Deand estiation and arket definition for broadband internet services 1 Mélisande Cardona, 2 Anton Schwarz, 3,4 B. Burcin Yurtoglu, 5 Christine Zulehner 5 July 2007 Abstract This paper analyses residential deand for internet access in Austria with a focus on broadband internet connections. Austria has a cable network coverage of about 50% and is, therefore, a good candidate to analyse the elasticity of deand for DSL where cable is available and where it is not. We also include obile broadband via UMTS or HSDA in our analysis. We estiate various nested logit odels and derive conclusions for arket definition. The estiation results suggest that the deand for DSL is elastic and that cable networks are likely to be in the sae arket as DSL connections both at the retail and at the wholesale level. Keywords: Estiation of discrete choice odels, broadband internet access, arket definition, regulation JEL classifications: L51, L96 1 2 3 4 5 We thank Florian Heiss, Frank Verboven and Klaus Gugler for useful coents and suggestions. All errors are our own. Ludwig-Maxiilians-University Munich, Schackstr. 4/III, D-80539 München Austrian Regulatory Authority for Broadcasting and Telecounications (RTR), Mariahilfer Straße 77-79,1060 Vienna, Austria. All views expressed are solely the author s and do not bind RTR or the Teleko-Control-Koission (TKK) in any way nor are they official position of RTR or TKK. Corresponding Author. E-ail: anton.schwarz@rtr.at, Tel.: +43 (0) 1 58058-609, Fax: +43 (0) 1 58058-9609 University of Vienna, Departent of Econoics, Brünner Straße 72, 1210 Vienna, Austria 1
1. Introduction The 2003 regulatory fraework for electronic counications networks and services 6 of the European union requires the national regulatory authorities (NRAs) of individual countries to periodically review a nuber of electronic counications arkets which potentially ay be subject to ex ante regulation. One of these arkets is the arket for wholesale broadband access. According to the recoendation on relevant arkets, 7 [t]his arket covers bitstrea access that perit the transission of broadband data in both directions and other wholesale access provided over other infrastructures, if and when they offer facilities equivalent to bit-strea access. Wholesale broadband access, also called bitstrea access (in particular if realised over a copper network), is a wholesale product which allows alternative operators to offer broadband internet access to the final consuer without having an own access line. The alternative operator receives traffic at a higher network level (e.g. an ATM node) and will forward this traffic to the public internet. It is usually anages the custoer relation, provides the internet connectivity, e-ail addresses and web-space, and can influence the quality of service (e.g. by setting the overbooking factor). 8 According to the 2003 regulatory fraework, NRAs are required to periodically analyse the state of copetition on the wholesale broadband access arket. If an undertaking is found to have significant arket power (SM), 9 NRAs have to ipose appropriate ex ante reedies 10 6 See Directives 2002/19/EC, 2002/20/EC, 2002/21/EC and 2002/22/EC, OJ L108, 24.4.2002. 7 Coission Recoendation of 11 February 2003 on relevant product and service arkets within the electronic counications sector susceptible to ex ante regulation in accordance with Directive 2002/21/EC of the European arliaent and of the Council on a coon regulatory fraework for electronic counication networks and services, OJ L 114/45. ( Recoendation on Relevant Markets ). 8 For details on bitstreaing see ERG (2005). 9 The concept of SM is based on the concept of doinance in general copetition law (see European Coissions Guidelines on arket analysis and the assessent of significant arket power under the Counity regulatory fraework for electronic counications networks and services ( SM-Guidelines Official Journal 2002/C 165/03)). 2
to prevent anti-copetitive or exploitative abuses. Before the process starts, however, a relevant arket has to be defined. The Recoendation on Relevant Markets quoted above is a starting point, however, NRAs have to check which products (and geographic areas) exactly to include or whether their national circustances are such that they have to deviate fro the Recoendation. The instruent applied to define arkets is, as in general copetition law, the hypothetical onopolist test (HM-Test). This test asks whether, starting fro the copetitive level, a nontransitory 5-10% price increase would be profitable for a hypothetical onopolist in the arket under consideration. The sallest set of products for which the price increase can be sustained constitutes the relevant arket. 11 Market definition of wholesale broadband access arkets has attracted soe attention since Oftel (now Ofco) and Coreg (the NRAs of the UK and Ireland) have notified their decisions to the European Coission in 2003 and 2004. 12 One of the ain questions was whether access via cable networks (CATV-networks) fors part of the sae arket as access via copper networks (digital subscriber line - DSL). Whereas DSL wholesale products (bitstrea acess) provided by the incubent telecounications operator (in ost cases due to regulatory obligations or regulatory pressure) are available in any EU Meber States, 13 wholesale broadband access via cable networks is only provided rarely. Therefore, a direct copetitive constraint fro cable on DSL at the wholesale level is unlikely to exist. 14 10 Ex ante reedies available to NRAs are listed in the access directive (Directive 2002/19/EC) and include obligations of access, non-discriination, price control, accounting separation, and transparency. 11 For a description of the HM-Test see, for exaple, Bishop/Walker (1999), OFT (2001), and 49 et sqq. of the SM-Guidelines. 12 See Oftel (2003) and Coreg (2004). Decision on arket definition and arket analysis have to be notified to the European Coission which has a veto power. 13 See ERG (2005) pp. 9-11. 14 See however the notification of the Maltesian NRA, MCA (2006) which argues that there would be sufficient wholesale substitution if cable networks offered a wholesale product. This notification has been withdrawn later on, however. 3
However, as Ofco, Coreg and later a nuber of other NRAs (although not all of the) argued, there is an (indirect) constraint fro cable on DSL via the retail level. The arguent is that a hypothetical onopolist for DSL access could not increase his bitstrea prices profitably by 5-10% as this would also increase retail prices which would ake custoers switch fro DSL to cable access at the retail level. This would also reduce access deand and if retail substitution is strong enough, the price increase would not be profitable. The elasticity of retail deand is therefore crucial not only for the definition of retail arkets but also for the definition of the wholesale broadband access arket. 15 Only few papers have analysed the extent of retail deand elasticities for broadband internet services so far. Rappoport et al. (2002) use a nested logit discrete choice odel to describe the deand for internet access of residential custoers in the US. They conclude that deand for DSL is elastic (own price elasticity of -1.462) and that therefore DSL and cable belong to the sae retail arket. Crandall et al. (2003) confir these results (DSL own price elasticity of -1.184). Ida and Kuroda (2006) estiate a siilar odel for Japan including fibre (FTTH) a rapidly growing access technology in Japan in their choice set. They conclude that deand for DSL (at this tie the ain access technology with a share of 75%) is inelastic (own price elasticity of -0.846) but deand for cable and FTTH is elastic (own price elasticities of -3.150 and -2.500). They also find that the upper and lower end of the DSL arket (very high and low bandwidths) are highly elastic as they directly copete with FTTH and cable on the high end and dial-up and ISDN (narrowband) on the low end. 16 This paper analyses residential deand for internet access in Austria with a focus on broadband internet connections. Like the studies quoted above, we use a nested logit discrete choice odel, however, there are several new features: (i) To our knowledge, this is the first analysis for an European country. Austria has a cable network coverage of about 15 See also Inderst and Valletti (2007) and Schwarz (2007) 16 Another recent study by Goel et al. (2006) reports inelastic deand for internet services in OECD countries using ordinary least squares estiation. 4
50% and therefore is a good candidate to analyse the elasticity of deand for DSL where cable is available and where it is not. (ii) We include obile broadband via UMTS or HSDA in our analysis. These broadband services, offered by obile operators, are available for 2-3 years and experienced high growth rates in 2006 with the introduction of HSDA, which currently allows download speeds of (theoretically) up to 7.2 Mbit/s. The total nuber of residential users is still liited but there appears to be potential for becoing a substitute to fixed broadband in the future. (iii) In addition to several two-level nested logit odels, we also use a three-level specification. (iv) We are able to derive conclusions for arket definition by calculating the effects of a 5-10% price increase of wholesale broadband access products on the profits of a hypothetical onopolist. The rest of the paper is structured as follows: Section 2 gives an overview of the Austrian retail and wholesale arket for broadband access including the current status of regulation. Section 3 describes the data and discusses the question how to allocate non-chosen alternatives to households. Section 4 describes epirical odels of consuer behaviour. Section 5 discusses the estiation ethod and the odel selection. It further presents the estiation results for the area where all four types of internet access (DSL, cable, obile and narrowband) are available and for an area where only DSL and narrowband are available. Section 6 discusses the iplications for arket definition. Section 7 suarizes and concludes. 2. The Austrian arket for broadband internet services By the end of 2006, 52% of all private households in Austria had an internet connection. While narrowband connections (dial-up and ISDN) still have a significant share (19% of all households in 2006), the share of broadband connections is increasing rapidly, fro 10% in 5
2002 to 33% in 2006. 17 The Austrian arket for broadband internet services therefore is like in any other countries - characterised by high and steady growth. Broadband internet via cable networks becae available in 1996 and DSL followed in 1999. By the end of 2006 there were 1.36 Mn. broadband connections, ore than 1 Mn. of which were held by residential custoers. The broadband penetration in Austria was 39% (households and businesses) and slightly above the EU average. 18 The cable network coverage is approxiately 50% of all households and relatively high copared to ost other EU countries. 19 There are ore than 100 cable network operators which offer broadband services in different regions of Austria (cable networks usually do not overlap), however, aost 90% of all cable connections are offered by six bigger operators. The DSL coverage is, like in ost other EU countries, above 90% of all households. In Septeber 2006, the shares of the different infrastructures on the arket for fixed broadband connections were as follows: DSL: 61%, cable: 37%, other (fixed wireless access, fibre): 2%. 20 While ost business users prefer DSL over cable, cable still holds a strong position in the residential arket. Most DSL and cable operators offer a enu of three to five (and soeties ore) tariffs, which vary by price, download speed and download volue. Since 2003, obile broadband via UMTS and since 2006, obile broadband via HSDA is available. By the end of 2006, there were about 220,000 obile broadband connections, ore than half of which were used by business custoers often copleentary to fixed access. Residential custoers on the other hand see to use obile broadband rather as a substitute than a copleent. Mobile broadband via HSDA is usually available in cities with 17 See Statistik Austria (2006). 18 See RTR (2007). 19 Exceptions are the Netherlands, Belgiu, Luxebourg and Switzerland with aost full cable coverage. 20 See RTR (2007). 6
ore than 5,000 inhabitants where in ost cases also cable networks exist. Hereby, Austria is one of the leading countries in the deployent of obile broadband services via UMTS/HSDA. Mobile tariffs are designed soewhat differently fro fixed network tariffs insofar as price does not vary with download speed but only with download volue. While on-off connection fees are usually lower than in the fixed network, volue is still ore expensive. There are two regulated wholesale products based on which alternative operators, which do not own infrastructure all the way to the final consuer, can enter the retail arket: local loop unbundling (LLU) and bitstrea access (described in section 1), both of which allow alternative operators to offer DSL connections on the retail arket. While ore own infrastructure is needed for local loop unbundling there is also ore value added and there are ore degrees of freedo for designing the products (e.g. bundles with voice telephony, etc.). 3. Data and descriptive statistics The data we use is fro a survey coissioned by RTR (the Austrian National Regulatory Authority) which was conducted in Noveber 2006. 4,029 households were interviewed about the type and characteristics of the internet connection they use as well as their onthly expenses. Individual specific data such as age, education and household size were also collected. After eliinating issing and iplausible values we are left with aost 3,000 observations. For the estiation, these observations are divided into two sub-saples: One for the area where all four internet access technologies (DSL, cable, obile, and narrowband) are available and another one for the area where only DSL and narrowband are 7
available. 21 The nuber of observations is reported in Table 1. The share of the different types of access in the saple is unbalanced. There are too any DSL households and too few narrowband and cable households. We therefore use weights in order to correct for that. The weights are taken fro the distribution between the different access types fro a icro census carried out by the central bureau of statistics (see Statistik Austria (2006)). Table 1: Nuber of observations Nuber of observations DSL cable obile narrowb. no internet total total unweigted 644 234 29 236 1682 2825 total weighted 375 278 34 452 1650 2789 cable/obile area unweighted 387 234 29 139 964 1753 cable/obile area weighted 229 278 34 269 968 1778 dsl/narrowband area unweighted 257 0 0 97 718 1072 dsl/narrowband area weighted 146 0 0 183 682 1011 Table 2 reports the descriptive statistics of the ain product-specific variables which we use in the analysis. Table 2: Descriptive Statistics product specific variables ean std. dev. in ax price in per onth DSL 31.73 9.40 9.90 73.00 cable 40.73 13.87 19.00 75.00 obile 32.20 10.24 9.50 59.00 narrowband 18.98 12.94 4.00 60.00 download speed in kbit/s DSL 1,365 999 210 6,144 cable 3,180 2,492 128 16,384 obile 900 0 900 900 narrowband 56 0 56 56 download volue included in MB (for non-flat rate products) DSL 1,561 2,418 250 21,000 cable 4,187 5,494 400 20,000 obile 777 844 250 4,000 narrowband 0 0 0 0 share of flat rate products (duy_flat) DSL 8.40% cable 57.80% obile 0.00% narrowband 0.00% 21 Data on cable network and DSL coverage are available fro the operators. Data on obile coverage are not available. However, it can be concluded fro the operators press releases that HSDA coverage by end of 2006 was available in urban areas where usually also cable networks are present. We therefore use the cable network coverage as a proxy for the obile broadband coverage. 8
As can be seen, users on average spend less on DSL than on cable products, however, the DSL products coe on average with lower speed and volue. Cable products are also uch ore frequently bought with flat rate (57.80%) than DSL products (8.40%). For obile products it is difficult to deterine a download rate as this depends on the nuber of users in the cell. We have taken 900 kbit/s as a axiu value a consuer could expect to get by end of 2006. The included volue for obile broadband is uch lower than for fixed broadband connections. Individual specific variables are reported in Table 3. Table 3: Individual specific variables DSL cable obile narrowb. no internet age (head of household - ean) 44.90 42.67 40.97 45.54 60.50 household size (ean) 3.1 2.7 2.7 3.1 2.0 education: copulsory school 37% 28% 40% 30% 70% education: high school without graduation 22% 14% 21% 25% 17% education: high school with graduation 25% 33% 26% 29% 10% education: university degree 15% 25% 12% 16% 3% gender: feale 45% 42% 43% 54% 68% The discrete choice approach we use for the analysis of deand requires us to allocate each household all four internet access types, which iediately raises the question of how to deterine the price and characteristics of the non-chosen alternatives. This is a crucial point as it can significantly influence the result of the analysis. We describe our approach by type of internet access. Narrowband: Narrowband is ost difficult to atch as it is quite hard to say how uch a broadband or no internet household would spend on narrowband services which are etered by inute. Narrowband expenses are also hard to explain by the individual specific variables available, however, they see to vary with region and age. We therefore for eight 9
groups (four regions cobined with age saller or larger than 50), calculate group averages and ipose those on broadband and no internet households. 22 Cable: We use the tariffs fro the web pages of seven big cable operators which cover all of the nine federal states of Austria. 23 We assue that saller cable operators (which have very low arket shares in total) charge siilar prices as the bigger local operator we use. No internet households and narrowband households which spend less than 25 per onth are assigned the low user package of the local operator. Narrowband households which spend ore than 25 per onth are assigned the ediu package. DSL households are assigned the package which is closest in bandwidth and obile households are assigned the package closest in price. DSL: DSL is basically done in the sae way as cable. We use the tariffs fro the web pages of the largest three DSL operators. 24 Geographic differences result fro two operators which offer DSL access based on local loop unbundling in different parts of Austria. Where only Teleko Austria the incubent DSL operator is present, we take the packages of Teleko Austria. Where also one or two of the LLU operators are present, we take average values (weighted by national arket shares). For both DSL and cable there is a proble with households which are assigned a non-flat rate tariff. It can be assued that these households on average have to pay an additional aount per onth for exceeding their included download volue. We solve this proble by coparing actual aounts paid to onthly fixed charges for DSL and cable households over several groups of included download volue. We find that the ean difference between actual aounts paid and onthly fixed charges is significantly different fro zero for low 22 This causes an endogeneity proble as pointed out by Ida and Kuroda (2006, footnote 9). However there does not see to be a good alternative and since we use this ethod only for narrowband we think that this should not be too probleatic. 23 These operators are: UC, LIWEST, Salzburg AG, kabelsignal, B.net, Telesyste Tirol and Cableco. 24 These operators are Tleko Austria, UC/Inode and Tele2UTA. 10
download volues. The result of the t-test is depicted in Table 4. We use the ean as arkup for atched DSL and cable products if it is significantly different fro zero at least at the 10% level. Table 4: Results of testing the null hypothesis H 0 : Difference between actual paid aounts and onthly fixed charges for DSL and cable = 0 volue obs. ean std. err. t-value <500 268 4.003 0.565 7.083*** [500, 1000] 233 4.903 0.693 7.071*** [1000, 5000] 100 1.910 1.031 1.853* >5000 184-0.314 0.997-0.315 flat rate 152-0.414 1.094-0.378 ***, ** and * denote significance on 1%, 5% and 10% level Mobile: Mobile products are assigned according to the onthly charge which is currently paid by the household. We use averages of the prices of the four obile operators weighted by arket share. 4. Epirical odels of consuer behaviour To epirically analyse consuer behaviour, we assue a rando utility odel of internet access in which consuers choose fro a set of five choices. These are no internet access, dialup internet access, cable network, DSL and obile internet access. 25 The utility a consuer derives fro a particular product depends on characteristics of that consuer and on the characteristics of the product. To account for characteristics that are unobserved by the econoetrician, the utility of consuer i for product j is of the for (1) Uij = V ij + ε ij, where i and j are the indices for consuer i, i=1, I, and product j, j=1, J, and where the ter Vij reflects the deterinistic part of consuers utility. The error ε ij is a residual that 25 These choices are available in area 1. Area 2 is odelled analogously. 11
captures for exaple, the effects of uneasured variables or personal idiosyncrasies. It is assued to follow an extree value distribution of type I. Consuers are assued to purchase that internet access that gives the the highest utility. The probability ij that consuer i purchases product j is equal to the probability that Uij is larger than the utility consuer i experiences fro any other product, i.e. j. This probability is equal to U ij > U ij' for all j' (2) = U > U j' j] = [ ε ε V V j' j] ij [ ij ij' ij' ij ij ij' Under the assuption that ε ij follows an extree value distribution of type I, the probability ij has a closed for solution (McFadden, 1974). It is equal to (3) ij e ij =. Vij ' e j' V This is the well-known conditional logit odel. Within this odel, we have to assue independence of irrelevant alternatives (IIA). To additionally odel correlations between choices, nested logit odels have been developed. 26 We use two- and three-layer specifications to odel the choice of internet access. Since the two-layer choice probabilities can readily be found in the literature, 27 we only report choice probabilities (and elasticities see Appendix B) for the three-layer odel. In a nested logit odel with three layers, the probability ij that consuer i purchases product j is equal to 26 See, for exaple, Maddala (1983) or Greene (2003). 27 See, for exaple, Train (2002) 12
(4) ij = i il ij where we index the first layer alternatives as and, the second layer alternatives as l and l, and the third layer alternatives as j and j. The probability ij is equal to (5) ij e = e j' V ij / λ Vij ' / λ. The probability il is equal to (6) V e e il = V / λ + λ IV l ' / λ + λ IV / λ l ' l ' l ' / λ with the inclusive value IV to be equal (7) IV = lne j' V j ' / λ. The probability i is equal to (8) i V e e + λiv = V ' + λ ' IV ' ' with the inclusive value IV() equal to 13
(9) IV = lne l' Vl ' / λ + λl ' IVl ' / λ. In areas where all types of internet access are available we odel consuers decisions using four alternative nested logit odels, three two-layer odels and one three-layer odel. The decision trees are depicted in Figure 1. While in trees 1, 2 and 4 we use no internet as the outside option, we use narrowband as the outside option in tree 3. While no internet is the ore natural outside option, we also believe that narrowband as an outside option is sensible since ost households which have a coputer (any of which nowadays have a built-in narrowband ode) also have a telephone line available and therefore the choice between no internet and narrowband ight not only be viewed as the choice of taking internet or not but as the choice of buying a coputer or not. In soe (predoinantly rural) areas, cable network and obile broadband cannot be chosen. Here we odel the decision process as follows: Consuers first decide whether to have internet access or not. If they choose internet access, they decide between narrowband and DSL. We specify the deterinistic part V ij, respectively V ij, as a linear function of consuer characteristics, Z, and product characteristics, X including the price of the product, such that (10) Vij = γ Zi + βx j, where and are the paraeters to be estiated. We assue a siple specification so that the s are constant over all choices j. Consuer characteristics are for exaple, age, educational duy variables or household size. roduct characteristics are the price, the download rate and the download volue. 14
Figure 1: Decision trees for the nested choice odel Decision trees for area where all four alternatives are available (area 1) Tree 1 (2-layer) Tree 2 (2-layer) No Internet Narrowband Broadband No Internet Internet Cable DSL Mobile Narrowband Cable DSL Mobile Tree 3 (2-layer) Tree 4 (3-layer) No Internet Internet Narrowband Broadband Cable DSL Mobile Narrowband Broadband Cable DSL Mobile Decision tree for area where only DSL and narrowband are available (area 2) Tree 5 No Internet Internet Narrowband DSL 5. Estiation results This section shortly discusses the estiation ethod, the odel selection (section 5.1) and presents the estiation results (section 5.2 and 5.3). Section 5.2. describes the results for 15
the region where all four types of internet access are available section 5.3 then discusses the results for the region where only DSL and narrowband are available. 5.1. Estiation ethod and odel selection We estiate the above described odels with sequential axiu likelihood in which the estiation is decoposed into two or three stages depending on the odel. We weight the observations as discussed in section 2. Model selection is based on Hausan specification tests. In addition to the nested logit odels depicted in Figure 1, we estiate a conditional logit odel with all alternatives (no internet, narrowband, cable, DSL, obile) in one nest and test for independence of irrelevant alternatives (IIA). For both areas, we reject the conditional logit odel. We also test for the IIA property in the first layer of tree 1 and in the second layer of tree 2. We reject both odels, i.e. no internet, narrowband and broadband do not belong to one nest as well as narrowband, cable, DSL, and obile do not belong to one nest. This leaves us with two odels (tree 3 and tree 4) for area 1 and one odel (tree 5) for area 2. 5.2. Area 1: DSL, cable, obile, narrowband The estiation results of the two-layer (tree 3) odel and the three-layer odel (tree 4) are given in the tables 5a, 5b and 5c (see the Appendix). Table 5a gives the results for the botto stage, tables 5b and table 5c those for the second and the first stage, respectively. At the botto level, the independent variables are price, the download rate, the download volue, a duy variable for flat rate tariffs and two duy variables indicating the fixed ter of DSL and obile alternatives. Furtherore, we use age, household size, educational duies (with the highest education university degree as the oitted duy variable), a gender duy and a duy for the capital city Vienna. Individual specific variables are interacted with duies for DSL and obile respectively. The price has the expected negative sign and is highly significant. The download rate and the download volue both have the expected positive signs and they are significantly different fro zero at conventional 16
significance levels. The flat rate duy is negative and significant which ight see odd but can be explained as follows: In order to be able to use the observation in the estiation, we have to assign soe value to the variable volue even if the product has a flat rate. We have chosen a value of 55,000 MB (55 GB), which is just a little larger than the largest volue of non-flat rate products. The negative sign on the flat rate duy variable can now be interpreted such that the actual utility a consuer derives fro a flat rate easured in volue is less than 55,000 MB. Most individual specific variables are insignificant and therefore do not appear to influence the choice between different broadband technologies. Exceptions are the duy variables for Vienna, which indicate that it is less likely to have DSL or obile copared to cable in Vienna. This can be explained by the strong position of the cable network operator UC in Vienna. Many Viennese households also have cable TV and obviously prefer to buy broadband fro the sae operator. The goodness of fit of the odel is evaluated using the McFadden R 2 or the likelihood-ratio index, which copares the likelihood for the intercept only odel to the likelihood for the odel with the predictors. The value of the McFadden R 2 of the first stage is 0.34. 28 At the second stage we use a duy variable for narrowband and the sae individual specific variables as in the botto level (interacted with a duy variable for narrowband). However, only the fixed-effects duy for narrowband is significant at a better than the 10% level. The McFadden R 2 shows that the odel perfors oderately well with a value of 0.25. At the third layer, we use a duy variable for no internet and again the sae individual specific variables (interacted with a duy variable for no internet). All variables (with the exception of the Vienna-duy) have the expected sign and are significant at the 1% level. The McFadden R 2 is relatively high (0.41). 28 The McFadded R 2 is useful for coparing odels. The odel fit can also be based on easures of inforation such as Akaike's inforation criterion (AIC) and the Schwarz inforation criterion (BIC). The values of AIC and BIC are 0.383 and 0.397, respectively. 17
Tables 5b and 5c also report the estiated coefficients of the inclusive value. The estiated coefficient of the inclusive value of the third stage of tree 4 is 13.46 and significantly different fro zero at the 1% level. The estiated coefficient of the inclusive value of the second stage is 1.57 and significantly different fro zero at the 1% level. Coing now to the own price elasticities of access deand, the two-layer odel iplies elasticities for broadband services in the range of -2.765-2.570 (see Table 7 in the Appendix). The elasticity of DSL services is -2.765 indicating that a one percent increase in the price decreases the deand for DSL services aost by 3 percent. The corresponding figures for obile and cable services are -2.570 and -2.751, respectively. The lowest elasticity is estiated for narrowband services, which is equal to -1.926. The elasticities derived fro the three-layer odel are siilar although soewhat lower in absolute value. The results indicate that deand for all services is elastic. However, broadband services appear to be ore elastic than narrowband services. An interpretation of this ay be that different broadband services (in particular DSL and cable) constrain each other, while those consuers still using narrowband do not consider broadband as an equally good substitute. 5.3. Area 2: DSL and narrowband As it was entioned earlier cable networks and obile broadband are not available in all regions. Mobile broadband via HSDA is only available in cities with ore than 5000 inhabitants where in ost cases also cable networks exist. In this section we consider only those observations where these two alternatives are not in the choice set of consuers. For this saple we estiate a two-layer nested logit odel where the consuers decide first whether they would like to have internet access or not. At the botto level, the consuers are then faced with the choice between DSL and narrowband access. The one-layer odel is rejected by the Hausan test. 18
At the botto level (Table 6a in the Appendix), the independent variables are price, the download rate, the download volue, a duy variable for flat-rate tarifs and a duy variables indicating the choices for narrowband. The individual specific variables age, household size, educational duies and a gender duy have been interacted with the narrowband duy. The estiated coefficient on the price is negatively significant tough its agnitude is substantially lower than in the two- and three-layer specifications reported for the full saple. The coefficients on download rate and download volue are also significantly different fro zero at conventional levels and have the expected sign. The McFadden R 2 is saller than for area 1 (0.13). On the second stage we again use a duy variable for no internet and the sae individual specific variables as before (interacted with the no internet duy variable). Most variables are significant and have the expected sign. The fit of the odel is relatively good (McFadden R 2 of 0.37). The estiated coefficient of the inclusive value is 1.91 and significantly different fro zero. Turning now to the elasticities iplied by this saple (see Table 7 in the Appendix), we note that the own price elasticity of deand for DSL is equal to -0.97 and for narrowband to -0.77. Both of these elasticities are saller than the estiates for the full saple. It appears that cable and obile not only constrain DSL but to soe extent also narrowband. 6. Iplications for arket definition The instruent used for arket definition is the hypothetical onopolist test (HM-test also called SSNI-test). 29 This test asks whether, starting fro the copetitive level, a nontransitory 5-10% price increase would be profitable for a hypothetical onopolist in the 29 SSNI is the acrony for sall but significant non-transitory increase in prices. 19
arket under consideration. In the case of wholesale broadband access arkets, the question is, whether a hypothetical onopolist for DSL lines on the wholesale level could profitably increase prices by 5-10% above the copetitive level. This can be ipleented by coparing profits before and after the price increase. If profits decrease after a 5-10% price increase, i.e., (11) ( w1 c) x1 > ( w2 c) x2 this eans that the closest substitute to DSL has to be included in the arket (w denotes the wholesale price, c arginal costs and x the nuber of DSL lines sold; subscripts 1 denote prices and quantities before and subscripts 2 prices and quantities after the price increase). Section 5 has derived an estiate of the retail deand elasticity for DSL services. However, in order to define the arket for wholesale broadband access we need the elasticity of deand at the wholesale level. As deand for inputs at the wholesale level is derived fro deand at the retail level, the elasticity of deand at the wholesale level will be related to the elasticity of deand at the retail level. 30 Under the assuptions that (i) one unit of the wholesale input is used to produce one unit of the retail good, (ii) there is no alternative input at the wholesale level and (iii) wholesale and retail supply is copetitive, the relation between retail and wholesale deand elasticity is (12) W = w/p R, where W is the wholesale elasticity R the retail elasticity, w the wholesale price and p the retail price. Assuptions (i) and (ii) are uncritical in the case of wholesale broadband access arkets: One bitstrea-access line is used to provide one DSL line at the retail level, and 30 For a discussion on the relation between wholesale and retail deand elasticities see, for exaple, Inderst and Valletti (2007) and Schwarz (2007). 20
other infrastructures do not offer wholesale access or only to a liited extent. Assuption (iii) is consistent with the HM-test ethodology (5-10% price increase fro the copetitive level) and also appears to be justified in the case of Austrian broadband arkets, in particular with the current wholesale regulations (local loop unbundling and bitstrea access) in place. The wholesale elasticity therefore can siply be calculated by ultiplying the retail elasticity with the share of wholesale costs in the retail price. This share can be estiated to be about 75% in the case of bitstrea products in Austria. 31 If we take the retail elasticity fro the twolayer odel (tree 3), -2.765, the wholesale elasticity would be -2.074. 32 Marginal costs can be estiated to be between 20% and 40% of total costs. This estiate is based on detailed (but confidential) cost data available fro operators. Costs can be divided into categories, soe of which are fixed costs, soe vary directly with the nuber of connections, and soe are in between (step fixed costs, which do not vary by a single connection but, e.g., by 50 or 100 connections). Costs which vary by custoer are, for exaple, the ode and the installation. Step-fixed costs include the DSLAM and soe personnel costs, fixed costs are, for exaple, backhaul, internet connectivity and other types of personnel costs. Considering step-fixed costs as fully fixed and fully variable, respectively, results in a range of 20-40% variable costs in total costs. With this inforation the profits before and after the price increase can be copared as in (11). The result of the coparison is depicted in Table 5. Table 5: Coparison of profits before and after a 5-10% price increase 20% arginal costs 40% arginal costs after 5% price increase -4.8% -2.9% after 10% price increase -10.8% -7.5% 31 This estiate is based on confidential data available fro operators. 32 This estiate is based on deand of residential users. The deand of business users for DSL is likely to be ore inelastic, however, business users ake up only about 20% of all DSL lines and this share is decreasing. 21
It can therefore be concluded that a 5-10% price increase fro the copetitive level would not be profitable for a hypothetical onopolist of DSL lines at the wholesale level due to substitution at the retail level. This conclusion can also be upheld with the soewhat lower DSL-elasticity fro the three-layer odel. This eans that the next best substitute (at the retail level) would have to be included into the relevant arket. Within the broadband nest, cable has the highest cross-price elasticity. This also appears plausible taking into account siilarities between tariffs and product characteristics as well as current penetration rates. It therefore appears that cable exerts the highest copetitive constraint on DSL and would have to be included in the arket. Since the penetration rate of obile broadband was still very low in 2006, we do not investigate whether DSL and cable taken together would be constrained by obile broadband. As the penetration rates of obile broadband are increasing rapidly, we think that it is likely to becoe ore relevant in future investigations. 7. Conclusion We use several nested logit discrete choice odels to estiate the price elasticity of deand for internet services in Austria. Our results indicate that deand for broadband internet access services is rather elastic ( >2.5 for DSL, cable and obile) in those areas where several types of broadband access (DSL, cable and obile) are available. This would indicate that different broadband access technologies are close substitutes and constrain each other. DSL and cable probably for a single arket at the retail as well as at the wholesale level. The elasticity of narrowband is lower (-1.93) which ay indicate that those users which are still using narrowband do not perceive broadband as an equally good substitute. In areas where only DSL and narrowband are available, the DSL elasticity is uch lower (-0.97) which suggests that the constraint fro narrowband on DSL is liited. 22
References Bishop, S., Walker, M. (1999). Econoics of E.C. Copetition Law. Concepts, Application and Measureent. Sweet & Maxwell, London. Crandall, R.W., Sidak, J.G., Singer, H.J. (2002). The Epirical Case Against Asyetric Regulation of Broadband Internet Access. Berkeley Law and Technology Journal 17(1), pp. 953 87. Coreg (2004). Market Analysis. Wholesale Broadband access. Docuent No. 04/83, available at www.coreg.ie. ERG (2005). Revised Coon osition on wholesale bitstrea access Adopted on 2nd April 2004 and aended on 25th May 2005. ERG (03) Rev2, available at www.erg.eu.int. Goel, Rajeev K., Edward T. Hsieh, Michael A. Nelson, Rati Ra. (2006). Deand elasticities for internet services, Applied Econoics, 38, 975-980. Goldberg,. (1995). roduct differentiation and oligopoly in international arkets: The case of the U.S. autoobile industry, Econoetrica, 63(4), pp. 891 951. Greene, W.H. (2003). Econoetric Analysis. Fifth Edition. rentice Hall. Ida, T., Kuroda, T. (2006). Discrete Choice Analysis of Deand for Broadband in Japan. Journal of Regulatory Econoics, 29:1, pp. 5-22. Inderst, R., Valletti, T. (2007). Market Analysis in the presence of indirect constraints and captive sales, Journal of Copetition Law and Econoics, published online on May 21, 2007, http://www3.iperial.ac.uk/portal/pls/portallive/docs/1/15263697.df 23
Maddala, G.S., (1983). Liited-Dependent and Qualitative Variables in Econoetrics, Cabridge University ress, Cabridge. MCA (2006). Wholesale Broadband Access Market. Identification and Analysis of Markets, Dterination of Market ower and Setting of Reedies. Consultation Docuent, 25 th July 2006. http://www.ca.org.t/infocentre/openarticle.asp?id=869&pref=6 McFadden, D., (1974). Conditional logit analysis of qualitative choice behaviour. In. Zarebka, ed., Frontiers in Econoetrics, Acadeic ress, New York, pp. 105-142. OFT (2001). The role of arket definition in onopoly and doinance inquires. A report prepared for the Office of Fair Trading by National Econoic Research Associates. Econoic Discussion aper 2. Oftel (2003). Wholesale Broadband access arket. Identification and analysis of arkets, Deterination of arket power and Setting of SM conditions. Available at www.ofco.gov.uk. Rappoport,., Kridel, D., Taylor, L., Duffy-Deno, K., Alleen, J. (2003). Residential Deand for Access to the Internet. Chapter 5 in the International Handbook of Telecounications Econoics, Volue II, ed. G. Madden, Edward Elgar. RTR (2007). RTR Teleko Monitor. 1. Quartal 2007. http://www.rtr.at/web.nsf/deutsch/ortfolio_berichte_nach%20kategorie_berichte_tkmonitor Q12007/$file/Monitor%20Kurz%20Qu%201-07.pdf 24
Schwarz, A. (2007). Wholesale arket definition in telecounications: The issue of wholesale broadband access. Telecounications olicy, 31, pp. 251 264 Statistik Austria (2006). IKT-Einsatz. Ergebnisse der Europäischen Erhebungen über den Einsatz von Inforations- und Kounikationstechnologien in Unternehen und in Haushalten 2006. http://www.statistik.at/neuerscheinungen/download/2006/ikt2006_www.pdf Train, K.E. (2002). Discrete Choice Methods with Siulation. Cabridge University ress, http://elsa.berkeley.edu/books/choice2.htl 25
Appendix A: Tables Table 5a: Botto stage of tree 3 and tree 4 (cable, DSL, obile) rice -0.0889 (0.01)*** Download rate 0.0256 (0.01)** Download volue 0.0042 (0.00)* Duy variable for flat tariffs -3.1182 (0.94)*** Duy variable for DSL -1.3232 (0.59)* Duy variable for obile -1.237 (1.05) Age interacted with duy variable for DSL 0.0157 (0.01) Age interacted with duy variable for obile -0.017 (0.02) Housholdsize interacted with duy variable for DSL 0.0584 (0.08) Household size interacted with duy variable for obile -0.1587 (0.15) Duy variable for copulsary school interacted with duy variable for DSL 0.1682 (0.29) Duy variable for highschool without graduation interacted with duy variable for DSL 0.687 (0.32)* Duy variable for highschool with graduation interacted with duy variable for DSL 0.2523 (0.29) Duy variable for copulsary school interacted with duy variable for obile 0.3322 (0.58) Duy variable for highschool without graduation interacted with duy variable for obile 0.9787 (0.6) Duy variable for highschool with graduation interacted with duy variable for obile 0.2862 (0.56) Duy variable for feale interacted with duy variable for DSL 0.1169 (0.21) Duy variable for feale interacted with duy variable for obile -0.0156 (0.4) Duy variable for Vienna interacted with duy variable for DSL -0.8897 (0.23)*** Duy variable for Vienna interacted with duy variable for obile -1.1604 (0.45)** seudo R-squared 0.338 Nuber of observations 650 26
Table 5b: Second stage of tree 3 and tree 4 (narrowband, broadband) Inclusive value 1.5726 (0.16)*** Duy variable for narrowband 2.8507 (0.63)*** Age inteacted with duy variable for narrowband -0.0029 (0.01) Houshold size inteacted with duy variable for narrowband 0.0444 (0.06) Duy variable for copulsary school interacted with duy variable for narrowband -0.5448 (0.25)* Duy variable for highschool without graduation interacted with duy variable for narrowband -0.499 (0.27) Duy variable for highschool with graduation interacted with duy variable for narrowband -0.1955 (0.24) Duy variable for feale interacted with duy variable for narrowband -0.1437 (0.17) Duy variable for Vienna interacted with duy variable narrowband 0.0152 (0.21) seudo R-squared 0.246 Nuber of observations 789 Table 5c: Third stage of tree 4 (no internet, internet) Inclusive value 13.4602 (0.95) *** Duy variable for no internet -21.9235 (1.38)*** Age inteacted with duy variable for no internet 0.0278 (0.00)*** Houshold size inteacted with duy variable for no internet -0.5610 (0.06)*** Duy variable for copulsary school interacted with duy variable for no internet 3.7386 (0.28)*** Duy variable for highschool without graduation interacted with duy variable for no internet 1.6175 (0.26)*** Duy variable for highschool with graduation interacted with duy variable for no internet 1.2765 (0.26)*** Duy variable for feale interacted with duy variable for no internet 0.5408 (0.14)*** Duy variable for Vienna interacted with duy variable no internet 1.7959 (0.19)*** seudo R-squared 0.406 Nuber of observations 1830 27
Table 6a: Botto stage for area2 (DSL, narrowband) rice -0.0324 (0.01)*** Download rate 0.0769 (0.0113)*** Download volue 0.0171 (0.01)** Duy variable for flat rate tariffs -0.0048 (47.08) Duy variable for narrowband 0.1671 (0.21) Age interacted with duy variable for narrowband 0.0148 (0.00)*** Housholdsize interacted with duy variable for narrowband 0.0865 (0.04)** Duy variable for copulsary school interacted with duy variable for narrowband -0.45 (0.26) Duy variable for highschool without graduation interacted with duy variable for narrowband 0.4176 (0.34) Duy variable for highschool with graduation interacted with duy variable for narrowband 0.7682 (0.28)*** Duy variable for feale interacted with duy variable for narrowband 0.1902 (0.12) seudo R-squared 0.131 Nuber of observations 708 Tabelle 6b: Second stage for area 2 (no internet, internet) inclusive value 1.9134 (0.55)*** Duy variable for no internet -2.5755 (0.62)*** Age interacted with duy variable for no internet 0.0235 (0.01)*** Household size interacted with duy variable for no internet -0.3019 (0.07)*** Duy variable for copulsary school interacted with duy variable for no internet 2.7415 (0.45)*** Duy variable for highschool without graduation interacted with duy variable for no internet 1.4814 (0.49)*** Duy variable for highschool with graduation interacted with duy variable for no internet -0.1481-0.57 Duy variable for feale interacted with duy variable for no internet 0.1577 (0.18) seudo R-squared 0.370 Nuber of observations 532 28
Table 7: Elasticities own price elasticity cross price elasticity Tree 3 (2-layer, without no internet) cable -2.751 0.364 dsl -2.765 0.621 obile -2.570 0.273 narrowband -1.926 0.478 Tree 4 (3-layer) cable -2.617 0.230 dsl -2.545 0.402 obile -2.481 0.183 narrowband -1.679 0.231 Area 2 (DSL, narrowband) dsl -0.969 0.455 narrowband -0.773 0.507 29
Appendix B: Derivation of elasticities Let us define the own price elasticity to be µ(jj). It is equal to j X j µ jj =, j j with j the probability of choice j and X j the price of that choice. We index the first layer alternatives as and, the second layer alternatives as l and l, and the third layer alternatives as j and j. For siplicity, we suppress the consuer-specific index i. The probability j is equal to j l j =. Its derivative with respect to X j is equal to j l j = l j + j + l j j j j The probability j is equal to = e j V j ' / λ j' V j / λ e. The probability = l' l V e e is equal to / λ + λ IV / λ l V ' / λ + λ ' IV ' / λ l with the inclusive value l l IV = ln j' e V j ' / λ. The probability is equal to = e V + λ IV V' + λ' IV e ' ' with the inclusive value 30
IV = ln l' e λ + λ IV l ' / l ' l ' V / λ. The derivative of j with respect to X j is equal to j j V = j j 1 λ j (1 j ) The derivative of l with respect to X j is equal to l j V = j j 1 λ l (1 l ) j The derivative of with respect to X j is equal to j V = j j ( 1 ) l j The own price elasticity is then equal to µ jj V = j j 1 λ X j [(1 ) l λ j + (1 l λ ) λ j + (1 j )] Let us define the cross price elasticity to be µ(j j). It is equal to µ j' j = j' j X j j' with j ' the probability of choice j and X j the price of an alternative choice. The probability j ' is equal to j' = l j' Its derivative with respect to X j is equal to j ' l j ' = l j ' + j ' + l j j j j The derivative of j ' with respect to X j is equal to 31
32 j j j j j j X V X ' ' = The cross price elasticity is then equal to ] ) (1 ) [(1 1 ' j j l j l j j j j j X X V + + = λ λ λ λ µ