Standards Wars in Wireless Communications David J. Salant March, 2005
Introduction • Mobile wireless communications industry has been locked in a series of standards battles for approximately 20 years. • Currently, there are over 1.3 billion people using cellular and PCS service in the world. • Gross new adds are approximately 500 M per year, and annual IP royalties likely exceed $5 billion. • Licensing fees accrue to developers of standards in handsets and in terminal equipment. • Network effects suggesting standardization include communication between individuals, geographical area of coverage and cost of complementary products (handsets). • Telecommunications policy makers adopt different approaches • Market approaches – market can find the standard. • Government mandate. • Trade associations, not always voluntary. • Competitive dynamics suggest tipping eventually result in a standard. • Yet, at least in wireless, the market has yet to settle on any one standard for very long.
Some Related Literature • Arthur’s (1989) – tipping effect of network externalities. • Economides, Matutes and Regibeau on compatibility. • Church and Gandal, on variety of complementary components with network effects. • Farrell & Saloner, Katz and Shapiro, Budd, Harris & Vickers and Cabral & Riordan on competitive dynamics and network effects. • David & Greenstein survey types of standard setting processes. • Farrell on converters.
The Wireless Industry • The EC has mandated a harmonized standard, GSM, for “2nd generation” mobile services • North America, and other countries, have allowed the market to determine standards • In North America, and especially the US, there is coverage with multiple 2g standards • Economic theory, and experience in other industries, suggest that mandated standards has some advantages and market standards has other advantages
Competing 2nd Generation Standards First wave of digital, second generation, wireless standards includes • GSM – Global Standard for Mobile Communication, based on a time division multiple access protocol • CDMA – Code Division Multiple Access or IS95 • TDMA/DAMPS – Time Division Multiple Access or Digital AMPS, another standard based on a time division multiple access protocol
1G to 2G Migration in the US and EC • The US mandated the analog AMPS standard for 1G • In the EC, countries did not use a uniform approach, using both AMPS and NMTS and in different bands • EC chose GSM for 2G • US chose market based approach
2G in US • In US the FCC • Initially allocated two, 25 MHz, or 2x12.5 MHz, licenses for 1G AMPS networks • Mandated the AMPS, analog, standard • Subsequently, FCC allocated 120 MHz or 2x60 MHz for PCS • First 60 MHz in 51 MTAs auctioned in 4Q ’94, service began sometime later • Subsequently, additional 60 MHz auctioned in a series of auctions beginning ’95 in 493 BTAs • Most recent auction was concluded in January 2001 • No PCS standard mandated, only out of band interference, service (had to be PCS) and modest build out requirements • FCC allowed refarming of AMPS frequencies
2G in the EC • The EU has delegated choice of wireless standards first to CEPT and then to ETSI • GSM was selected in the 900 MHz band in 1987 as the harmonized standard • GSM service was introduced in EC as early as 1989 (in Germany) • DCS1800, an upbanded version of GSM, was introduced in 1993 • EC had over 1M GSM subs by 1993 • By 1995 all EC countries had at least one GSM license in operation
2G Elsewhere • A number of other countries have followed EC (e.g., Greece, Turkey, South Africa, most of Eastern Europe) • Some countries allowed multiple standards • Australia started with only GSM and has now allocated spectrum which is being used for CDMA • Japan has PHS standard unique to Japan • China and India have permitted allocated spectrum for both GSM and CDMA • Israel has TDMA, GSM, and CDMA
The Competition • Competition occurs at several levels • Network infrastructure – Lucent, Motorola, Nortel, Ericsson, Siemens, Alcatel. • Terminals/handsets – Motorola, Sony/Ericsson, Nokia, Samsung, Kyocera. • Operator • Chipsets – Qualcomm, Nokia, TI, NTT • IP – Qualcomm, patent pool for GSM • Network equipment and handsets are both spectrum and technology specific.
2G to 3G in EU • Initial ETSI Special Mobile Group (SMG) planned a variant of 2G CDMA for 3G. • New 3G CDMA standard would be backwards compatible with 2G CDMA. • Backwards compatibility would benefit US chipset suppliers and also CDMA handset suppliers, which were largely non-European. • European handset suppliers preferred non-compatible standard.
EU Spectrum Management Policy • EU, and especially UK and Germany, were anxious to allocate 3G spectrum. • Tipping effect of first to deploy a standard was thought to have an effect on subsequent global adoptions. • EU designates CEPT/ETSI to set telecom standards. • ETSI voting power weighted by European turnover • SMG decided to make a few small adjustments basic 3G CDMA standard • SMG included Nokia, Siemens and Ericsson. • New 3G standard, UMTS aka WCDMA was incompatible with basic 3G CDMA standard, renamed cdma2000 • US (QUALCOMM) complained • ETSI standard setting policy explicitly favored EU suppliers over non-EU ones. • ETSI standard setting process violated US/EU competition policy treaty. • EU agreed to some modifications.
Band Plan Conflicts • Equipment is specific to bands and technology. • For 1G, • AMPS was deployed 824 – 849 x 869 – 894 MHz. • EU used both AMPS and NMTS bands at 450MHz • For 2G • US – 1G + 1850 – 1910 x 1930 – 1990 MHz • EU – 880 – 915 x 925 – 970 MHz + 1710 – 1785 x 1805 – 1880 MHz. • For 3G • US – TBD + 700 MHz + AWS (1710 – 1755 x 2110 – 2155 MHz) • EU – 1920 – 1980 x 2110 – 2170 MHz • EU has not allowed re-farming, US has • Other countries left to decide. • GSM works in US, CDMA not available for EU bands, and has not been permitted in EU.
PCS 1900 PCS 1900 AWS AWS DCS1800 DCS1800 WCDMA WCDMA 2100 1700 1750 1800 1900 2000 2G and 3G Spectrum Over 1700 MHz
Coverage vs. Compatibility • Theory is incomplete – discusses network effects and compatibility • Compatibility is of limited importance in wireless standards • Matters to the extent that compatibility affects production scale, and therefore costs, of network infrastructure and terminals • Reduces risk of orphans, but does not eliminate it, e.g., CT2 • Coverage determines value much more than compatibility
For Harmonization Costs, due uniformity Enhances coverage Reduced risk of orphan technology Reduced consumer uncertainty For Market Based Standards Competition Product/quality differential can enhance value Costs, due to competition No risk of locking into wrong standard Mandated Standards vs. Market Determined Standards • Harmonization can lead to higher or lower penetration • If harmonization improves end user welfare, it should result in higher penetration, other things equal
Dynamic Factors • When the market, and not regulatory mandate, determines standards, early adoption decisions can have a tipping effect • For wireless networks, interconnection means that compatibility is not essential • Two or more standards can be viable, and tipping effect is less persistent • When there are different firms deciding on standards in different regions, the choice in one region can affect optimal choice of standards by all firms in another region
Coordination Across Regions • In each region, there are multiple licenses available • With market based standards, different firms will make decisions in each region • One firm will not usually want to adopt a standard that will not have coverage in other regions • Multiple equilibrium can exist, depending on the sequencing of the decisions • Equilibrium can exist in which all standards are in use in each region • Equilibrium can also exist in which there is complete harmonization
A Simple Duopoly Model • Technology starts at base level, 0, and can progress to new level 1. • Two periods, and each has two stages. • First stage firms determine whether or not they will produce to the same specifications. • Second stage, firms invest in R&D • Probability of success = ½ ρ2 • Let ПS(j) = each firm’s profit with a common technology j = 0, 1. • Let ПD(j,k) = profit for a firm with technology j when its rival has technology k and they disagree. • We assume • ПS(j) > ПD(j,j), j =0,1 and 2 ПS(1) > ПD(0,1) + ПD(1,0) • At state (1,0) or (0,1), bargaining is costless and perfectly efficient, so that industry profits are the same as at (1,1).
Model Features • If one firm gets new technology, it will earn an advantage. • Firms cannot capture all benefits of investment. • Can be under-incentives to invest if firms agree to one standard.
Incentives to Disagree Proposition: If ПS(0) - ПD(0,0) is small , then standardization occurs in all states other than (0,0). • In other words, firms have incentives to deviate from any single proposed standard. • Result does not require differentiation or inefficient bargaining. • Result should extend to dynamic models.
Data • Data for 1994-9 from 25 countries, Australia, Austria, Belgium, Denmark, Finland, France, Germany, Greece, Hong Kong, Iceland, Ireland, Israel, Italy, Japan, Korea, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, United States. • Six countries had multiple standards, Australia, Hong Kong, Japan, Israel, New Zealand, and the USA • Test to see if impact of multiple standards on penetration was positive or negative • In most cases, multiple standards meant possibly higher overall market penetration
Variables • Number of subscribers in each year • Price is cost in US$ for 120 minutes of peak service • Analog dummy is 1(0) if analog service was (un)available • Multiple was 1(0) if multiple (one) digital standards were available • Connect is the connection charge for residential service • Payphone_100 is number of payphones available per 100 people • Also, Yn, n = 1995,6,7,8,9, is a year n dummy, POP is population, GDP_capita is per capita gross domestic product
Variable Mean Std. Dev. Minimum Maximum SUBSCRIBERS 5,787,062 1.19e+07 12,000 8.07e+07 PRICE 68.62808 22.73185 14.82 121.5 ANALOG 0.70 0.46 0 1 MULTIPLE 0.15 0.36 0 1 POPULATION 3.50e+07 5.73e+07 266,783 2.76e+08 GDP_CAPITA 24,254 8,505 6,901 44,962 PAYPHONE_100 0.39 0.23 0.074 1.22 CONNECT 161.58 154.10 0 832.1 Summary Statistics
Model 1 Dependent variable, subscribers/log subscribers
Model 2: Results Dependent Var. = Log of Installed Base R2 = .929, Adj. R2 = .923
Model 3: Results Dependent Var. = penetration R2 = .870, Adj. R2 = .860
Model 4: Results Dependent Var. = penetration R2 = .718, Adj. R2 = .655