finally, the paper recommends three policy alternatives that the Ministry of Information and Communication of Korea can draw on to increase ef? ciency in the Korean mobile telecommunications market. & 2008 Elsevier Ltd. All rights reserved.

Keywords: Scale economy estimation Economic value-added (EVA) Telecommunications Asymmetric regulation 1. Introduction Empirically estimating the long-run average cost (LRAC) curve and the minimum ef? cient scale (MES) of production has been an important research topic in the ? eld of regulatory economics because data relating to both are critical for measuring production ef? ciency in industries, especially in public utility variants. One major problem in estimating the LRAC curve and the MES has been obtaining the appropriate data, especially data related to production factors other than capital and labor.

Nowadays, ? rms are capitalizing more than ever before on the bene? ts of intangible assets, such as computer and management software, brand, and intellectual property rights. 1 The traditional estimation method for the LRAC curve and the MES focusing on traditional production factors in manufacturing industries has been of decreasing validity and usefulness in the information age. According to Brynjolfsson and Hitt (2000, 2003), information and communication technology increases the productivity of ? rms by complementing organizational capital and streamlining business processes.

Because of the increasing importance of intangible assets, researchers who pay attention only to traditional production factors are likely to omit a proportion of input cost items. Therefore, this paper develops a new method of estimating the LRAC curve, free from changes in the composition of production factors. The purpose of this paper is to propose a new model to estimate the LRAC curve of ? rms, and to apply it to Korean mobile network operators (MNOs). This paper estimates the LRAC curve of production using annual sales data, estimated economic value-added (EVA) data, and annual This research was supported by the Ministry of Information and Communication (MIC), Republic of Korea, under the Information Technology Research Center (ITRC) support program supervised by the Institute of Information Technology Assessment (IITA) ‘‘(IITA-2006-C1090-0603-0041)’’ A Corresponding author. Quello Center, Michigan State University, MI, USA. Tel. : +1 517 803 0497; fax: +1 517432 8065. E-mail addresses: [email protected] ac. kr (C. Nam), [email protected] ac. kr (Y. Kwon), [email protected] ac. kr (S. Kim), [email protected] ac. kr (H. Lee). 1 Refer to Whitwell, Lukas, and Hill (2007) on the growing importance of intangible assets. 308-5961/$ – see front matter & 2008 Elsevier Ltd. All rights reserved. doi:10. 1016/j. telpol. 2008. 10. 005 ARTICLE IN PRESS 30 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 production. In other words, this paper is not utilizing traditional structural models derived from the cost and production functions, but devises a new model. When this model was applied to a Korean MNO, SK Telecom, it was found that the LRAC curve is downward sloping as expected and asymmetric regulation is a source of the pro? t to the dominant player in the market. One limitation in using the model is that it can only be applied to ? ms listed on the stock market. This paper is organized as follows. The Section 2 brie? y reviews traditional approaches to estimating LRAC curves used in previous research and discusses their limitations. In Section 3, a new model is proposed and compared with traditional models to identify the strengths and weaknesses of the model. Section 4 introduces data and output measurement issues. Section 5 presents the results of the estimation and Section 6 discusses the policy implications of the results for asymmetric regulation that have been used in the Korean telecommunications industry since 2000. Section 7 concludes the paper. . Traditional approaches of estimating the LRAC curve The main reason for estimating the LRAC curve is to ? nd out the magnitude of scale economies and the MES. This information is necessary and useful for regulators and business managers. Using this information, regulators can deduce the appropriate market structure of an industry and managers can calculate the optimal size of their business in terms of production ef? ciency. Unsurprisingly, managers can also use the information for such strategic purposes as deterring market entry by competitors and consolidating pro? ts. As Huettner (1973, pp. 27–429) summarized, previous studies have used one of two methods: estimating the LRAC curve or the short-run average cost curve or directly predicting the level of scale economies by estimating the production function. These two approaches are intrinsically the same because the model used in the cost function approach is also derived from the production function and budget constraints. 2 In other words, if the production technology has an increasing-returns-to-scale property, the estimated average cost function should be downward sloping either locally or globally when factor prices are stable or constant.

The cost function approach, estimating the LRAC and its curvature or estimating the LRAC and long-run marginal cost (LRMC) to calculate the level of scale economies, has been widely used in ? research. 3 Two recent studies drawing on the cost function approach are Asai (2006) and Lorincz (2006). 4 Huettner (1973) pointed out that one critical problem in measuring scale economies is obtaining appropriate data on cost variables and on the quantities of output and factor inputs. In addition, existing empirical papers suffer from measurement problems.

Some papers used only variable costs, ignoring capital costs, and others ignored the age of capital facilities. Recently, as information technology advances and competition becomes intense worldwide because of globalization, ? rms have tended to increase their investment in intangible assets such as software, brand, product design, and marketing in order to reduce production costs. This trend seems to be eroding the robustness of traditional approaches of estimating scale economies. Not taking into account the costs of invisible assets is likely to distort the estimation results of scale economies even though the direction of distortion, i. . , under- or overestimation of scale economies, is not yet well known. 3. The model This paper proposes a new method of deriving the LRAC, which is calculated by subtracting economic pro? t (EP), whose equivalent term is EVA in management, from total revenue (total sales). By de? nition, EP is equal to total revenue less total cost. Therefore, if EP is equivalent to EVA, and EVA can be somehow estimated, total cost can also be estimated. Then, the LRAC can be estimated by dividing the estimated total cost by quantity sold in each year.

In short, this paper adopts a reversed approach to deriving the LRAC curve compared with the traditional approaches that ? rst obtain the costs of major factor inputs and secondly sum them to obtain total average costs. The reversed approach reduces the hassles of obtaining data and avoids the issue of missing cost variables. Total revenue (TR) is easily observable, but total cost (TC) is not. The model starts from the idea that TC can be estimated as long as EVA is estimable as shown in the following equation: EVA ? TR A TC (1) Regarding Eq. (1), two points are worth mentioning. Eq. 1) is based on the assumption that EVA is the same as EP, or at least the assumption that EVA is an unbiased estimate of EP. EP is the residual remaining after the opportunity costs of all inputs are subtracted from TR. By the same token, EVA is the residual a ? rm retains after compensating for all explicit and implicit factor costs. In other words, the two are conceptually equivalent. The second point of discussion is whether the estimated EVA is comparable to the estimated EP. Conceptually, they are the same, but EP is TR less TC, where TC is the cost of producing the optimal quantity of production, which in turn is determined by a ? m in an effort to minimize production costs given that the prices of the ? nal product and inputs are ? xed. The point is that TC, used in estimating EP, is the 2 3 4 Refer to Fare and Logan (1983) for duality between cost and production functions. See Considine (1999) for how to use average cost and marginal cost concepts to understand the level of scale economies. Two more recent empirical papers are Sung and Gort (2000) and Katrishen and Scordis (1998). ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 31 minimum cost of production evaluated at the optimal factor proportion of inputs.

In other words, if the factor proportion resulting in maximum EP is different from that for maximum EVA, even though EVA and EP are conceptually the same, their estimated values are likely to be different. However, if the assumption that a ? rm is minimizing costs in business holds, EVA should be equal to EP. It would be fair to say that EP is the lower bound of EVA. In brief, using EVA as a surrogate of EP requires justi? cation. As explained later, estimating EVA in the model requires that ? rms be listed on the stock market. If ? rms are so listed, it is not an unreasonable assumption that they are pursuing maximization of ? m value, which in turn implies that they are trying to maximize their pro? t. Pro? t maximization requires cost minimization. Therefore, the EVA of a listed ? rm can be seen as a fair approximation of its pro? t. Finally, EVA can be considered an unbiased estimate of EP and be used as a proxy of EP. Because the equivalence of EVA and EP are proven at least intuitively, the next step is to estimate EVA. EVA ? NOPAT A WACC A IC (2) EVA is usually measured in the literature as the difference between a ? rm’s net operating pro? t after taxes (NOPAT) and its total cost of capital, which is the ? m’s invested capital (IC) times the weighted-average cost of capital (WACC), as shown in Eq. (2). 5 However, it is dif? cult to estimate EVA directly in practice because various accounting adjustments are required for calculating the ? rm’s NOPAT and IC. Lovata and Costigan (2002) point out that more than 100 possible adjustments are needed to estimate EVA directly. Therefore, estimating EVA with publicly available accounting data does not seem to be an ef? cient and objective way to estimate EVA. This paper, instead, attempts to estimate EVA indirectly, using the functional relationship between the EVA and market value-added (MVA) of a ? m. MVA ? EVA WACC (3) Conceptually, MVA is the present value of the stream of future EVAs of a ? rm. Therefore, although previous empirical studies such as Kramer and Pushner (1997) did not ? nd a strong relationship between EVA and MVA, the two measures of business performance should be closely correlated with each other. As Young (1997) mentioned, if a ? rm is expected to achieve the same magnitude of EVA every year, the ? rm’s MVA becomes the present value of its expected future EVAs as shown in Eq. (3). In addition, MVA is equal to the market value (MV) of a ? rm less the value of IC, as shown below MVA ?

MV A IC Then, from Eqs. (3) and (4), Eq. (5) is derived. This paper draws on Eq. (5) to estimate EVA. EVA ? WACC A ? MV A IC? (5) (4) If a ? rm is listed on the stock market, MV can be estimated by multiplying the stock price of a ? rm and the number of stocks outstanding, and then adding it to the book value of its debt. 6 IC also can be obtained from the current balance sheet of the company, although some asset values are based on historical data. WACC can also be calculated using both the stock price and ? nancial statements data. This paper calculates the data for MV, IC, and WACC, and then estimates EVA and TC, respectively.

Compared with the traditional methods, the model has several strengths as well as limitations. First, it subsumes all the cost items, tangible or intangible. As Weaver (2001) discussed, invisible assets such as R&D and advertising are not expensed but capitalized in the process of calculating EVA. Secondly, the estimation method in this paper might be more appropriate than traditional methods in estimating a ? rm’s total cost and the LRAC curve because the EVA estimated by the new method not only emphasizes a historical accounting performance, but also re? ects the future cash ? ow of the ? rm in the long run. Because a ? m’s total cost, obtained by subtracting the estimated EVA from its total revenue, represents the present value of future expected costs, the estimated LRAC curve is expected to overcome the shortcoming of the traditional approach that input prices and technology are assumed to be ? xed in the long run. 7 Thirdly, the reliability of the estimated EVA is also expected to be improved because the proposed method does not need to estimate the ? rm’s NOPAT with any adjustments. One limitation of the model is that EVA is derived from Eq. (3), based on the rather strong assumption that the current EVA continues into the future.

Another limitation is that the model is applicable only to ? rms whose stocks are traded publicly in the market because data for MV cannot otherwise be obtained easily. 5 6 7 Refer to Kramer and Pushner (1997) and Weaver (2001) for a detailed discussion of EVA concept. The book value of debt is added to the value of the ? rm’s equity, based on the assumption that the ? rm holds debts until maturity. See Huettner (1973) for a discussion on the shortcomings of the traditional methods. ARTICLE IN PRESS 32 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 4. Data and output measure 4. 1.

Data The key variables for the estimation of EVA are MV, IC, and WACC, as shown in Eq. (5). The calculations of these three variables and data sources are summarized in Appendix A. Financial statements of the Korean MNOs are obtained from the Korean Financial Supervisory Service,8 stock price data were taken from the Korea Exchange,9 and the annual number of subscribers is from the Korean Ministry of Information and Communication. 10 The average stock price for December is used in calculating the equity value of a year, and then the equity value data of 2 consecutive years are used for the calculation of the average equity value of a year.

The average stock price of December is used instead of the end of year stock price to use a more representative stock price of the year. The values of ? rms’ debts, IC, and WACC are calculated following the standard methods used in ? nance textbooks. Finally, the TC variable derived from Eq. (1) is de? ated by the gross domestic product (GDP) de? ator of the telecommunications industry in order to re? ect the decreasing equipment costs of the telecommunications sector. The GDP de? ator of the telecommunications sector is obtained from the Bank of Korea website. 4. 2.

Output measure Obtaining an appropriate output measure is a prerequisite for calculating the LRAC, especially when the industry concerned is producing multiple heterogeneous products. As Carlton and Perloff (2000) point out in their textbook, if a ? rm is producing multiple products such as oranges and apples, then deriving the average cost of production becomes problematic. In previous work, such as Nemoto and Asai (2002), the quantity produced is derived from the TR variable. TR, by de? nition, is simply the product prices times the quantities sold in the market.

Therefore, the quantity can be derived by dividing TR by the appropriate price index. However, this approach is subject to a few critical problems. First, if heterogeneous products are produced by a ? rm, it is dif? cult to interpret the quantity variable derived from TR because the unit of the quantity variable cannot be determined. In addition, the quantity variable derived from TR has problems if the composition of the product mix changes from year to year. Secondly, ? nding an appropriate price index for the quantity measure of multiple products is very dif? ult, especially when there are a large number of products. This paper uses a direct measure of quantity supplied, a publicly available output measure in the telecommunications industry—the number of subscribers. It is well known that traf? c is a major cost driver in the telecommunications industry. Therefore, using traf? c volumes as an output measure seems appropriate. Nowadays, however, almost all telecommunications ? rms deliver voice as well as data services in the market and different metrics are used for voice and data traf? c: call minutes for the former and bytes of data transferred for the latter. 1 It is obvious that conceptually two metrics of different units cannot be merged into one. In the telecommunications industry, the number of subscribers has been used as an output measure in addition to traf? c volume. The number of subscribers can de? nitely be considered a major cost driver because they are generating voice and data traf? c. Therefore, this paper uses the number of subscribers, N, rather than traf? c as the output measure. TC is a function of voice and data traf? c, as presented in the following equation: TC ? f ? t v ; t d ; w; T? (6) where tv and td are voice and data traf? respectively; w is the input price vector; and T is the technology characteristic vector. Because tv and td are increasing functions of the subscribers, Eq. (6) can be rewritten as TC ? f ? t v ? N? ; t d ? N? ; w; T? ? g? N; w; T? LRAC ? TC? N; w; T? N (7) (8) This paper ? nally calculates LRAC using Eq. (8). This way of measuring the LRAC is especially appropriate for present purposes because time series data spanning 16 years are used, for which output composition has changed from solely voice services to voice and data services in the telecommunications industry. 5.

Empirical results: an application to the Korean telecommunications industry The model is applied to the Korean wireless communication market, where three MNOs—SK Telecom (hereafter SKT), KTF and LG Telecom (hereafter LGT)—are competing. At the end of 2006, the three MNOs’ market shares were 50. 4%, 32. 2%, 8 9 http://www. fss. or. kr http://www. krx. co. kr http://www. mic. go. kr 11 See Kelly and Woodall (2000) for trends in traf? c data. 10 ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 33 12000 10000 TR TR (billion Won) 8000 6000 Subscribers 5 20 Subscribers (million) 15 10 4000 2000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 5 0 Year Fig. 1. SKT’s TR (left Y-axis) and subscribers (right Y-axis). and 17. 4%, respectively. The LRAC and LRMC curves are derived for only SKT, the largest MNO in Korea, primarily because of data availability. SKT’s stocks began to be traded on the Korean Exchange from 1989, whereas KTF and LGT were publicly listed on the stock market in 1999 and 2000, respectively. Therefore, there are scant available data for EVA estimation for KTF and LGT.

The model, however, is also applied to them in order to compare the TC and LRAC of the three MNOs. 5. 1. Estimation of the scale economies This section estimates the TC of SKT, whose ? nancial data are available since 1990 and derives LRAC and LRMC curves from the estimated TC function. Since 1990, SKT’s TR has grown continuously, with the number of subscribers increasing from 88,000 in 1990 to 19. 5 million in 2005, as shown in Fig. 1. Following the process shown in Appendix A, this paper estimates the MV, IC, WACC, and EVA of SKT from 1990 to 2005, as shown in Table 1.

For the whole estimation period, the EVA of SKT has been positive and growing in size up to a maximum of about 1. 5 trillion Korean Won in 2001. The estimated EVA of SKT has decreased since 2001 because of increased IC, mainly caused by new investments for network upgrades to accommodate wireless Internet services, and market saturation, but returned to nearly 1 trillion Won in 2005. The WACC of SKT fell continuously until 1997 and since then has slowly been returning to its 1990s level. By de? nition, WACC is the weighted average of the cost of equity and cost of debt, so the trend of SKT’s WACC re? cts the fall in the risk-free interest rates of the early1990s and the rise in the cost of equity and debt in the latter part of the decade. The TC of SKT, estimated by subtracting the estimated EVA from TR, is de? ated by the GDP de? ator of the telecommunications industry. Fig. 2 presents the trend of TC in nominal and real value terms. According to Fig. 2, TC has increased continuously, keeping pace with the number of subscribers, except for 1999 and 2001 when SKT recorded exceptionally high values of EVA. TC is unlikely to decrease in the real business world especially when a ? m grows, and the idiosyncratic events of 1999 and 2001 stem from the TC estimation method of this paper. EVA re? ects changes in the stock price, which in turn re? ect investors’ expectations on SKT’s future pro? ts. Therefore, when SKT commanded exceptionally good business prospects as it did in 1999 and 2001, TC would be biased downward temporarily because of an overestimated expected EVA. From the 1990s to the present, the GDP de? ator of the telecommunications industry has fallen consistently, so the real TC curve rotates counter-clockwise compared with the nominal TC curve.

Fig. 3 presents the real TC curve, which shows the relationship between the real total costs and the subscriber variable (the quantity produced by the ? rm). The smoothly curved TC curve, ? tted using the ordinary least-squares method to estimated total costs, is shown in Eq. (9) and also presented in Fig. 3. TC ? 2:8147N3 A 57:262N2 ? 653:32N (9) The ? tted TC curve looks like the typical TC curve, found in microeconomics textbooks, i. e. , TC initially rises at a decreasing rate until it reaches the in? exion point and, after that, at an increasing rate.

In deriving the TC curve, the intercept is set to zero because TC is a long-run concept in this paper. ARTICLE IN PRESS 34 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 Table 1 Trends of major variables (billion Korea Won). Year 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 MV (A) 238 467 859 1558 3293 4705 4667 5137 6512 28,430 26,433 26,823 27,783 23,416 23,253 22,064 IC (B) 53 83 92 166 294 631 1242 1751 2374 2999 3399 4099 5898 8231 9526 9868 MVA (C ? AAB) 185 383 768 1392 2999 4074 3425 3386 4139 25,431 23,034 22,724 21,885 15,184 13,726 12,196 WACC (D) (%) 10. 4 8. 49 7. 75 7. 15 5. 74 5. 46 5. 32 3. 28 4. 63 5. 47 4. 05 6. 51 5. 32 5. 22 4. 64 8. 08 EVA (E ? C A D) 19 33 59 99 172 222 182 111 191 1392 932 1479 1164 793 637 985 14000 12000 10000 8000 6000 4000 2000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 TC (billion Won) TC (real value) TC (current value) Year Fig. 2. Total production cost of SKT. The LRAC curve is obtained by dividing the TC equation by N and LRMC curve by taking the derivative of the TC equation with respect to N. The LRAC and LRMC curves derived from the TC curve are drawn in Fig. 4.

From Fig. 4, it is easy to identify the MES, the size of service production that minimizes LRAC. The MES of SKT turns out to be about 10 million subscribers, half its current size. The MES derived from the nominal TC is about 13 million subscribers, which is greater than that derived from real values. 12 The scale economies index, (LRAC–LRMC)/LRAC, is calculated and presented in Fig. 5. 13 The scale economies index can be greater than, less than, or equal to zero, when returns to scale increase, decrease, or are constant. The index is maximized when the subscriber number is about 6 million.

From Figs. 4 and 5, it can be easily noted that the gap between LRAC and LRMC begins to widen rapidly after the number of subscribers exceeds 10 million. However, based on this result it cannot be said that decreasing returns to scale exist and intensify as the number of subscribers rise above 10 million. This is because the rising portion of the LRMC curve above the minimum LRAC is likely to be caused by the large investment in network upgrade needed for new data services, while the number of subscribers is stagnant and the revenue from the data service is not robust. 4 Considering this, it is fair to say that the recent LRAC and LRMC of SKT are overestimated. The equation of the nominal TC curve is not presented here but is available upon request from the corresponding author. Willig (1979) used a different but equivalent measure of scale economies. The share of revenue from wireless Internet services in monthly average revenue per user was only 1% in January 2000, but it increased to 27% in December 2005, which indirectly reveals the investment in the wireless Internet service made by SKT. 13 14 12 ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 35 4000 12000 10000 TC (billion Won) 8000 TC (real value) 6000 4000 2000 0 0 2 4 6 8 10 12 14 16 18 20 Subscribers (million) Fig. 3. Fitted TC curve of SKT. TC = 2. 8147N3 – 57. 262N2 + 653. 32N R2 = 0. 987 2000 1800 1600 AC/MC (thousand Won) 1400 1200 1000 800 600 400 200 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Subscribers (million) LRAC LRMC Fig. 4. AC and MC curve of SKT. 5. 2. Estimation of cost curves of other MNOs in Korea The same estimation method of TC is applied to KTF and LGT, and Fig. 6 compares their TCs with that of SKT. Even though data from KTF and LGT are scant, some ? dings are worth attention. First, when the number of subscribers is around ? ve million, the estimated total costs of the three MNOs are roughly the same; as the number of subscribers rises, the total cost curves begin to diverge. Considering that SKT’s estimated TC was exceptionally low because of a high expected EVA when it reached 10 million subscribers, KTF’s TC curve could be viewed as SKT’s hypothetical TC curve without any overexpectation on SKT’s business performance on the stock market. Secondly, LGT and KTF use a 1. 8 GHz spectrum, while SKT uses an 800 MHz spectrum.

It is frequently argued that the difference in spectrum bands results in a difference in costs, (mainly in investment costs), which is in turn caused by the difference in the cell coverage of wireless communication. However, it is also argued that in metropolitan areas, where the majority of subscribers are living, the cost variation owing to spectrum band differences is almost zero and, as a result, the overall cost difference between SKT and the other two ? rms should be insigni? cant. Fig. 6 seems to support the latter view, i. e. , the proposition that the cost difference among MNOs is not signi? ant, controlling for the number of subscribers. Three points regarding this are important. First, there is no reason for LGT, using the adjacent spectrum with KTF, to have a different cost function, given the assumption that the two ? rms adopt ef? cient technologies and management methods. Secondly, because SKT has invested in its network upgrades for wireless Internet service during the past few years, the ARTICLE IN PRESS 36 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 0. 5 0. 0 1 Scale Economies Index -0. 5 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 -1. 0 -1. 5 -2. Subscribers (million) Fig. 5. Scale economy index. 14000 12000 10000 TC (billion Won) 8000 6000 4000 2000 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Subscribers (million) KTF LGT SKT Fig. 6. Estimated TCs of Korean MNOs. other two ? rms followed the same path even though they are underdogs in the market. 15 This can be observed in Fig. 6. In recent years, the three MNOs’ TC curves exhibit the same pattern, digressing from the previous trend. Thirdly, LGT’s TC curve appears to be an upward sloping straight line with a very steep slope, which implies that LGT does not bene? from scale economies. This TC curve suggests that LGT invested or was forced to invest, because of competition with other MNOs, in building a wireless data network before realizing economies of scale in the voice communications market. Fourthly, subscribers to KTF have increased from 5. 3 million in 2000 to 12. 3 million in 2005; they have exceeded 10 million, the MES derived from SKT’s LRAC curve, since 2002. Therefore, even though there are only six sample data points for estimation of the TC curve, the ? tted TC curve is derived and appears in Eq. (10) with R2 ? 0. 96. TC ? :2718N3 A 106:3N 2 ? 830:1N (10) The basic shape of KTF’s TC curve is quite similar to SKT’s. From Eq. (10), the LRAC curve of KTF is drawn and juxtaposed with that of SKT in Fig. 7. The MES, calculated from KTF’s LRAC curve, is about 7. 3 million, smaller than the MES derived from SKT’s LRAC curve. However, considering that KTF has invested in network upgrades for its data service since 2001, it is highly likely that the MES of KTF is underestimated. LGT’s subscriber number was about 6. 5 million at the end of 2005, which is smaller than either of the MESs and does not appear to be an ef? ient scale. 15 According to our estimation, the ICs of SKT, KTF, and LGT increased by 174%, 152%, and 72%, respectively, between 2000 and 2005. ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 37 1800 1600 LRAC_KTF 1400 AC (thousand Won) 1200 1000 800 600 400 200 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Subscribers (million) LRAC_SKT Fig. 7. AC curves of SKT and KTF. According to Fig. 7, SKT looks more ef? cient than KTF over the whole range of domain denoted by the number of subscribers.

Above all, the available data for the estimation of KTF’s TC curve is too small to derive a general conclusion. However, the differences in spectrum ef? ciency, economies of scale, and economies of scope can be designated as potential contributors that result in differences in the LRAC curves, even though the reasons this occurs cannot be determined exactly. 16 The estimated cost curves will change as the customer base grows and new services such as mobile banking, digital media broadcasting, and wireless Internet connection begin to earn pro? s, so additional time and data are needed to draw a more de? nite conclusion. 6. Policy implications 6. 1. Asymmetric regulation in Korean telecommunication market In the Korean telecommunications market, wireless telephony service had been provided solely by SKT since 1984. Competition was introduced in the mid 1990s after four MNOs using cellular and personal communications service (PCS) technologies launched their mobile telephony services in 1995 and in 1996, respectively. However, the competition among ? e MNOs existed only for 4 years because two MNOs merged with two other MNOs (SKT and KTF) in 2000. The mergers of 2000 brought forth current market structure and since then the market shares of three MNOs have rarely changed as shown in Fig. 8. The Ministry of Information and Communication (MIC) of Korea has implemented an asymmetric regulation policy since 2000 to help late market entrants such as KTF and LGT compete with the incumbent, SKT, in the market, based on the belief that economies of scale and the network externality effect hinder fair competition. 7 Examples of asymmetric regulation policy implemented in Korea are regulating interconnection charges based on the individual ? rm’s network cost rather than on industry average network cost, regulating only the rate of SKT voice service, and applying number portability policy to MNOs sequentially, not concurrently, with 6 months time lags among them from LGT to SKT. Asymmetric regulation prohibited price competition among MNOs in Korea’s mobile phone service market and instead resulted in non-price competition through handset subsidies and various membership card promotions. 18 6. 2.

Implications for policy makers The MIC is arguing that asymmetric regulation is necessary for promoting effective competition in the Korean mobile telephone service market despite the criticism that it has done more harm than good, especially in terms of subscriber welfare. SKT has been making considerable positive (excess) economic pro? ts for more than a decade, which indicates that it has plenty of room for maneuver. The MIC continues to worry that price competition would drive LGT out of the market, which would eventually harm consumers by making it easier for the remaining two operators to collude in price setting. 6 In the Korean metropolitan area, the coverage area of a cell of an 800 MHz spectrum is 6. 36 km2 while that of a 1. 8 GHz spectrum is 2 km2. The number of base stations of KTF (7051) is higher than that of SKT (5243) according to Kim (2006). 17 Refer to Perrucci and Cimatoribus (1997) for general discussions on asymmetric regulation and Peitz (2005a, b) for recent theoretical studies on asymmetric regulation. 18 See Park, Kwon, and Park (2005) for a comparison of price reduction and membership card promotion in Korea. ARTICLE IN PRESS 38 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 60. 0% 2001. 2 50. 0% 40. 0% 30. 0% 20. 0% 10. 0% 0. 0% SKT KTF LGT Fig. 8. Changes in market shares of Korean MNOs. 2007. 7 The authors’ calculation shows that LGT is currently making a small amount of economic pro? t and the ? nancial data from an accounting ? rm reveal that LGT incurred economic losses between 2002 and 2003. That is, LGT is surviving on the margin; thus price competition could drive it out of the market as suggested by the MIC. From an alternative perspective, however, prohibited price competition can be seen as playing the role of guaranteeing positive pro? ts for SKT and KTF and the survival of LGT. 9 Focusing on the number of players in the mobile telephone service market, the MIC has been ignoring consumer welfare for more than six years. Asymmetric regulations have not produced effective competition in the Korean mobile telecommunication industry so far as illustrated by Fig. 8. Based on the analysis of this paper, three policy recommendations regarding asymmetric regulation in Korea are addressed below. The ? rst policy option is to abolish asymmetric regulation and simply let the market work. However, abolishing asymmetric regulation does not mean removing rate regulation altogether. The MIC can egulate the rate of the dominant player in the market by adopting price cap regulation. 20 This option could result in two players in the Korean mobile telephone service market after a period of ? erce price competition as discussed above. The removal of asymmetric regulation might not produce spontaneous price competition if SKT and KTF want to make economic pro? ts. If this is the case, the MIC can make the regulated rate fall gradually until the EVA of the dominant player falls to an appropriate level for the sake of consumer welfare. The second option is to reshuf? e the current market structure by intensifying asymmetric regulation.

Such regulation does not automatically allow LGT to expand its market share and enjoy economies of scale effects. Therefore, the success of this policy suggestion will depend on how effectively the MIC reshuf? es the market structure by using policy measures. According to the present analysis, the MES in the Korean mobile telephone service industry is located at between 10 million and 13 million subscribers, and derived from nominal values of TC. At the end of 2006, the total number of subscribers to mobile telephone services in Korea was about 40 million, which is large enough for three MNOs to attain the MES.

A balanced market share among three MNOs in Korea would result in a win–win outcome for producers and consumers even though the dominant ? rm obviously would not be happy with the outcome. The key problem is how to implement it in a free market economy, considering that notwithstanding asymmetric regulation by the MIC since 2000, the market share among the three did not change signi? cantly until recently. The policy that forces the dominant MNO to resell wireless services (airtime) to competitors (especially to LGT) at wholesale rates seems to be a feasible way to attain a more balanced market structure. 1 This means that LGT becomes a reseller of mobile airtime compared with being a traditional facilitybased network service provider. The third option is to induce MNOs that cannot attain the MES to turn from facility to service-based competitors, i. e. , becoming mobile virtual network operators (MVNOs). This policy would increase ef? ciency in production as well as in consumption by inducing those that cannot attain the MES to drop off from facility-based competition and boost competition in retail market.

In addition, the MIC needs to boost price competition in Korean mobile telecommunications market to motivate LGT to get out of facility-based competition and to enter into service competition. 7. Conclusion This paper proposes a new method of estimating scale economies by deriving TC from TR using estimated EVA. This method overcomes the problem of using historical accounting data and uses readily available ? rm-level data. One major 19 20 21 One reviewer also mentioned that the survival of LGT could stem from the asymmetric regulation.

The UK is using this method of price regulation (Ofcom, 2003). KT, the dominant wired network operator in Korea, has been reselling airtime at retail prices by purchasing it at wholesale rates from KTF. ARTICLE IN PRESS C. Nam et al. / Telecommunications Policy 33 (2009) 29–40 39 drawback is that this method is only applicable to ? rms whose stocks are traded on the stock exchange. The new method is applied to the Korean MNOs. The empirical results show that the MES lies between 10 million and 13 million subscribers, although the result cannot be generalized because of a lack of data.

The estimated MES implies that production ef? ciency and consumer welfare might not fall without asymmetric regulation even though only two MNOs exist in the Korean mobile telephone service market. Considering the size of the estimated MES, three MNOs can coexist without lowering production ef? ciency and consumer welfare if the market share is rearranged in a more balanced way. Available policy options to the MIC would either abolish asymmetric regulation, which has not been effective in changing market structure in the past, or reinforce it to increase LGT’s market share.

While this paper presents some meaningful implications, it is not without limitations. The estimation method of predicting scale economies is based on several strong assumptions as discussed above, even though it is believed that relaxing these assumptions would not signi? cantly change the results of the analysis. These assumptions should be relaxed to improve the generality of the analysis. Secondly, the results of the authors’ estimation and the implications drawn from the empirical results might be limited in applicability mainly because of the scarcity of samples.

Therefore, a useful area of future research would be to extend the empirical analysis to other parts of the telecommunications industry or to an international context. Appendix A. The estimation process of EVA for the Korean MNOs in year X See Table A1. Table A1 Items EVAX MVX EVX PX CSX ATDX AICX ICX TAX NIB_CLX APX OAPX AEX OAFCX IMMX URX UIX FDX OCLX NIB_LLX LURX LOAPX LUIX OLPX OLLX NOAX LAPX LFIX ISX ISSRX DDCTX BAX AX WACCX Meanings Economic value-added (economic pro? ) Market value Market value of equity The daily average closing price during Decembera The number of issued common stocks Average total debtsb Average invested capital Invested capital The total assets Non-interest-bearing current liabilities Accounts payable Other accounts payable (including accrued dividends and tax payable) Accrued expenses Other advance from customers Import margin money Unearned revenue Unearned income Financial derivatives Other current liabilities Non-interest-bearing long-term liabilities Long-term unearned revenue Long-term other accounts payable Long-term unearned income OTHER liability provisions Other long-term liabilities Non-operating assets Loans to af? liated companies Long-term ? ancial instruments Investment securities Investment securities of person with a special relationship Differences of deferred corporate taxes Building account Allowanced Cost of capital Calculation (source) (MVXAAICX) A WACCX EVX ? ATDX PX A CSX (Korea Exchange) (Korean Financial Supervisory Service) TDX ? TDXA1 2 ICX ? ICXA1 2 TAX A NIB_CLX A NIB_LLX A NOAX (Korean Financial Supervisory Service) APX ? OAPX ? AEX ? OAFCX ? IMMX ? URX ? UIX ? FDX ? OCLX (Korean Financial Supervisory Service) (A. 2) (A. 1) LURX ? LOAPX ? LUIX ? OLPX ? OLLX (Korean Financial Supervisory Service) LAPX ? LFIX ? ISX ? ISSRX ? DDCTX ? BAX ? AX (Korean Financial Supervisory Service) AIBLX ATEX A CODX ? 1 A TX ? ? A COEX ATAX ATAX (A. 3) ARTICLE IN PRESS 40 C. Nam et al. / Telecommunications Policy 33 (2009) 29–40

Table A1 (continued ) Items ATAX ATEX AIBLX IBLX CODX IEX COEX RFX Meanings Average total assets Average total equity Average total interest-bearing liabilities Total interest-bearing liabilities Cost of debt Interest expenses Cost of equity Annual rate of national housing bond type 1 Beta Risk premium Annual average market return for the last ten years Corporate tax rate Calculation (source) TAX ? TAXA1 2 TEX ? TEXA1 2 IBLX ? IBLXA1 2 TAXANIB_CLXANIB_LLXc IEX AIBLX (Korean Financial Supervisory Service) RFX ? bX A RPX (Bank of Korea) (Korea Exchange)d RFX A MRX (Korea Exchange) (Korean Financial Supervisory Service)e bX RPX MRX TX a The last closing price is ideal but inappropriate for estimating the market value because of its volatility.

We also attempted to estimate the market value using the last closing price and found that there is almost no difference compared with the results when using the daily average closing price during December. b Because the market value of the debt is not publicly available, the book value is used instead. c It is easily calculated by estimating IC. d This is obtained from the regression between a market return and a rate of return of the MNO, from the listed day to the last day of year X. e The corporate tax rate is constant at 20. 7% in Korea. References Asai, S. (2006). Scale economies and scope economies in the Japanese broadcasting market. Information Economics and Policy, 18(3), 321–331. Brynjolfsson, E. , & Hitt, L. (2000). Beyond computation: Information technology, organizational transformation and business performance.

Journal of Economic Perspectives, 14(4), 23–48. Brynjolfsson, E. , & Hitt, L. (2003). Computing productivity: Firm-level evidence. Review of Economics and Statistics, 85(4), 793–808. Carlton, D. W. , & Perloff, J. M. (2000). Modern industrial organization (3rd ed). New York: Addison Wesley Longman. Considine, T. J. (1999). Economies of scale and asset values in power production. Electricity Journal, 12(10), 37–42. Fare, R. , & Logan, J. (1983). The rate-of-return regulated ? rm: Cost and production duality. Bell Journal of Economics, 14(2), 405–414. Huettner, D. A. (1973). Shifts of long run average cost curves: Theoretical and managerial implications.

Omega, 1(4), 421–450. Katrishen, F. , & Scordis, N. A. (1998). Economies of scale in services: A study of multinational insurers. Journal of International Business Studies, 29(2), 305–323. Kelly, T. , & Woodall, M. (2000). Telecom traf? c indicators. Telecommunications Policy, 24(2), 155–159. Kim, Y. (2006). The impact of cellular-spectrum expansion on the structure and performance of mobile telecom market. Korea Telecommunications Policy Review, 13(1), 1–25 (in Korean). Kramer, J. K. , & Pushner, G. (1997). An empirical analysis of economic value added as a proxy for market value added. Financial Practice and Education, 7, 41–49. Lovata, L. M. & Costigan, M. L. (2002). Empirical analysis of adopters of economic value added. Management Accounting Research, 13, 215–558. ? Lorincz, S. (2006). Cost structure and complementarity in US Telecommunications, 1989–1999. Information Economics and Policy, 18(3), 285–302. Nemoto, J. , & Asai, S. (2002). Scale economies, technical change and productive growth in Japanese local telecommunications services. Japan and the World Economy, 14(3), 305–320. Ofcom. (2003). Fixed narrowband retail services market: Identi? cation and analysis of markets, determination of market power and setting SMP conditions. Explanatory Statement and Noti? cation. Park, J. , Kwon, Y. & Park, M. (2005). Membership card promotion vs. price reduction in the mobile telephony market. In 2005 Korea management information system international conference, Seoul, Korea (in Korean). Peitz, M. (2005a). Asymmetric access price regulation in telecommunications markets. European Economic Review, 49, 341–358. Peitz, M. (2005b). Asymmetric regulation of access and price discrimination in telecommunications. Journal of Regulatory Economics, 28(3), 327–343. Perrucci, A. , & Cimatoribus, M. (1997). Competition, convergence and asymmetry in telecommunications regulation. Telecommunications Policy, 21(6), 493–512. Sung, N. , & Gort, M. (2000).

Economies of scale and natural monopoly in the US local telephone industry. Review of Economics and Statistics, 82(4), 694–697. Weaver, S. C. (2001). Measuring economic value added: A survey of the practices of EVA proponents. Journal of Applied Finance, 11, 7–17. Whitwell, G. J. , Lukas, B. A. , & Hill, P. (2007). Stock analysts’ assessments of the shareholder value of intangible assets. Journal of Business Research, 60(1), 84–90. Willig, R. D. (1979). Multiproduct technology and market structure. American Economic Review, 69(2), 346–351. Young, D. (1997). Economic value added: A primer for European managers. European Management Journal, 15(4), 335–343.

How to cite this page

Choose cite format:

Scale Economies on Wireless Telecommunications Industry. (2018, Jun 13). Retrieved from https://studymoose.com/scale-economies-on-wireless-telecommunications-industry-essay

We will write a custom essay sample onScale Economies on Wireless Telecommunications Industryspecifically for you

for only $16.38 $13.9/page
Order now

Sorry, but copying text is forbidden on this website. If you need this or any other sample, we can send it to you via email.

By clicking "SEND", you agree to our terms of service and privacy policy. We'll occasionally send you account related and promo emails.

Our customer support team is available Monday-Friday 9am-5pm EST. If you contact us after hours, we'll get back to you in 24 hours or less.

By clicking "Send Message", you agree to our terms of service and privacy policy. We'll occasionally send you account related and promo emails.
No results found for “ image
Try Our service