Japan has long been regarded as an exemplary mobile market, years - rather than months - ahead of its closest rivals. Operators the world over turn their eyes to Japan when making promises about where they are headed and what they will become. Is this deference justified and do all those high-end users actually make Japanese carriers any money?

A vicious price war erupted in Tokyo this July. Yoshinoya, the nation's leading operator of fast food-style gyudon (rice topped with seasoned beef) restaurants slashed the price of its regular-size serving by 30 per cent. The move - immediately matched by leading rivals - was the latest in a series between gyudon operators struggling to attract hungry consumers looking for lower prices amid the never-ending economic slump.

But another kind of domestic operator - mobile, not restaurant - is currently staring similarly tough times in the face. Despite the hype surrounding Japan's wireless internet revolution, carriers here are faced with weak ARPU growth, fewer new subscribers, and technical difficulties with handsets and networks. It's no surprise that tired telco execs are taking two-hour lunches; at least they can enjoy some peace and quiet and a cheap bowl of gyudon.

Despite the tough times, however, Japan's mobile market remains staggeringly successful by many measures, including subscriber numbers. Between them, the major cellular carriers had 64,180,200 subscribers as of July 31st.

When you add in PHS subscribers, the total climbs to almost 70 million - that's more than the population of France. The top four carriers, NTT DoCoMo, KDDI, J-Phone, and Tu-Ka, weigh in with 58.9, 18.1, 16.7, and 6.3 per cent of the cellular market respectively.

In addition to size, there's a lot of innovation hitting the market this year. The dominant carrier, NTT DoCoMo, is leading the world in deployment of WCDMA. In May, DoCoMo launched its limited-area trial of FOMA in Tokyo and Kawasaki employing some 4,500 public volunteer 'monitors' (selected from over 147,000 applicants) in what has to be the largest free cellular beta test ever. Monitors received a new 3G phone for free (but have to pay for time and packet usage) in exchange for agreeing to complete surveys on their experience. The trial was slated to end on October 1st when the network opens up to general subscribers.

In addition to the launch of 3G, Japan continues to be in the forefront of the wireless internet, with mobile web access, Java, BREW (Qualcomm's Binary Runtime Environment for Mobile - due for deployment in December 2001), and location-based services all making a mark. As of August 27th, i-mode had 26,779,000 fee-paying subscribers; the copycat EZweb and J-Sky services from KDDI and J-Phone weren't doing too poorly either, with some 8.2 and 7.9 million users each. The three wireless information services are packet-switched, low-bandwidth, and always on, and all deliver text, graphics, and mobile services to a microbrowser resident on the handset; the browser also serves as the handset interface for email, short mail, and telephony control functions.

But whether the start of new services like 3G, the continuing hyper-popularity of i-mode, EZweb, and J-Sky, and Japanese consumers' love of mail and character downloads will prove sufficient to boost operators' ARPUs and minutes of use, and keep new subscribers trooping into the shops, is still not clear. In particular, DoCoMo would dearly like to know what subscribers will pay for 3G services. Takeshi Takahara, a pseudonym for an independent mobile consultant who requested anonymity, summed up the Japanese market for the second half of 2001 as, "decreasing new subscribers, handset development is slowing down, and it's a big wait and see for 3G."

While the absolute subscriber numbers for all of Japan's cellular carriers are looking good, there's trouble on the horizon, particularly for NTT DoCoMo. The carrier estimates that total (voice plus data) monthly ARPU for the November quarterly report should come in at Y8,580 per month, while analysts at the Tokyo office of investment bank ABM Amro think the number will be somewhat higher, at Y8,668. Both these figures are roughly the same as KDDI's Y8,570, and somewhat higher than J-Phone's Y7,800.

Minutes of use on all the networks is very stable, and appears to be strong. DoCoMo has said it expects to report 195 minutes per month per user in November. But what the ARPU figures don't reveal is that unless industry-leader DoCoMo can leverage Java and 3G to boost data ARPU, the 900lb gorilla of Japan's domestic market may be in for a rough ride. DoCoMo's monthly voice ARPU has been falling - from Y15,930 in 1996 to Y8,620 in 2000 - and it appears that this trend will continue. Senior DoCoMo managers have already publicly stated that they expect voice ARPU to fall another 7.9 per cent by the end of fiscal 2001 (year ending March 2002), and no-one appears to expect this trend to reverse.

Industry watchers in Japan attribute the cause to the marginal subscriber syndrome - where DoCoMo is faced with ongoing costs to acquire additional subscribers and run the network, but the newcomers don't use the network as much as early adopters who have already signed up. "All carriers eventually face diminishing returns due to the marginal subscriber effect," says Kieran Calder, senior telecommunications analyst at Indosuez W.I. Carr Securities in Tokyo. "DoCoMo is facing a period where growth will slow and returns will fall. Voice ARPU continues to fall by ten per cent per year," wrote Calder in a May 2001 research report. In short, Calder is arguing that the marginal subscriber effect, long acknowledged with respect to voice usage on DoCoMo's network, is now affecting data usage as well.

There is some strong evidence to support Calder's conclusions. Last June, when i-mode was a little over a year old, the carrier added 1,175,000 new i-mode subscribers. In June this year, the net adds were only 974,000 - a decrease of some 18 per cent. Even the January launch of Java only gave a two-month boost in net adds. Some, like Calder, see little chance for improvement in the carrier's data ARPU situation. "DoCoMo is unlikely to turn this around," he says, although with July 2001 net adds back up above one million the company may yet confound the sceptics.

Meanwhile, the competing carriers don't seem to have been as badly affected, if only due to their smaller size and the fact that they can gain new subscribers through churn and defections from DoCoMo. KDDI, for example, has gained 1.9 million new cdmaOne subscribers since October 2000 largely based on a student-targeted discount plan (but even this growth is now waning).

Credit Suisse First Boston analyst Mark Berman expects KDDI's total ARPU to decline by four per cent year-on-year for FY2001, and a further five per cent in FY2002. "After that, we believe that ARPU will begin to trend upwards as growth in data and internet revenue finally begins to offset the decline in overall voice revenue," he says. In 2000, EZweb users contributed a tidy Y1,500 per month in data ARPU on top of voice, and generated some 35 million page views. Berman estimates that by 2005, total monthly ARPU should be up around Y10,000, with data and internet fees comprising about 20 per cent of the total.

J-Phone should see Java subscribers ramp up nicely - similar to DoCoMo's experience - and its market share is likely to continue rising. J-Sky is expected to take over second place from EZweb sometime before Christmas on the strength of the carrier's youth branding (the company says that J-Sky subscribers use more combined email and SMS than subscribers on either i-mode or EZweb). The operator also claims to carry more mail messages and page views than individual voice calls. "Our marketing aim is the under-25 demographic," says Goichi Yoshizawa, executive vice president and COO at J-Phone Communications. "These people are heavy, high-value users," he adds.

Overall, CSFB's Berman is bullish on mobile in general, and expects to see the three main carriers extend and consolidate their businesses in the next 12 months. For DoCoMo, he forecasts continued growth in Java subscriptions (boosting ARPU), a slightly lower market share and strong earnings. Like analyst Takahara, Berman is taking a wait-and-see stance on 3G, although he thinks that industry concern with the delayed network roll-out is "likely to wane."

For J-Phone, Berman is calling for growth in market share and for new owner Vodafone to fold J-Phone into parent Japan Telecom. "The negative scenario," he says, "is for J-Phone to list separately, leaving Japan Telecom shareholders holding the bag."

Finally, Berman is more than a little frustrated with KDDI. "PHS and PDC subscribers continue to churn off the network, KDDI's fixed-line market share is getting crushed by NTT, and management is not coming through with restructuring," he said. He's happy with the cdmaOne network and its evolution to cdma2000, but he thinks the PHS and PDC networks (operated by group companies DDI Pocket and Tu-Ka, respectively) should be disposed of.

But the really big news this year has to be DoCoMo's roll-out of 3G, even if just on a trial basis. The move puts DoCoMo significantly ahead of competitors KDDI and J-Phone, which will not start their own 3G services, based on cdma2000 1X EV-DO and WCDMA, respectively, until at least the second half of 2002. There will also be significant differences between the networks.

KDDI's cdma2000 1X EV-DO is a data-only transmission format with a data transfer rate averaging 600Kbit/s downstream, and is based on Qualcomm's CDMA technology. The network is due to roll out via upgrades to cdma2000-1X (offering some 144Kbit/s) in December 2001, and then to 1X EV-DO in the second half of 2002.

J-Phone's WCDMA network will be based on Release 5 of the 3GPP 3G standard, due in December 2001, while DoCoMo's is based on Release 4. The result is that DoCoMo can expect little roaming compatibility with any other carrier, while J-Phone should be able to leverage roaming as a key differentiating factor through its tie-up with Vodafone, although it is still too early for this to be certain. "We want to have global roaming on 3G with Vodafone," says J-Phone Communications' Yoshizawa, "but we're still in early stages."

Despite delaying its 3G launch from spring 2001 to June 2002, J-Phone is "in no hurry to meet that deadline," says one senior engineer working on IMT-2000 technology at YRP (Yokosuka Research Park - home of Japan's 3G development efforts). J-Phone, for example, has yet to fix any initial targets for 3G subscribers. It's widely assumed that J-Phone will simply wait and watch what difficulties DoCoMo encounters, then benefit from the hindsight (as well as from the test results of Vodafone's own UMTS network). "Ericsson is the biggest supplier of 3G infrastructure for both DoCoMo and J-Phone," says CSFB's Berman, "so that's where J-Phone will benefit."

And there are certainly plenty of difficulties. The May-to-September trial phase has been marked by handset glitches, dropped calls, and extremely poor battery life (lasting less than one day, using i-mode and Java only with no voice calls, according to one triallist), as well as limited service areas and complaints about lack of content. The first three are engineering problems, which the operator will doubtless fix; the latter two are a tougher nut to crack.

At the September 3rd press conference confirming the October full launch, the company said that only the "i-motion" movie-clip distribution service will be ready by the end of this year; the M-stage Visual and M-stage Music video and audio services won't come online until 2002. Dr. Keiji Tachikawa, DoCoMo president, said that these services were delayed due to "technical problems, not due to lack of content." He was likely correct: the current M-stage Video service (distributed via 64Kbit/s PHS to the Eggy pocket video device) appears to have lots of content. It's just tough to get that content to flow over the packet-switched high-speed portion of the 3G network. When 3G users do get on to the packet network, rates for one packet will be Y0.1 on the Y2,000 monthly package option, Y0.05 on the Y4,000 package, and Y0.02 on the Y8,000 package.

Similarly, the network will operate this year only within an area 30km from central Tokyo, but will expand to major cities throughout 2002. It must be tough to build a network when you don't know how you're going to pay for it, and there's at least some industry sympathy for DoCoMo's sheer persistence. "A lot of people don't appreciate how difficult it is to build 3G," says the senior engineer at YRP. He points to the difficulties of working in an unknown spectral range, the entirely new air interface, and the significant differences in algorithms between WCDMA and CDMA.

"There are so many new parameters. The only way to test it is to build it. I think DoCoMo is very courageous," he adds. Unfortunately, it appears that no-one - least of all DoCoMo - really has an idea as to what 3G means. The problem centres on DoCoMo's traditional mindset as a consumer-focused telco that provides voice services via a handset that looks an awful lot like a mobile telephone, and in this aspect, DoCoMo is no different from competitors J-Phone and KDDI. "The big operators have the networks, but their mindset is consumer. That will last for a long time," says Punnamas Vichitkulwongsa, founder of Tokyo-based mobile enterprise application developer Arriya Solutions.

It is exceedingly difficult for DoCoMo to shift mindset and corporate culture to focus on data, enterprise and corporate applications, and non-telephone-like terminals (for data, the terminal doesn't have to look like a phone). Yet data is considered by many industry insiders to be the true value of 3G. The 3G network's pricing plan - which would confuse even the most ardent of early-adopters amongst consumers and is utterly unmanageable for any corporate comptroller - is a good example of this consumer-centric, non-data orientation.

Others are far more critical of DoCoMo, and assert that for 3G success, pricing and marketing for voice and data have to be completely separated. "For 3G and [related] data services to be an economical, viable business, you have to price separately," says Seiji Sanda, president and CEO of Japan's first MVNO, Japan Communications. Sanda argues that as long as the large carriers such as DoCoMo are restricted by the terms of their existing, voice-centric licences and tariffs, they won't be able to successfully offer the full range of data services that the network can provide.

Further, the competition isn't letting DoCoMo steal all the high-speed bragging rights. While DoCoMo's system is nominally third generation, the network isn't guaranteed to provide the full 384Kbit/s packet speed. In contrast, KDDI's cdma2000-1X network at 144Kbit/s and the recently boosted 128Kbit/s PHS network (operated by DDI Pocket, but leased to Japan Communications) will both provide almost-as-good speeds but much cheaper.

For this reason, Japan Communications' Sanda is bullish on MVNOs. He hopes to see other MVNOs in the next 12 to 18 months that, like Japan Communications, offer 3G-like data services via PHS combined with full enterprise application solutions targeting vertical industries. "We price the telecom and data costs at zero, since we want the customer to focus on the solution, rather than the pipeline," says Sanda.

In the meantime, consumer data use on i-mode, EZweb, and J-Sky continues to enjoy wide popularity. But there's some question as to what percentage of subscribers actually use the data portals, as none of the operators will release any hard numbers. J-Phone, for example, claims that its monthly ARPU is split between content subscription fees (46 per cent), voice calling (41 per cent), mail (four per cent), and web access (nine per cent). The operator says that top usages are melody download, character download, portal menu access, train timetables, games, fortunes, train routes, sports news, and dictionaries - all used by at least 84 per cent of subscribers.

What's clear is that each of the portal operators is seeing a lot of usage - and that translates into data and site access revenues that would otherwise not be available. J-Sky's J-Navi (map look-up system) revenues, in particular, may be quite impressive (neither the operator nor the site builder will confirm figures). One knowledgeable estimate says that the carrier is earning Y4 in airtime fees for each map delivered, for some Y36 million per month. Subscribers also pay Y20 for each map (content access fee), so J-Phone could be pulling in some Y200 million per month.

The widely publicised and highly embarrassing handset glitches that have affected both 2.5G and 3G networks here point to the increasing complexity of the onboard environment and to the weaknesses of Japan's traditional, operator-controlled handset industry structure. On existing 2.5G networks, the problems seem to centre on the onboard system integration - getting the Java virtual machine to work with the browser, the real-time OS, and any required middleware.

3G has these problems plus the added complexity of the WCDMA chipset functions, which require, for example, the system to provide variable power output. The spring round of Sony handset recalls (there were also recalls from Panasonic, Hitachi, Ericsson, Japan Radio and others) led the giant manufacturer to take a $95 million earnings charge in a July financial report.

In an April interview, mobile browser maker Access CTO Tomihisa Kamada explained that three years ago, "we only saw small applications on the keitai." Now, cellphone software includes Java, sophisticated internet and SMS mail applications, and plug-ins, like navigation, multimedia and web browsing. "This is all new to the makers." He says that the handset manufacturers are all trying to hire more software engineers, an observation confirmed by the senior engineer at YRP. "Look at Panasonic here in YRP. They've built up 400 staff," he says.

Nonetheless, the likes of Panasonic, NEC, Sony, and Fujitsu - proud engineering companies - are likely to solve the problems, if only by throwing money and people at them. The makers will also look to partners for help. Sony has tied up with Ericsson for global handset development (the new joint venture will launch on October 1st), while NEC is tying up with Matsushita.

What will have to change, however, is the traditional, top-down control exerted by NTT DoCoMo over the manufacturers. When the handsets were simple, and the only software was deeply embedded, this command-style of development worked. There was little need for co-ordination between third parties, and DoCoMo called all of the research, design, and engineering shots.

But now, there are others involved - like Access (which makes the onboard browser used in some 80 per cent of i-mode handsets) and middleware, application, security, and multimedia vendors. The software vendors and hardware manufacturers have to work together; otherwise there will be more $95 million bills to pay. In a tacit acknowledgement of this reality, Access received a Y932 million capital infusion in July from none other than Matsushita, NEC, Fujitsu, and Mitsubishi - all major cellphone manufacturers. That the investment came quietly from the handset makers is no surprise - they stand to get stung the most if there are any additional glitches in handset software.

Overall, the only thing certain about the Japan mobile scene is there's going to be a heck of a lot more change, and 3G will not become significant until much later next year. Until then, data over 2.5G will continue to generate significant returns, and incremental technologies - like location-based services, Java, and BREW - will prove to be much more significant to the average consumer than anything new, flashy, and hyped.

JAPAN'S TOP CARRIERS AND THEIR WIRELESS WEBS

Carrier: NTT DoCoMo Co. Ltd. (www.nttdocomo.com)</h3>

Total subscribers: 37,784,000

Wireless web: i-mode

i-mode subscribers: 26,779,000

Fee: Y300 monthly basic; Y0.3 per packet

Official sites: 1,760-plus

Unofficial sites: 45,810-plus

Data speed: 9.6Kbit/s

Browser: Access's Compact NetFront

Image format: GIF, animated GIF (5KB; 94x72)

Colours: 65,536 (latest models)

Markup language: i-mode-compatible cHTML

Maximum page size: 5,000 bytes (text and images)

Attachments: images, ringtones

Carrier: KDDI/Au and Tu-Ka (www.kddi.com/english)</h3>

Total subscribers: 15,655,100

Wireless web: EZweb

EZweb subscribers: 8,184,900

Fee: Y200 monthly basic; Y0.27 per packet

Official sites: 2,000-plus

Unofficial sites: Thousands; subscribers can also access unofficial i-mode sites

Data speed: 14.4 and 64Kbit/s (KDDI); 9.6Kbit/s (Tu-Ka)

Browser: Openwave UP browser

Image format: BMP, PNG, wBMP (1.2KB; 96x38)

Colours: 65,536 (latest models)

Markup language: HDML

Maximum page size: 1,492 bytes

Attachments: images, ringtones

Note: KDDI/Au and Tu-Ka are separate voice carriers that offer the same wireless information data service (EZweb).

Carrier: J-Phone (www.j-phone.com/h-e/)</h3>

Total subscribers: 10,741,100

Wireless web: J-Sky

J-Sky subscribers: 7,896,800

Fee: Y2 per page view (no monthly basic fee for data)

Official sites: 6,000-plus

Unofficial sites: Thousands

Data speed: 9.6Kbit/s

Browser: Proprietary

Image format: BMP (32KB, 640x480); PNG (6KB, 96x68)

Colours: 65,536 (latest models)

Markup language: MML (Mobile Markup Language; based on HTML 4.0, 3.2, 2.0)

Maximum page size: 6,000 bytes

Attachments: images, ringtones

Note: All subscriber figures current as of July 31st 2001

Sources:

Telecommunications Carrier Association (www.tca.or.jp)

Operator websites (www.nttdocomo.com, www.kddi.com, www.j-phone.com)

J@pan Inc magazine (www.japaninc.com)

Mobile Media Japan (www.mobilemediajapan.com

NTT DOCOMO 3G PRICING: EFFECTIVE OCTOBER 1ST 2001

3G pricing is a confusing cross-matrix of packet, circuit-switched data call, and circuit-switched voice call fees, all rolled-up in a package carefully designed to confound any hope of early corporate adoption. The company's basic principles for setting fees included:

1 Cheaper packet communication charges suitable for high-speed and large volume transmission

2 Charges for voice communications will be similar to those for current cellular phone services

3 Communication charges are set to align better with the charges adopted in Europe and the US. Specifically, the billing unit will be set to 30 second units (currently set to Y10 units) and there will be only two time zones (business hours and other).

BASIC FEE</h3>

The basic monthly fees comprise a data-only plan and five different comprehensive usage plans.

USAGE FEE</h3>

64Kbit/s circuit-switched data and voice calls will be billed on time usage, reduced under a number of loyalty, frequent-usage, and other discount plans. Fees will range from Y10.5 to Y39.5 per 30 seconds (under Plan 39), with a basic monthly fee of Y3,900. Packet fees will be billed under three different package plans and an SMS plan, and range from Y0.05 to Y0.2 per packet, before discounts.

Source: NTT DoCoMo http://www.nttdocomo.com/new/contents/01/whatnew0903.htm