Information Economics and Policy 22 (2010) 4–17
Contents lists available at ScienceDirect
Information Economics and Policy journal homepage: www.elsevier.com/locate/iep
The relationship between mobile and fixed-line communications: A survey Ingo Vogelsang Department of Economics, Boston University, 270 Bay State Road, Boston, MA 02215, United States
a r t i c l e
i n f o
Article history: Available online 16 December 2009 JEL classification: L86 L88 L96 Keywords: Fixed-to-mobile substitution Network effects Cross-price elasticity
a b s t r a c t The dramatic worldwide increase in mobile communication that has led to more than 4 billion users has over the last few years been accompanied in wealthy countries by a significant decline in fixed network subscriptions. Such fixed-to-mobile substitution (FMS) is at the center of this literature survey. Theoretical models explaining FMS are scarce and are inconclusive regarding the balance between substitution and complementarity of the fixed and mobile sectors. Empirical explanations hinge on the interaction of positive cross-elasticities of demand and reductions in mobile relative to fixed communications prices. FMS is also supported by relative declines in mobile network costs, network effects in demand and quality improvements of mobile services. The policy consequences of FMS stem from the potential reductions in market power of operators in fixed-line markets and from the ability of mobile operators to enable universal service. The survey reveals ample opportunity for further empirical and theoretical research in the area of FMS. Ó 2009 Elsevier B.V. All rights reserved.
1. Introduction In 2002, with one billion users worldwide, mobile communications for the first time surpassed fixed-line subscribers (ITU, 2003; Garbacz and Thompson, 2007). At present, there are over 4 billion mobile users against about 1.2 billion fixed-lines (ITU, 2008).1 For about a decade it has been well known that the number of mobile lines in low-income countries has been outpacing the number of fixed lines by an increasing margin. During this period, the number of fixed-lines was increasing as well. By contrast, fixed-lines in high-income countries peaked around 2000 (Albon, 2006), with fixedline penetration in high-income countries remaining above 90% and mobile penetration increasing rapidly to over 100%. Most recently, however, substantial declines in fixed-line penetration have been observed. A particularly striking case is that of high-income Austria, where, as the result of substantial decline in fixed-line penetration and
E-mail address:
[email protected] Since, in contrast to mobile, fixed-lines are usually shared by household members and since mobile lines include prepaid services, fixed and mobile user numbers are not strictly comparable. 1
0167-6245/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.infoecopol.2009.12.002
the growth of mobile telephones, the ratio between fixed telephone lines and mobile users is nearing 1:5 and fixed-to-mobile telephone usage is nearing 1:3 (RTR, 2008). Mobile is also making rapid and substantial inroads into the broadband market that was previously dominated by DSL and CATV. A consequence – and a major difference with low-income countries – is that fixed network capacity is now being under-utilized. This generates policy questions that few other high-income countries currently face with the same urgency but which they are likely to confront in the next few years. Is this an efficient development that should be encouraged? Is it the result of regulatory handicaps imposed on incumbents to help fixed-line entrants that ultimately have instead favored mobile networks in their competition with fixed networks (Vagliasindi et al., 2006)? As already suggested, the relationship between fixed and mobile networks in developing economies differs fundamentally from wealthy economies. In the former, mobile networks accelerated due to a lack of fixed network penetration, because mobile networks could be constructed more rapidly, were not necessarily linked to an entrenched fixed network bureaucracy and were substantially cheaper as the result of low fixed-line network density. In contrast
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
to wealthy economies in developing countries (LDCs), mobile networks have largely fulfilled a new demand for telephony, whereas in wealthy countries (DCs) second line or replacement demand accounted for a larger share of mobile demand. Although the following review offers further discussion of low-income countries and empirical studies about them, greater emphasis is placed on wealthy economies, where network externalities appear to be less of an issue and market power may still exist. The key concept running through the literature identified here is that of fixed-to-mobile substitution (FMS). Once one has access to both fixed and mobile networks, they are clearly substitutes in making calls. Thus, it is necessary to distinguish between calls and access. However, mobile services may attract a premium due to their mobility component that fixed services cannot have, while fixed services have, until now, enjoyed a premium due to high speed Internet capacity which also influences demand for access. On the supply side, even if fixed and mobile service are substitutes for the end consumer, they are complements in transport because connections between cells is done over the fixed network. With these features in mind we examine the literature for answers to four main questions: 1. What is the empirical evidence on the occurrence, extent and patterns of FMS? 2. What factors explain the evidence? 3. What are the policy consequences of FMS? 4. What are the gaps in and shortcomings of the literature and how can they be remedied? In the next section we introduce FMS in more depth and develop a set of hypotheses about its occurrence. Section 3 then surveys the empirical evidence on the occurrence of FMS and the factors explaining it. Section 4 builds on the largely positive analysis of the previous sections to draw normative or policy conclusions, leaving out the strategy implications for telecommunications operators. Section 5 ends with recommendations for further research.
2. Hypotheses on fixed-to-mobile substitution (FMS) 2.1. Characterization This section first defines and characterizes FMS. It then asks how FMS may be explained and gives tentative answers in the form of several hypotheses. FMS means the replacement of fixed-line services with mobile services (Albon, 2006) or the use of mobile instead of fixed phone for calls or access to telecom services (Vagliasindi et al., 2006). It does not necessarily mean substitution in the technical sense of positive cross-price elasticities of demand. It therefore is not necessarily a shift in demand in response to a change in the relative prices of the two services. FMS could also be the result of supply side effects, which the literature treats less deeply than the demand side effects. FMS is considered to be a relatively new phenomenon. When mobile networks are young, they nurture fixed net-
5
works. This is interpreted as complementarity, meaning that mobile growth strengthens fixed networks. Mobile services were seen as a luxury good, very expensive and only useful for being reachable and being able to call on the go. Calls from mobiles mostly went to fixed-lines and calls to mobiles came from fixed-lines. This is also how the fixed-line incumbents considered mobile services that they themselves provided and how regulators ended up treating the supply of mobile services by fixed network incumbents. When mobile networks mature, however, they appear to become substitutes to fixed networks, meaning that mobile growth reduces the size of fixed networks and ultimately can lead to their demise. 2.2. Potential explanations for FMS Explanation of FMS begins with the observation that fixed network subscriptions and telephone usage decline, while mobile network user numbers and usage increase. Thus, (H1) The first and most basic hypothesis from this observation is that the decline of fixed and rise of mobile communications are linked to each other. The only literature that potentially describes this link directly and in a holistic fashion is based on penetration models, treated below in Section 3.1. 2.3. Hypotheses from the demand perspective The term FMS suggests that the observation itself is based on substitution, meaning that the rise of mobile somehow ‘‘causes” the decline of fixed communication. How can FMS be explained and why is it a new phenomenon? We begin with the demand perspective. Complications in demand relationships governing FMS include complementarity between subscription/access and calling for each network, the two-sided markets for calling (caller and receiver), the choice of CPP vs. RPP,2 non-linear pricing (optional calling plans, flat rates), different Lancasterian (attribute) values for fixed and mobile services (i.e. the mobility premium), and last, but not least, network effects. For regular telephone calls, there exists a microeconomic literature that integrates network externalities into demand analysis (based on Rohlfs, 1974). However, what is largely missing is an application to FMS and complementarity issues. These issues are distinguished from the existing literature by the fact that consumer utility for mobile calling and fixed calling (and, in both cases, of being called) depends on subscribership in both types of networks. Sorting out these issues seems difficult and the only published paper I could find, Jeon (2001), is buried in a theoretical journal, has never been cited, and is quite complex and idiosyncratic, though also quite insightful. Jeon models subscription demand and call demand for two differentiated networks (mobile and fixed), building up from the microfoundations of individual calling relations between subscribers of networks leading to individual user demands and then aggregating to market demands. 2
CPP = calling party pays, RPP = receiving party pays.
6
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
Jeon introduces a parameter measuring the degree of substitutability between calls to different networks. Network externalities are assumed to relate the utilities of individual subscriptions to the number of subscribers in a network. Network externalities diminish as substitutability increases, because an additional subscriber for network i is worth less if i is a close substitute to j (to which the user already subscribes) than if it is not a close substitute. As a result of network effects that are not overcome by substitution effects, a price decrease for calls or subscriptions in one network contributes to the growth of both networks. In his model, network externalities have the general effect of turning different types of calls into complements (noting that income effects could increase this complementarity). Thus, the greater the network effect relative to the substitution parameter the more likely that an overall complementarity between calls on different networks exists. Under the conjecture that network effects lessen as subscribership increases, substitutability should at some point outweigh the complementarity effect that he identifies. This could therefore explain the switch between complementarity and substitution observed or conjectured by many authors to be a consequence of increased mobile penetration. A very different and very simplified demand model for calls is presented in Andersson et al. (2009) for the related issue of the relationship between SMS and voice calls in mobile networks. Beginning again with the premise that two products can partially be substitutes and partially be complements, Andersson et al. incorporate a feedback effect from calls, meaning that, on average, outgoing calls result in a percentage of return calls (net of cases where an outgoing call replaces a return call).3 Such effects have been reported by Taylor (2002) and are discussed more broadly in Cambini and Valletti (2008). Under this assumption Andersson et al. get a result that reverses Jeon (2001) in that, with a large feedback effect, services that are originally substitutes turn into complements if a service with the formerly small number of users becomes large. Applied to the fixed–mobile relationship, the reason for this switch would be that, with a larger number of mobile users, a price reduction in fixed-line (mobile) calling leads to more calls to mobile (fixed) networks, which lead to more return calls from mobile (fixed) networks. With a large number of users this indirect effect can overcome a substitution effect in the opposite direction. Gans et al. (2005) note that at (relatively) low mobile penetration rates most calls involving mobile are FTM and MTF calls.4 With feedback effects this would suggest that mobile operators benefit from a large fixed-line penetration leading to complementarity. Could this mean that at very high penetration rates of mobile relative to fixed subscriptions, a complementarity arises for fixed networks because most calls will be FTM and MTF?5
(H2) The second hypothesis arising from this discussion is that, from a theoretical perspective, mobile and fixed communications services could be complements or substitutes. However, in order for FMS to occur, they should be substitutes. This hypothesis is addressed in demand estimations surveyed in Sections 3.2–3.4.
3 The introduction of feedback effects into telecommunications demand analysis goes back to Larson et al. (1990). While feedback effects may signify the presence of call externalities, there may well exist call externalities without feedbacks and feedback effects may actually be a way to internalize call externalities. 4 FTM = fixed-to-mobile, MTF = mobile-to-fixed. 5 This might explain why, in some LDCs, fixed penetration was pulled along by mobile penetration that went far ahead.
6 Economies of scale (and density) or the lack thereof can explain mobile penetration in two different ways. In poor countries and rural areas of rich countries a lack of economies of scale in mobile relative to fixed networks can mean that mobile networks are less expensive at low density. On the other hand, the presence of economies of scale (and density) in mobile networks would imply cost reductions through increased penetration, while fixed-line penetration is already saturated and hence not subject to further cost reduction.
2.4. Hypotheses from the supply perspective The demand properties discussed so far only explain FMS in conjunction with changes in other relevant variables. Thus, a positive cross-price elasticity of demand for fixed network services with respect to the price of mobile services can explain substitution away from fixed networks only if the price of mobile services falls relative to fixed prices. As usual, it is the interaction of demand and supply that explains market developments and this is no different for FMS. The supply side clearly has a crucial part in explaining the incredible advances of mobile communication in developing countries, where mobile network costs are estimated to be only half those of fixed networks and where the build-out for mobile is much more flexible and faster. This contrasts with wealthy countries, where – at the time of mobile takeoff – full fixed-line penetration had been reached so that, because of sunk costs, the forward-looking costs of fixed networks had to be much lower, although they also provided the terrestrial backbone for the mobile networks. In addition, fixed networks in wealthy countries are cheaper because of high subscriber density. This combination of factors makes FMS in wealthy countries much more of a puzzle. Based on positive cross-elasticities between fixed and mobile services (in both directions) a price decline of mobile relative to fixed services would help explain FMS. On top of that, we would expect that there is a critical price difference between fixed and mobile call charges, at which mobile calling turns from being a complement to being a substitute to fixed calling. A further price decline of mobile relative to fixed would not only shift demand because they are substitutes but also make them more substitutable. The relative price decline of mobile versus fixed services, however, needs explanation in order to help in drawing policy conclusions. Usually, a price decline can be due to unit cost reductions and/or reductions in the price-cost margin. Cost reductions again can be the result of economies of scale/scope and of technical advances.6 Reductions in price-cost margins tend to be the result of increased competition (and sometimes of regulation), which can also lead to cost reductions. Increases in demand elasticities are often also interpreted as increases in competition, although the two may differ. (H3) The third hypothesis is therefore that FMS is driven by a relative price decline of mobile against fixed services, given
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
that the two are substitutes in demand. This hypothesis is addressed in Section 3.5.1. Higher mobile penetration itself can spur FMS through network externality effects and quality improvements including expansion of the scope of mobile services. This means, among others, that the quality-adjusted price for mobile service has decreased. The network effect of mobile penetration is related to the tipping problem. A simple example of this would occur if every household member has a mobile phone, eliminating the need for (common) fixed-line service. A related development is the emergence of calling clubs that can have network effects and can provide benefits from on-net/off-net price discrimination even for firms with small market shares (Birke and Swann, 2006; Gabrielsen and Vagstad, 2008). (H4) The fourth hypothesis is therefore that network effects as well as quality improvements and an increased scope of services have all benefited mobile networks and thereby enhanced FMS. This hypothesis is taken up in Section 3.5.2.
3. Empirical evidence on fixed-to-mobile substitution 3.1. Penetration models for mobile networks The early work relevant for FMS was mostly concerned with the speed and pattern of mobile penetration rather than with its relationship with fixed network development per se, although the latter was considered as one of the explanatory variables. Mobile penetration is mostly characterized by S-shaped epidemic diffusion models (Gruber, 2001; Koski and Kretschmer, 2005). The main diffusion studies related to FMS are summarized in Table 1. In the penetration models, the effects of fixed networks are mostly captured by the influence that the size of fixed networks has on mobile diffusion. As a predecessor to the diffusion studies in Table 1, OECD (1995) demonstrated graphically that, in the second half of the 1980s, the Nordic countries led the OECD as a whole in fixed-line penetration. At that time, mobile penetration in the Nordic countries was already increasing at the expense of fixed-line penetration, whereas this was not the case for the rest of OECD countries. Unsurprisingly therefore Gruber and Verboven (2001) show a significant negative relation between of the number of fixed-lines and mobile penetration in the EU for the later period of 1992–1997. This is in line with Jang et al. (2005), who consider the period 1980–2001 for the OECD and Taiwan and find that the fixed penetration had a strongly negative effect on the diffusion of mobile, suggesting substitutability. In contrast, Gruber (2001), in his analysis of Central and Eastern European countries, finds that larger fixed networks are associated with more rapid diffusion of mobile networks. Hamilton (2003) also finds for the 1985–1997 period a positive correlation between fixed and mobile lines in Africa, which she interprets as complementarity. Koski and Kretschmer (2005) examine the 1991–2000 period for 25 predominantly wealthy countries. They find, among other things, that liberalization of the fixed networks accelerated mobile (2G) penetration. If one interprets ‘‘liberalization of fixed networks” as lower fixed
7
network prices this result would hint at complementarity. However, liberalization may be an indicator of many more differences between economies that would independently influence mobile penetration. For example, Gans et al. (2005) interpret liberalization as a possible supply side effect because investing in fixed networks may be less attractive after liberalization. Few of the penetration models address the reverse effect of mobile network development on fixed networks. Barros and Cadima (2001) allow for feedback from fixedto-mobile networks. They find a negative effect of mobile on fixed network development to be compatible with both fixed and mobile penetration expanding. In particular, a linear trend characterizes fixed network expansion and an exponential trend holds for mobile. Nevertheless, their results suggest that the evolution of the fixed network in Portugal had no bearing on the development of mobile penetration and they find a very modest slow-down of the fixed network diffusion as a result of mobile growth. Relating mobile usage to mobile and fixed-line penetration and to fixed-line prices Grajec and Kretschmer (2009) show fixed and mobile to be complements in usage intensity (negative cross-price effects), with evidence of becoming substitutes with higher mobile penetration (after the inflection point of the diffusion curve). They also find evidence that mobile usage is higher when there is low fixed-line penetration. The diffusion literature generally emphasizes supply side variables in explaining mobile diffusion. The relationship of mobile to fixed networks is analyzed selectively and is largely restricted to the effects of fixed networks on mobile penetration rather than effects in the reverse direction. The penetration models have shown some puzzling differences in individual results, although consistent patterns emerge. According to Banerjee and Ros (2004), in high-income countries the evidence from the literature is quite consistent with the life-cycle hypothesis. According to this hypothesis, fixed and mobile network services are complements at the beginning of mobile diffusion and, around 2000, switch to become substitutes, indicating that the beginning of FMS can be set at about that time.7 From the perspective of FMS, penetration models suffer from three major problems. The first is the necessity of long data series in order to establish a diffusion curve (Madden et al., 2004). This makes it difficult to characterize recent developments and it poses challenges to achieving data consistency. The second problem is that penetration models, in contrast to demand elasticity estimates, are generally non-structural and nonequilibrium in nature. Their main function therefore has been to generate rather than test hypotheses about fixed–mobile relationships. The third problem is that they generally say little about the effects of mobile on fixed penetration. Thus, from the FMS perspective, they have it backwards. The only, admittedly early, work on fixed-line development as a function of mobile penetration found only a mild effect going in that direction (Barros and Cadima, 2001). Also, hardly any work 7 Liikanen et al. (2004) already find some substitutability for the 1992– 1998 period. However, among the 80 countries substitutability dominates ‘‘only very close to the sample maximum of fixed-line penetration”.
8
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
Table 1 Empirical evidence on FMS: diffusion models. Authors
Countries/years
Main explanatory variables
Effect of fixed penetration on mobile
Remarks
Gruber and Verboven (2001) Gruber (2001)
EU, 1984–1997
Negative
Central and Eastern Europe, before 1998
Starting dates of 1G, of 2G and of competition Mobile competition, start date of mobile
Portugal, 1981–1999
Cross-price variables
No effect
80 countries, 1992–1998
1G penetration for 2G penetration, number of 2G technologies used Network effects, spectrum capacity, competition, CPP (+) Explain usage as function of penetration and price variables
Positive for poor countries, negative for rich countries
Large technological effects on mobile penetration Waiting list for fixed-line services increases mobile penetration. Positive effect from mobile to fixed penetration Use 1G diffusion to explain 2G and vice versa
Barros and Cadima (2001) Liikanen et al. (2004)
Jang et al. (2005)
OECD and Taiwan, 1990– 2001
Grajec and Kretschmer (2009)
41 countries, 1998–2004
Positive
Negative
Concentrate on shape of diffusion
Evidence of life-cycle effect in usage relationship between fixed and mobile; platform substitutes: Low fixed-line penetration ? high mobile usage
Emphasize usage heterogeneity in penetration
Table 2 Subscription cross-elasticities I: multiple countries. Author(s)
Country/year
Cross elasticities
Remarks
Mobile subscription Ahn and Lee (1999)
64 countries, ca. 1997
Garbacz and Thompson (2005) Garbacz and Thompson (2007) Heimeshoff (2008)
53 LDCs, 32 DCs, 1996–2001 53 LDCs, 1996–2003 30 OECD countries, 1990–2003
Fixed-line subscription
Positive quantity effects ? Complementarity Negative, turning positive later Positive
Negative
Study only cross-quantity effects; find weak own-price effects Imperfect price data
Negative
Imperfect price data
Positive (+0.94)
Insignificant
Low significance; uses call prices; endogeneity problem
covers time periods where drastic FMS can be observed, i.e. when fixed penetration was actually decreasing. Nevertheless, in my view, the results of penetration models are generally compatible with the existence of links between the rise of mobile and the decline of fixed networks (H1). 3.2. Cross elasticities of demand We now turn to demand estimates that may be relevant for FMS. Taylor (2002) in his literature survey on telecommunications demand laments the lack of non-confidential studies of mobile demand and therefore encourages estimation of the demand relationship between fixed and mobile. Since then, a limited number of empirical estimates have appeared in the literature. Most of these studies, however, refer to just a few data sources and are now 5–10 years old. Studies on both access and on calling demands are considered relevant for FMS and are discussed separately below. 3.2.1. Access The demand for access is derived from demand for usage, where both outgoing and incoming calls and possi-
bly option demands play a role. Substitution at the level of usage can therefore drive substitution at the level of access. There is likely to be a marked difference between households that are subscribers of only one of the two services as opposed to households with both. The latter are assumed to have no switching costs for usage. In discussing cross-elasticity estimates for access we differentiate multi-country from single-country studies. The evidence on multi-country studies is summarized in Table 2. Columns 3 and 4 of Table 2 show quite mixed results, finding both complementarity and substitutability from fixed-to-mobile. Garbacz and Thompson (2005, 2007) suggest that the sign of cross-elasticities between mobile and fixed may actually differ according to which direction the effect is being assessed. In contrast to the multi-country studies, the singlecountry estimates seem to agree on positive cross-elasticities in both directions.8 They are summarized in Table 3.
8 In addition, Reece (2007), using price and telephone tax variations across the US in 2001, finds positive effects of telephone connection charges on mobile penetration, but does not express them as elasticities.
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
9
Table 3 Subscription cross-elasticities II: single countries. Author(s)
a
Country/year
Cross elasticities
Remarks
Mobile subscription
Fixed-line subscription +0.22 to +0.26a
Rodini et al. (2003)
US, 2000–2001
+0.13 to +0.18
Rodini (2009)
US, 1999–2001
+0.223
Ward and Woroch (this issue) Narayana (2008)
US, 1999–2001 India, 2003
+0.25 to +0.31 Positive and very large
Positive and very large
Bill-harvesting data; fixed-line subscription is for second lines Bill-harvesting data; but uses outside price data Uses lifeline program Second line survey data; elasticities hard to interpret
Not statistically significant.
Ward and Woroch (this issue) make use of US low-income subsidy programs, which implicitly cause large changes in fixed subscription prices, to estimate crossprice elasticities between fixed and mobile subscription. These elasticities are larger than those found for second lines in Rodini et al. (2003), who used the same US survey and bill-harvesting data. By their clever use of the support programs, Ward and Woroch were able to generate enough variation in the price data to be able to estimate demand relationships. Lack of price variation is generally a major problem for subscription estimations in single-country studies. In multi-country studies variation is less of an issue, but the quality of the price data is usually quite low. Although Hausman et al. (1993) have established clearly that access decisions empirically depend on subscription as well as usage prices, only a few studies (including Rodini et al. (2003) and Rodini (2009)) use both prices as explanatory variables for subscription decisions. Narayana (2008) finds for a survey sample in India in 2003 that the two price components were correlated and that usage price had a much stronger and more significant effect than subscription prices. His results indicate much stronger substitutability in both directions than estimates in other studies. Most cross-demand studies only estimate one crosselasticity, which fails to address the question if the reverse elasticity is symmetric or, at least, has the same sign. Ward and Woroch (this issue), assuming away income effects, infer the reverse elasticities, but this may not be reliable if income elasticities are sizeable. When the income elasticity is large relative to cross-price elasticity even the signs may differ. As noted above, this occurred in Garbacz and Thompson (2007) but not in Narayana (2008).9 However, large income effects should have less relevance for telephony in wealthy economies than in LDCs. Sung and Lee (2002) is a study that cannot be characterized either as a penetration model or as a cross-elasticity study. They change the emphasis from the relationship between the stock of fixed-lines and mobile subscribers to the substitutability between new fixed-line and mobile subscribers (as do Yoon and Song, 2003). Using separate equations for new fixed connections and fixed disconnections they estimate that a 1% increase in mobile subscribership
3.2.2. Voice usage All the work surveyed on cross-elasticities for calling applies to single countries. Cross effects for fixed call demand are summarized in Table 4. The measured cross-elasticities for fixed call demand are in the +0.2 to +0.6 range.11 Higher mobile penetration should increase the substitutability of fixed and mobile calls (Thompson et al., 2007). This may be playing out in the findings of Briglauer et al. (2009), who estimated cross-price elasticities for fixed-line domestic calling in response to mobile price changes in Austria for the period 2002–2007, using monthly data on minutes of calling and taking average revenues per minute across the country as price data. However, while their best estimates of +0.50 for a long-run cross-elasticity indicate fairly strong substitution and are in line with the large increase in mobile subscriptions in Austria over this period, their estimates for short-run elasticities are substantially smaller and, in some cases, insignificant. Horváth and Maldoom (2002) estimate the effect of mobile subscribership on fixed-line usage based on UK survey data for 1999, 2000 and 2001. In an endogenous switching model they discover strong substitution effects. The later surveys show stronger substitution. Caution is warranted, however, because such survey results do not represent actual but only self-reported consumption and only contain quantity ranges rather than exact quantities.
9 Madden et al. (2004), using panel data for 56 countries 1995–2000, estimate large income elasticities for mobile subscription (about 4.8 overall).
See Bonfrer et al. (2006). The insignificant effect on FTM calls found by Yoon and Song (2003) could result from call-back effects discussed above in Section 2.3.
in Korea led to a 0.10–0.18% reduction in new fixed connections and a 0.14–0.22% increase in fixed disconnections over the period 1991–1998. Prices are not significant in their estimations. Since simultaneity of substitution and complementarity impede identification of fixed–mobile demand relationships, the Sung and Lee approach may be the more promising approach to these issues. Overall, the results on subscription cross-elasticities are not as strong or clear as one might have hoped. The reason could be that estimations of cross-price effects are typically less robust than own price effects regardless of the product or service.10 My assessment is that substitutability now prevails in wealthy countries for both fixed and mobile penetration. There is some empirical evidence that it does so in poorer countries as well.
10
11
10
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
Table 4 Cross elasticities for fixed-line call demand. Author(s)
Country/year
Cross elasticity
Remarks
Yoon and Song (2003)
Korea, 1997–2002
Differentiate FTF, FTM and MTM/MTF
Horváth and Maldoom (2002) Ward and Woroch (2004) Briglauer et al. (2009)
UK, 1999, 2000, 2001 US, 1999–2001 Austria, 2002– 2007
Positive FTF: +0.6 FTM: insignificant Positive +0.22 to +0.33 +0.50 long-run, +0.06 and +0.11 short-run
Empirical work on mobile demand containing fixed-line call prices seems to be restricted to studies for the US by Ingraham and Sidak (2004) and Rodini (2009). They both use the same 1999–2001 bill-harvesting data as Rodini et al. (2003) for a large number of households. Ingraham and Sidak find a very small but significant cross-elasticity of +0.022. Since they use individual households as their unit for price observations, the price variable will likely introduce a bias because of the self selection associated with quantity discounts. In contrast, Rodini’s (2009) small and positive elasticities of mobile usage with respect to the fixed long-distance usage price are all statistically insignificant. He avoids the potential endogeneity from constructing mobile prices from the bill-harvesting data by using outside price data.12 Summing up, cross-price elasticities for calls are consistently positive. There are some indications that they increase over time. However, more work is needed for backing this up. 3.3. Own-price elasticities of demand FMS can be associated with demand shifts (positive cross-elasticities) in which own demand is still inelastic. However, FMS could more likely be a sign of the fixed and mobile markets merging, which would mean that the own elasticity in each ‘‘market” would increase. It is therefore interesting to look at own elasticities as indicators of FMS as well. We start with mobile call and subscription elasticities and then proceed to their fixed-line counterparts. A number of estimates show market demands for mobile calls to be fairly inelastic, ranging from 0.183 for developing countries (e.g. Garbacz and Thompson, 2005, for 1996–2001) to around 0.5 in a wider sample of countries (e.g. Hausman, 1997, for 1993).13 The low level of mar12 Somewhat different from cross-elasticity estimates Grzybowski (2005), using EU price panel data from 1998 to 2002 that include usage and monthly charges, suggests that liberalization of fixed networks leads to lower mobile call charges and increased mobile demand. To the extent that liberalization reduces fixed call prices, this can be interpreted as indicating complementarity of fixed services to mobile services. 13 Garbacz and Thompson (2005) estimates are around 0.5 for developed countries only. On the low side is Rodini (2009) with a pooled estimate of 0.119 for the years 1999–2001. Dewenter and Haucap (2008) found a range of market segment-specific elasticities between 0.2 and 0.7 with prepaid tariffs being the least elastic and business tariffs being the most elastic. They estimate firm-specific demand elasticities for calls between 0.47 and 1.1.
Use survey data; control for selection bias; also find access substitution Use bill-harvesting data; estimate mobile price Average call prices per minute as price variable
ket elasticities found under oligopolistic market structures indicates that mobile calling services form their own market and that the market is fairly competitive (i.e., not close to monopoly pricing).14 Mobile subscription elasticities of 0.313 for 2000 and 0.270 for 2001 derived by Rodini (2009) relate to monthly access charges. The fixed-line access elasticity appears to be generally very low (Hausman et al., 1993), close to zero, but mostly significant. As an exception, it turns out to be essentially zero in Garbacz and Thompson (2005), who only find a significant (short-run) connection charge elasticity of 0.054 for business subscribers to fixed networks. More recently, the fixed access elasticity may have increased, though, due to FMS (Briglauer et al. (2009), find a short-run elasticity of 0.10 and a long-run elasticity of 0.25). Fixed-line elasticities for calls also have tended to stay in the inelastic range. Taylor (2002) reports about 0.50 for long-distance and 0.25 to 0.40 for shorter distances. Ward and Woroch (2004) find them to be 0.3 for intraLATA and 0.7 for interLATA calls during the 1999–2001 period. However, for Austria during 2002–2007, Briglauer et al. (2009) recently estimated a long-run calling elasticity of 1.37 and a short-run elasticity of 0.74. This (together with the high positive cross-elasticity mentioned above) suggests that the advanced FMS experienced by Austria may already have resulted in mobile calling becoming part of the fixed calling market. In summary, there have been many more studies on fixed access and fixed usage elasticities than on the corresponding mobile elasticities. However, the work on fixedlines which shows inelastic demand largely precedes any serious move towards FMS. Elasticities for access are, as would be expected, lower than for calling. While the mobile access elasticity has been larger than the fixed-line access elasticity, both calling elasticities appear to be quite similar.
14 I am thereby discounting the own-price elasticities of 1.12 and 1.29 found by Ingraham and Sidak (2004). Their price variable is the customerspecific total wireless bill amount per subscriber divided by total wireless minutes. It is well known from the literature on non-linear pricing (surveyed, e.g., in Wilson (1993)) that for optional pricing the profitmaximizing average and marginal price declines in the number of minutes. It is not clear that their first-stage estimation of predicted prices eliminates the resulting selection bias. Thus, the large absolute value of elasticities found by Ingraham and Sidak is likely to reflect mostly the self selection of subscribers into ‘bucket’ service plans, where customers with a high propensity for mobile calling choose larger buckets with a lower average price than customers with a lower propensity for mobile calling.
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
3.4. Conclusions on demand elasticity estimates As explained in Section 2.3, telecommunications services have complicated demand properties and are associated with complicated pricing schemes that are hard to capture in quantitative studies. It is therefore not surprising that there are very few convincing elasticity estimates and that the empirical support for FMS from such estimates continues to be quite weak. The demand for telephone services is determined by at least four types of prices: the prices for hardware (telephones/handsets), one-time connection, monthly subscribership, and calling (under CPP and RPP) or being called (under RPP). Not having data on one or more of these prices can, in principle, lead to serious biases in the results. Mobile off-net call charges, for example, may be high because of high termination rates, which in turn help subsidize mobile handsets or reduce monthly subscription charges. As another example, low monthly subscription charges for fixed services may be cross-subsidized by high long-distance charges. If one only has data on subscription charges, a rebalancing may be counted as a price increase and may (for reasons given in Hausman et al. (1993)) lead to unchanged or even increased subscribership. The main problem with price data in multi-country analysis seems to be the aggregate nature of information used compared to the more structural approaches that typically concentrate on a single-country with better micro-data as, for example, Rodini et al. (2003). In aggregate studies, such as Garbacz and Thompson (2005), the authors control for the worst of the unmeasured price variations by employing some control variables in the models – the most general control being fixed (country) effects dummy variables. In their 2007 paper Garbacz and Thompson corrected for possible endogeneity effects in prices, thereby removing some of the unmeasured cost and regulatory differences attributed to prices. Koski and Kretschmer (2005), whose price data are ‘‘average monthly cost of 120 min peak calls,” experimented with other call intensities, monthly subscription rates and cost per three-minute call, finding that all these lead to very similar results. Estimates with imperfect data are therefore all we have. My conclusion from this evidence is that substitution between mobile and fixed services now generally prevails in wealthy countries and that the strength of substitutability is increasing (H2).
3.5. The supply side 3.5.1. Price changes in favor of mobile With positive cross-elasticities between fixed and mobile services (in both directions), a price decline of mobile relative to fixed services would help explain FMS. Although mobile calling prices in the late 1980s and early 1990s did not change much (Jain et al., 1999), price declines in the following years are well documented (Rodini et al., 2003). Particularly strong price reductions have occurred through ‘bucket plans’ for mobile services, which simultaneously encouraged greater use through flat rating within
11
the plan limits.15 Such plans allow customers to choose larger buckets of mobile services (e.g. a monthly allowance of specific types of calls) with a lower average price than customers that have lower propensity for mobile calling will experience. However, to the best of my knowledge, there exists no empirical work relating fixed and mobile service prices to each other and to the relationship between fixed and mobile penetration or usage. Thus, we can only survey literature that deals with parts of these relationships. As explained in Section 2.4, price reductions are the result of cost reductions and/or reduced price-cost margins. We first turn to cost reductions. There appears to be a lack of microeconomic modeling of FMS from a supply perspective, which would include estimates for the movement of costs over time and a characterization of cost structures of the two types of networks. Other features would include cost reductions of mobile relative to fixed networks and of lower costs of mobile networks in low-density areas. All this is not well researched in published papers and is probably available only from studies of analytical cost modeling. The properties of such models have been quite thoroughly analyzed for fixed networks (Gasmi et al., 2002) but not for mobile networks. Although such models are not yet as mature as for fixed networks, it should be possible to derive such relationships. The early work on economies of scale in mobile networks was mixed in that McKenzie and Small (1997) found evidence for constant or even slightly decreasing returns to scale, while Foreman and Beauvais (1999), employing more detailed data found economies of scale, though not to the extent found in fixed networks. The case for constant returns comes from the observation that a doubling of traffic leads to cell splitting and increases the number of cells required even in the same area, roughly doubling costs. However, analytic cost models have identified substantial fixed costs of coverage, which may or not include the sunk costs of acquiring radio spectrum rights. Thus, the cost function appears to be affine linear, costs increase linearly with a constant base that is not affected by the number of connections or call volume. The complementarity found between voice and SMS for mobile (Grzybowski and Pereira, 2008) will hasten the exhaustion of scale economies and lead to lower optimal price-cost margins. Higher penetration and the advancing time could also lead to learning economies, which could be enhanced by the growth of mobile multinationals spreading new mobile networks across countries (Hodge, 2005). Furthermore, technical change has obviously occurred in conjunction with different generations of mobile networks, as shown in penetration models, such as Liikanen et al. (2004) and Koski and Kretschmer (2005). There has been very little work on any reduction in price-cost margins that might be a consequence of competition. Early work on the competitiveness of mobile communications markets has been ambivalent. Parker and Röller (1997) and Busse (2000) estimated collusive effects from mobile operators competing in a multi-market set-
15
See US Dep’t of Justice (2008, p. 63).
12
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
ting, while Crandall and Hausman (2000) found distinct price reductions in a duopoly setting compared to the monopoly outcome, with much weaker price effects coming from further entry.16 While price reductions for mobile services have been dramatic, large price reductions have also occurred for fixed-line services. The latter have been due to privatization, market liberalization and incentive regulation, although these effects are not as clear cut (Sappington, 2002) as the pronounced effect of technical change in the form of digital switching and fiber-optic lines (Spulber and Yoo, 2009). Long-distance services were the main beneficiaries, while local and subscriber charges changed little (Kaserman and Mayo, 2002). In spite of the substantial price reductions for fixed services, the absolute difference to mobile services seems to have decreased or vanished, because mobile prices where higher to begin with. A very different source of (relative) price reductions has in the past arguably been provided by high mobile termination rates that would have translated into lower mobile charges via the waterbed effect. 17 The main argument for the existence of such an effect is based on regulated FTM termination rates, where MTF termination charges under CPP are going to be at or above monopoly levels (Armstrong, 1997; Gans and King, 2000).18 Competition among mobile operators would translate these high rates into incentives for attracting mobile customers, leading to higher mobile penetration (Wright, 2002). At the same time, fixed network customers would be disadvantaged by these conditions, assuming that they do not sufficiently benefit from the increase in mobile subscribership (Thompson et al., 2007).19 Since, in contrast to CPP, RPP is associated with small or vanishing FTM rates, the effect of termination rates could be tested by differences in mobile penetration of CPP vs. RPP countries.20 There is some empirical evidence of this negative effect of RPP on penetration (Jang et al., 2005). However, given the steep decline of termination charges in CPP countries over the last decade it is doubtful that mobile termination charges can explain FMS which seems to be accelerating rather than slowing down. Another factor contributing to price changes in favor of mobile networks may have been the overregulation of fixed relative to mobile networks (Vagliasindi et al., 16 Jain et al. (1999) show that under competition the price of mobile calls may stay high, while the price of handsets falls. 17 For the theory behind the waterbed effect, see Schiff (2008) who defines it as the phenomenon in which ‘regulation of one of the prices charged by a multiproduct firm will result in a change of the firm’s other unregulated price(s).’ (p. 392) and, for its empirical relevance, see Genakos and Valletti (forthcoming). A different view is expressed in Albon (2006) and, to some extent, Littlechild (2006). 18 See, however, Armstrong and Wright (2009) and Binmore and Harbord (2005) for arguments why unregulated termination rates may turn out to be lower. The results also do not hold if fixed termination charges are not regulated. For a more extensive review of the termination-related literature, see the working paper version of this article (available from the author). Also see Laffont and Tirole (2000), Armstrong (2002), Vogelsang (2003) and Littlechild (2006). 19 High mobile-to-mobile (MTM) termination charges can lead to on-net/ off-net price differentials that can spur FMS (Briglauer et al., 2009). 20 The direction of causality between RPP/CPP and termination rates is in dispute. See, for example, Doyle and Smith (1998), Gans et al. (2005), Littlechild (2006), Thompson et al. (2007) and Cambini and Valletti (2008).
2006). In particular, price-squeeze regulation may have prevented fixed-line incumbents from responding to flatrated mobile offerings with their own flat rates. This has been conjectured, for example, for Austria (Kruse, 2007).21 Overregulation would prevent the larger amount of sunk costs in fixed networks (over-capacity) from leading to lower prices.
3.5.2. Nonprice factors explaining FMS While it appears to be obvious that quality improvements of mobile services are contributing substantially to FMS, there seems to be little or no direct literature on this issue. Indirectly, it is covered in some of the penetration work that captures the effects of a move between different generations of mobile networks or the role of standards (for example, Koski and Kretschmer, 2005).22 Digitalization certainly increased service quality and efficiency of spectrum use. However, quality aspects, such as coverage, voice clarity, dependability, speed, functionality (voice mail, caller ID, emergency calls), broadband or additional features such as handsets having a built-in camera, are all very hard to capture in the quantitative work that economists prefer. FMS can also be affected by switching costs between networks. Regulators have tried to reduce switching costs between fixed and mobile and within mobile networks, for example, by requiring mobile number portability. This is consistent with the finding of Grzybowski (2005) that introduction of mobile number portability reduced mobile calling charges. However, mobile number portability also reduces the identifiability of mobile numbers, which under CPP leads to caller ignorance about FTM calling rates (Bühler and Haucap, 2004). This can lead to inefficient calling patterns and to increased FTM termination charges.23 Switching costs from fixed-to-mobile have also been reduced through the move from post-paid contracts to prepaid services. Getting mobile access via prepaid services can be very easy (Hodge, 2005). FMS may be enhanced if one-time fees for installation are lower for wireless than for fixed-lines and if installation is faster. For example, college students move very often, and as mobile-only customers they may remain on mobile after graduation. Moving homes is no problem for mobile subscribers. They simply take their phone with them. Potentially dominating the other non-price factors, network effects may actually be as strong as or stronger than price effects in explaining FMS (Grajek, forthcoming; Doganoglu and Grzybowski, 2007). However, Grajec and Kretschmer (2009) show that network effects may be 21
A similar issue is taken up by Reiffen et al. (2000). Most work on the relationship between fixed and mobile communications either addresses the whole population of users or concentrates on households. An exception is Vagliasindi et al. (2006), whose cross-country analysis emphasizes (small) business mobile subscribership in transition economies. In particular, they are concerned with the role of deficient fixed networks and find that low fixed-line penetration and low fixed line investment spur the advancement of mobile penetration for small enterprises. 23 It is well known that the results of switching costs on competition can be quite ambiguous. See Klemperer (1987). Analogous results could also hold for the effects of switching costs on FMS. 22
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
dominated by rank effects, when it comes to usage. As a result, usage per subscriber may fall with increasing mobile penetration, simply because the later subscribers are lower users. 3.6. Conclusions on explanatory factors In line with hypotheses H3 and H4 the main explanatory factors for the increased FMS in wealthy countries seem to be the demand substitution from large price reductions in mobile relative to fixed services and demand shifts arising from network effects and the relative quality increase of mobile handsets or services. While asymmetric termination rates may have played a role at an intermediate stage of mobile penetration, this role has probably diminished due to FTM termination charge reductions through regulation and possibly through end-users being reachable over several channels. Lower or reduced switching costs may have played some role, as has the stimulating effect of the universal spread of mobile. 4. Policy consequences 4.1. Effects of mobile competition and FMS on fixed network deregulation 4.1.1. Competition between fixed and mobile providers Early on OECD (1995) asked a key question: ‘‘To what extent will mobile communication services eventually compete with, or substitute for, fixed-link telecommunication services in terms of tariffs, traffic and subscribers?” Almost 15 years later, one of the main potential policy effects discussed both in the literature and before regulators is whether FMS is putting pressure on incumbent fixed network providers that should lead to deregulation. The main presumption behind the interest in fixed–mobile competition is either that mobile markets are more competitive overall than fixed networks and that this competition will constrain market power in fixed network markets or, at least, that the combination of fixed and mobile markets will be more competitive than each is individually. US Dep’t of Justice (2008), however, argues that to date FMS has not effectively constrained the prices consumers pay for access to landline services, that wireline carriers have not reduced prices for residential access and that there are no downward price trends where fixed local services have been deregulated. Market convergence is conventionally measured under the hypothetical monopoly test or SSNIP test,24 which answers the question if a hypothetical monopolist in the market to be defined could profitably sustain a price increase (over and above a contestable level) of 5–10% for some period of time (usually 2 years). The smallest entity, for which this can be answered in the affirmative, is the relevant market. This SSNIP test translates into a critical own demand elasticity = 1/(m0 + t), where m0 is the Lerner index at zero profit (capturing economies of scale in a contestable situa24 SSNIP stands for Small but Significant and Non-transitory Increase in Price.
13
tion) and t is the threshold markup (typically 0.05–0.10). In fixed networks m0 can be substantially larger than t. For example, Stumpf (2007) assesses m0 for fixed-line calls at 0.6–0.9 and for fixed-line access at 60.5. Thus, at t = 0.1 the critical demand elasticity for calls would be 1.0 to 1.4 and for access 6 1.7. While, for calls, this elasticity is within reach or may already hold in some countries, such as Austria (Briglauer et al., 2009), it still seems out of the possible range for fixed-line access. Since the regulatory emphasis is on constraining the local market dominance of incumbent fixed carriers, is competition by wireless mobile operators going to be or already constraining the market power of fixed-line incumbents sufficiently (Woroch, 2002)? This property is weaker than being in the same market, because it is testing market dominance of a firm, not market power of a (hypothetical) monopolist. Thus, an incumbent can lack dominance, due to the effect of mobile (as well as fixed-line) competition, although mobile and fixed operators are not in the same market. To the extent that mobile markets are deemed competitive and that mobile competition exerts enough pressure on fixed network operators, market dominance regulation of fixed networks could be revoked. This possibility already seems to hold true in the markets for fixed-line calling in some countries, such as Austria (Briglauer et al., 2009) and can be expected for other countries soon. It is, however, not particularly interesting, because such markets are often already sufficiently competitive, even without the additional pressure of mobile operators (Stumpf, 2007). A more interesting possibility would be if mobile competition was able to sufficiently constrain market power in fixed access networks that end-user access and/ or unbundled local loops and back-hauls could be deregulated. However, currently measured demand elasticities are far below that hurdle (Stumpf, 2007; Briglauer et al., 2009). It appears that continuing regulation of bottleneck services may handicap the fixed networks in their competition with mobile networks, because regulated interconnection and termination charges are based on minutes of calling, while mobile networks move in the direction of flat-rate end-user charges. This causes particular problems if most fixed-line calls are to mobile customers. With mobile penetration levels substantially exceeding fixed-line penetration, mobile ought not to be advantaged by handicapping fixed networks with burdensome regulation or asymmetric termination rates.25 It is questionable if the traditional approach of using long-run average incremental costs for regulating termination rates is still appropriate, given the move to all-IP networks.26 Rather, capacity-based charging and Bill and Keep may be more appropriate (Kennet and Ralph, 2007; Knieps and Vogelsang, 2007). At the same time, fixed networks should not be advantaged by universal service subsidies.
25 Specific policy consequences for mobile termination charges are surveyed in the working paper version of this article, available from the author. 26 For a discussion of cost-based termination rates, see Vogelsang (2003).
14
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
4.1.2. Fixed–mobile convergence Another development threatening the deregulation of fixed networks may come from quadruple play and joint fixed/wireless offerings. They could substantially disadvantage wireless-only or fixed-only carriers. This would, however, only happen if consumers sufficiently preferred such joint offerings to separate fixed and mobile services. In that case and if that required a joint physical setup, one could expect a merger wave of cable and smaller fixed network providers with independent wireless operators. It could, however, cause major problems in countries with no cable TV networks. On the other hand, integrated carriers have less interest in pursuing FMS than independent mobile operators (Zimmerman, 2007, 2008). The ambivalence of competition by wireless affiliates of fixed-line incumbents both in the fixed and mobile markets has become apparent in the attempts of US regulators to incorporate wireless market shares in concentration analysis for the purpose of determining dominant firm status. For example, the FCC has taken an intermediate approach to market definition by adding Verizon mobile customers that had cut the cord to its voice market share and adding all ‘‘cut the cord” customers to the denominator.27 Thus some, but not all mobile subscription was included. Earlier analysis by Loomis and Swann (2005) suggested that the adverse effect of mobile on fixed-lines of incumbents would justify including mobile lines in the market definition. The potential of integrated firms to compete with bundles of fixed and mobile services creates severe problems in evaluating anti-competitive behavior. Their ability to set low on-net calls may be a competitive advantage even if it is not caused by anti-competitive intent (Hoernig, 2007). It may be strongly influenced by large differentials between FTM and MTF termination charges. 4.2. Universal service Universal service is an established public policy goal in many countries. With FMS this goal changes its nature because, to the extent that fixed and mobile services are substitutes, it now has to include mobile services. In particular, if fixed-line penetration shrinks because of FMS it becomes absurd to pursue universal service via fixed lines only.28 At the same time the inclusion of mobile services in a universal service policy can accelerate the pace at which the goal may be achieved and reduce its costs. Fixed-line own-price elasticities of subscription demand are generally lower than corresponding mobile elasticities (Garbacz and Thompson, 2005). This implies that universal service policy should be aimed predominantly at mobile penetration. In addition, to the extent that universal service policy is directed at telephony (and to some extent broadband) in rural areas (and developing countries more generally) its lower cost and the faster deployment of mobile networks may make it cheaper as well. 27
‘‘Verizon 6-MSA Order” released December 5, 2007. Ward and Woroch (this issue) show that fixed-line subsidy programs can lower mobile penetration, thereby canceling the gain in fixed-line penetration. 28
In the past, universal service policy has largely worked through subsidization of end-users for subscription charges for fixed and more recently mobile telephone access and of network operators for the costs of low-density telephone services. This may, for the future, be replaced by broadband access. However, to the extent that universal telephone access is still the objective, it may be advisable to subsidize usage prices as well, because they can have a large effect on the subscription decision (Narayana, 2008). In practice this can be achieved by way of requiring a low-priced or free initial buckets of minutes. 5. Lessons for further research The fourth question of our opening section was ‘‘What are the gaps in and shortcomings of the literature and how can they be remedied?” We answer it in the order of the first three questions. 5.1. What is the empirical evidence on the occurrence, extent and patterns of FMS? The price data in cross-country studies are often not meaningful or are insufficient in that they only provide single prices out of a non-linear or bundled tariff schedule. In contrast, single-country price data may be superior, but generate fewer observations with less variation. Price data for baskets of services, such as provided for the OECD countries, may therefore be superior. Beyond the quality of the data it is particularly disappointing that in view of our initial observation about widespread FMS only one study was able to quantify effects that occurred in the last five years. Late publication dates (like Garbacz and Thompson, 2007) mask the fact that we have very few quantitative analyses of the latest and arguably most dramatic developments. Given the problem with multi-country and aggregate studies, more empirical work with micro-data would provide more insights. Because of their simpler pricing framework, it may also be worthwhile to study consumption behavior under prepaid arrangements. There does not, however, exist anything comparable for fixed networks. The openly accessible empirical literature lags behind the technological and market developments. It has, in my view, not yet captured the full dynamics of FMS. In view of the sizeable numerical reduction in fixed-line subscribers and increase in mobile-only subscribers, one would really like to know if the subscription demand elasticity of the remaining fixed-line users remains low or if these low estimates reflect the situation before the large shift in subscriptions. We also would like to know more about short-run and long-run elasticities and about the relationship between elasticities and the installed base in terms of subscriber penetration. At this point in time, the growth in broadband access saves the fixed network providers from the dismal outlook of losing all their customers. However, with all the exciting new developments in mobile services, fixed-to-mobile broadband access substitution emerges more and more as a possibility, if not a reality (Stumpf, 2007). Empirical
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
work that includes mobile broadband and other mobile data developments and the relationship to fixed broadband and data developments will therefore be needed sooner rather than later. One would particularly want to know if high Internet access is limiting FMS and if broadband mobile is increasing FMS.29 The current measurement of mobile penetration seems to be driven by the desire to reach new records. In my view, the numbers are seriously inflated by inactive users, mostly of prepaid services. Work on the inflation in mobile penetration counts seems warranted. 5.2. What factors explain the evidence on FMS? Since the publication of Reed (1992) detailed information on mobile cost functions has not been well researched in published papers and is available only from studies of analytical cost modeling. Such models can in principle be analyzed for mobile networks in similar ways as Gasmi et al. (2002) have done for fixed networks. A most fascinating question is, to what extent and in what way are broadband and ultra-broadband developments going to constrain FMS and possibly favor fixed– mobile integration? Answering this question could require both theoretical and empirical work. In fact, such constraints may already be at work and may explain why substitutability is not found to be as strong as expected. Although there has been some empirical work on the relationship between RPP/CPP and mobile penetration, there does not appear to exist any work that relates the development of FTM termination charges to FMS. It would be interesting to estimate the effect of those charges on both fixed and mobile penetration. Also, the theoretical analysis of the waterbed effect on subscription and calling prices and on FMS at high mobile penetration should be extended. This could be part of a broader modeling approach to fixed–mobile competition. Since Cremer et al. (1996), no theoretical models seem to have been developed on the market relationship between fixed and mobile operators. Instead of tailor-made models for fixed/mobile competition one could try to adapt well-known models from the literature. For example, the dynamic dominant firm model could be adapted by assuming that the fixed network is (initially) dominant and that the mobile network, as the competitive fringe, expands at a pace depending on the difference between the fixed network price and the mobile networks’ unit costs. Different from conventional dominant firm models, the mobility feature would allow the fringe to expand at a modest pace even if its costs exceed those of the dominant fixed network. Other models of relevance include those for two-sided markets (Armstrong, 2006) and for platform competition. Also, Thompson et al. (2007) suggest a model for competition in usage based on mobility patterns of people.
29 Cross effects between mobile penetration and broadband are contained in Bohlin et al. (this issue).
15
5.3. What are the policy consequences of FMS? As long as fixed and mobile communication services are in separate markets, the market definitions of services within each of them is unaffected by the other. However, if for example mobile and fixed call markets merge, new relationships with fixed access markets emerge that could have competition policy consequences that deserve future research. FMS is already leading to substantial excess capacity in some fixed networks, such as in Austria. This could require a rethinking of the pricing of bottleneck inputs, which is currently based on long-run average incremental costs (Vogelsang, 2003). However, if networks contract rather than expand, this may no longer be the right concept. Any new concept would have to balance efficiency arguments (which favor excluding sunk costs) and viability arguments (which favor historic costs). Such research could also go in the direction of retail-minus approaches like the efficient component pricing rule (ECPR).
Acknowledgements The author would like to thank Carlo Cambini, Ulrich Heimeshoff, Justus Haucap, Tommaso Valletti, Glenn Woroch, and an anonymous referee for insightful and constructive comments on earlier drafts. Late in 2008 I received an anonymous large parcel which I took for a Christmas present. It turned out to be sent by Glenn Woroch and contained papers for this survey that were unavailable on the Web. Many thanks!
References Ahn, H., Lee, M., 1999. An econometric analysis of the demand for access to mobile telephone networks. Information Economics and Policy 11, 297–305. Albon, R., 2006. Fixed-to-mobile substitution, complementarity and convergence. Agenda 13 (4), 309–322. Andersson, K., Foros, Ø., Steen, F., 2009. Text and voice: complements, substitutes or both? Industrial and Corporate Change 18 (6), 1231– 1247. Armstrong, M., 1997. Mobile Telephony in the UK. mimeo, September. Armstrong, M., 2002. The theory of access pricing and interconnection. In: Cave, M.E., Majumdar, S.K., Vogelsang, I. (Eds.), Handbook of Telecommunications Economics. North-Holland Elsevier, Amsterdam, pp. 295–384. Armstrong, M., 2006. Competition in two-sided markets. RAND Journal of Economics 37, 668–691. Armstrong, M., Wright, J., 2009. Mobile call termination. Economic Journal 119 (538), F270–F307. Banerjee, A., Ros, A.J., 2004. Patterns in global fixed and mobile telecommunications development: a cluster analysis. Telecommunications Policy 28, 107–132. Barros, P.P., Cadima, N., 2001. The impact of mobile phone diffusion on the fixed-link network. CEPR Discussion Paper Series No. 2598. Available at:
. Binmore, K., Harbord, D., 2005. Bargaining over fixed-to-mobile termination rates: countervailing buyer power as a constraint on monopoly power. Journal of Competition Law and Economics 1, 49– 72. Birke, D., Swann, G.P., 2006. Network effects and the choice of mobile phone operator. Journal of Evolutionary Economics 16, 65–84. Bohlin, A., Gruber, H., Koutroumpis, P., this issue. Diffusion of new technology generations in mobile communications. Information Economics and Policy 22, 51–60.
16
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17
Bonfrer, A., Berndt, E.R., Silk, A., 2006. Anomalies in estimates of crossprice elasticities for marketing mix models: theory and empirical test. NBER Working Paper 12756, Cambridge, MA, December. Briglauer, W., Schwarz, A., Zulehner, C., 2009. Is fixed–mobile substitution strong enough to de-regulate fixed voice telephony? Evidence from the Austrian markets. Mimeo. Bühler, S., Haucap, J., 2004. Mobile number portability. Journal of Industry, Competition and Trade 4, 223–238. Busse, M.R., 2000. Multimarket contact and price coordination in the cellular telephone industry. Journal of Economics and Management Strategy 9, 287–320. Cambini, C., Valletti, T.M., 2008. Information exchange and competition in communications networks. Journal of Industrial Economics 56 (4), 707–728. Crandall, R., Hausman, J., 2000. Competition in US telecommunications services: effects of the 1996 legislation. In: Peltzman, S., Winston, C. (Eds.), Deregulation of Network Industries: What’s Next? AEIBrookings Joint Center for Regulation Studies, Washington, DC. Cremer, H., Ivaldi, M., Turpin, E., 1996. Competition in access technologies. IDEI Working Papers 60, Institut d’Économie Industrielle. Dewenter, R., Haucap, J., 2008. Demand elasticities for mobile telecommunication in Austria. Jahrbücher für Nationalökonomie und Statistik 228 (1), 49–63. Doganoglu, T., Grzybowski, L., 2007. Estimating network effects in mobile telephony in Germany. Information Economics and Policy 19, 65–79. Doyle, C., Smith, J.C., 1998. Market structure in mobile telecoms: qualified indirect access and the receiver pays principle. Information Economics and Policy 10, 471–488. Foreman, R.D., Beauvais, E., 1999. Scale economies in cellular telephony: size matters. Journal of Regulatory Economics 16, 297–306. Gabrielsen, T.S., Vagstad, S., 2008. Why is the on-net traffic cheaper than off-net traffic? Access markup as collusive device? European Economic Review 52, 99–115. Gans, J.S., King, S.P., 2000. Mobile network competition, customer ignorance and fixed-to-mobile call prices. Information Economics and Policy 12, 301–327. Gans, J.S., King, S.P., Wright, J., 2005. Wireless communications. In: Majumdar, S.K., Vogelsang, I., Cave, M.E. (Eds.), Handbook of Telecommunications Economics, vol. 2. North-Holland Elsevier, Amsterdam, pp. 241–285. Garbacz, C., Thompson Jr., H.G., 2005. Universal telecommunication service: a world perspective. Information Economics and Policy 17, 495–512. Garbacz, C., Thompson Jr., H.G., 2007. Demand for telecommunication services in developing countries. Telecommunications Policy 31, 276– 289. Gasmi, F., Kennet, D.M., Laffont, J.-J., Sharkey, W.W., 2002. Cost Proxy Models and Telecommunications Policy. MIT Press, Cambridge, Mass, and London. Genakos, C., Valletti, T., forthcoming. Testing the ‘waterbed’ effect in mobile telephony. Journal of the European Economic Association. Grajec, M., Kretschmer, T., 2009. Usage and diffusion of cellular telephony, 1998–2004. International Journal of Industrial Organization 27, 238– 249. Grajek, M., forthcoming. Estimating network effects and compatibility: evidence from the polish mobile market. Information Economics and Policy. doi:10.1016/j.infoecopol.2009.07.002. Gruber, H., 2001. Competition and innovation: the diffusion of mobile telecommunications in Central and Eastern Europe. Information Economics and Policy 13, 19–34. Gruber, H., Verboven, F., 2001. The diffusion of mobile telecommunications services in the European Union. European Economic Review 45, 577–588. Grzybowski, L., 2005. Regulation of mobile telephony across the European Union: an empirical analysis. Journal of Regulatory Economics 28, 47– 67. Grzybowski, L., Pereira, P., 2008. The complementarity between calls and messages in mobile telephony. Information Economics and Policy 20 (3), 279–287. Hamilton, J., 2003. Are main lines and mobile phones substitutes or complements? Evidence from Africa. Telecommunications Policy 27, 109–133. Hausman, J.A., 1997. Valuing the effect of regulation on new services in telecommunications. Brookings Papers on Economic Activity Microeconomics, 1–38. Hausman, J., Tardiff, T., Belifante, A., 1993. The effects of the breakup of AT&T and changes in telecommunications regulation: what are the lessons? American Economic Review 83, 178–184.
Heimeshoff, G.U., 2008. Essays in Telecommunications Economics. Unpublished Doctoral Dissertation, Friedrich-Alexander-Universität Erlangen-Nürnberg. Available at: . Hodge, J., 2005. Tariff structures and access substitution of mobile cellular for fixed line in South Africa. Telecommunications Policy 29, 493–505. Hoernig, S., 2007. On-net and off-net pricing on asymmetric telecommunications networks. Information Economics and Policy 19, 171–188. Horváth, R., Maldoom, D., 2002. Fixed–mobile substitution: a simultaneous equation model with qualitative and limited dependent variables. DotEcon DP No. 02/02, London, August. Ingraham, A.T., Sidak, J.G., 2004. Do states tax wireless services inefficiently? Evidence on the price elasticity of demand. Virginia Tax Review 24, 249–261. ITU, 2003. Mobile Overtakes Fixed: Implications for Policy and Regulation. Available at: . ITU, 2008. Worldwide mobile cellular subscribers to reach 4 billion mark late 2008. Available at . Jain, D., Muller, C.E., Vilcassim, N.J., 1999. Pricing patterns of cellular phones and phonecalls: a segment-level analysis. Management Science 45, 131–141. Jang, S.-L., Dai, S.-C., Sung, S., 2005. The pattern and externality effect of diffusion of mobile telecommunications: the case of the OECD and Taiwan. Information Economics and Policy 17, 133–148. Jeon, S., 2001. Telephone demands in two-way interconnected networks. Journal of Economic Theory and Econometrics 7, 1–28. Kaserman, D.L., Mayo, J.W., 2002. Competition in long-distance markets. In: Cave, M.E., Majumdar, S.K., Vogelsang, I. (Eds.), Handbook of Telecommunications Economics, vol. 1. North-Holland Elsevier, Amsterdam, pp. 509–561. Kennet, D.M., Ralph, E.K., 2007. Efficient interconnection charges and capacity-based pricing. International Economics and Economic Policy 4, 135–158. Klemperer, P., 1987. The competitiveness of markets with switching costs. RAND Journal of Economics 18, 138–150. Knieps, G., Vogelsang, I., 2007. Digital economy and regulatory issues. Introduction. International Economics and Economic Policy 4, 101– 107. Koski, H., Kretschmer, T., 2005. Entry, standards and competition: firm strategies in the diffusion of mobile telephony. Review of Industrial Organization 26, 89–113. Kruse, J., 2007. 10 Jahre Telekommunikationsliberalisierung in Österreich. Schriftenreihe der Rundfunk und Telekom Regulierungs-GmbH, vol. 2, Vienna. Laffont, J.-J., Tirole, J., 2000. Competition in Telecommunications. MIT Press, Cambridge, MA. Larson, A., Lehman, D., Weisman, D., 1990. A general theory of point-topoint long distance demand. In: de Fontenay, A., Shugard, M.H., Sibley, D.S. (Eds.), Telecommunications Demand Modeling. NorthHolland, Amsterdam. Liikanen, J., Stoneman, P., Toivanen, O., 2004. Intergenerational effects in the diffusion of new technology: the case of mobile phones. International Journal of Industrial Organization 22, 1137–1154. Littlechild, S.C., 2006. Mobile termination charges: calling party pays versus receiving party pays. Telecommunications Policy 30, 242–277. Loomis, D.G., Swann, C.B., 2005. Intermodal competition in local telecommunications markets. Information Economics and Policy 17, 97–113. Madden, G., Coble-Neal, G., Dalzell, B., 2004. A dynamic model of mobile telephony subscription incorporating a network effect. Telecommunications Policy 28, 133–144. McKenzie, D.J., Small, J.P., 1997. Econometric cost structure estimates for cellular telephony in the United States. Journal of Regulatory Economics 12, 147–157. Narayana, M.R., 2008. Substituability between mobile and fixed telephones: evidence and implications for India. CIRJE, University of Tokyo, Discussion Paper CIRJE-F-550. Available at: . OECD, 1995. Mobile and PSTN Communication Services: Competition or Complementarity? Organisation for Economic Co-Operation and Development, Paris, OCDE/GD(95)96, Paris. Parker, P.M., Röller, L.H., 1997. Collusive conduct in duopolies: multimarket contact and cross-ownership in the mobile telephone industry. RAND Journal of Economics 28, 304–322. Reece, W., 2007. Household mobility and cellular telephone demand. Applied Economics Letters 14, 321–326.
I. Vogelsang / Information Economics and Policy 22 (2010) 4–17 Reed, D., 1992. Putting It All Together: The Cost Structure of Personal Communications Services. FCC Office of Plans and Policy, OPP Working Paper No. 28, Washington, DC. Available at: . Reiffen, D., Schumann, L., Ward, M.R., 2000. Discriminatory dealing with downstream competitors: evidence from the cellular industry. The Journal of Industrial Economics 48, 253–286. Rodini, M., 2009. A Discrete/Continuous Model of Mobile Telephone Demand using Household Data. Unpublished Dissertation, University of California, Berkeley. Rodini, M., Ward, M., Woroch, G., 2003. Going mobile: substitution between fixed and mobile access. Telecommunications Policy 27, 457–476. Rohlfs, J., 1974. A theory of interdependent demand for telecommunications service. Bell Journal of Economics and Management Science 5, 16–37. RTR, 2008. RTR Telekom Monitor, 4. Quartal 2008. Rundfunk & Telekom Regulierungs-GmbH, Vienna. Sappington, D.E.M., 2002. Price regulation. In: Cave, M.E., Majumdar, S.K., Vogelsang, I. (Eds.), Handbook of Telecommunications Economics, vol. 1. North-Holland Elsevier, Amsterdam, pp. 224–293. Schiff, A., 2008. The ‘waterbed effect’ and price regulation. Review of Network Economics 7 (3), 392–414. Spulber, D.F., Yoo, C.S., 2009. Networks in Telecommunications. Cambridge University Press, Cambridge, UK. Stumpf, U., 2007. Regulatory Approach to Fixed–Mobile Substitution, Bundling and Integration. WIK Discussion Paper No. 290, Bad Honnef, March. Sung, N., Lee, Y.-H., 2002. Substitution between mobile and fixed telephones in korea. Review of Industrial Organization 20, 367–374. Taylor, L.D., 2002. Customer demand analysis. In: Cave, M.E., Majumdar, S.K., Vogelsang, I. (Eds.), Handbook of Telecommunications Economics, vol. 1. North-Holland Elsevier, Amsterdam, pp. 97–142.
17
Thompson, H., Renard, O., Wright, J., 2007. Mobile termination. In: Dewenter, R., Haucap, J. (Eds.), Access Pricing: Theory and Practice. Elsevier, Amsterdam, pp. 277–302. US Dep’t of Justice, 2008. Voice, Video and Broadband: The Changing Competitive Landscape and Its Impact on Consumers. Available at: . Vagliasindi, M., Güney, I., Taubman, C., 2006. Fixed and mobile competition in transition economies. Telecommunications Policy 30, 349–367. Vogelsang, I., 2003. Price regulation of access to telecommunications networks. Journal of Economic Literature XLI, 830–862. Ward, M., Woroch, G., 2004. Usage Substitution between Mobile Telephone and Fixed Line in the US CRTP Working Paper, October. Ward, M., Woroch, G., this issue. The effect of prices on fixed and mobile telephone penetration: using price subsidies as natural experiments. Information Economics and Policy 22, 18–32. Wilson, R., 1993. Nonlinear Pricing. Oxford University Press, New York and Oxford. Woroch, G., 2002. Local network competition. In: Cave, M.E., Majumdar, S.K., Vogelsang, I. (Eds.), Handbook of Telecommunications Economics, vol. 1. North-Holland, Elsevier, Amsterdam, pp. 641–716. Wright, J., 2002. Access pricing under competition: an application to cellular networks. Journal of Industrial Economics 50, 289–316. Yoon, C.-H., Song, Y.-W., 2003. Telecom development in Korea: substitution and integration of fixed–mobile services and regulatory implications. Communications and Strategies 52 (4th Quarter), 257– 270. Zimmerman, P.R., 2007. Recent developments in US wireline telecommunications. Telecommunications Policy 31, 419–437. Zimmerman, P.R., 2008. Strategic incentives under vertical integration: the case of wireline-affiliated wireless carriers and intermodal competition in the US. Journal of Regulatory Economics 34, 282– 298.