This analysis explains why initial terminal forecasts for 1999 averaged 55 million units below what will likely be 260 million terminals shipped. It reveals that between 1996 and 1998 the world terminal replacement rate increased from 19.7 percent to 32 percent and explains why this has occurred. Based on the increased replacement rate, it introduces a component analysis forecast model. It applies the model to forecast terminal sales for 1999 and 2000.
Introduction
Forecasts of mobile terminal sales have failed with dismal regularity. Universally, analysts have underestimated the size of the market. Even forecasts for the current year have proven inadequate. Initially, we ourselves were no more perceptive than other analysts.
As of this writing (August 1999), we now foresee that mobile terminal sales for 1999 will reach from 250 million to as high as 262 million. These values are 22 to 28 percent higher than the average 205 million sales that we, as well as others, were estimating in late 1998 and early 1999.
As examples, in December 1998, Goldman Sachs forecast 205 million sales for 1999. In January 1999, Paine Webber forecast 210 million. As late as May 1999, another industry consultancy was forecasting 205 million.
However, by mid-year, it was clear these earlier estimates were far too low. In June 1999, Texas Instruments, based on discussions with manufacturers, was anticipating 245 million terminal sales. In August, we reviewed expectations of component suppliers. Based on these, we concluded that even TI was low. Assuming that suppliers could meet demand, eventual shipments would reach 250 million and possibly as high as 262 million.
In our following analysis, we examine the market shifts that lie behind the unexpected acceleration in terminal sales. In so doing, we reveal why more rather than fewer sales will be the norm for the future.
Approaches to forecasting terminal sales
Typically, analysts forecast terminal sales by applying “regression analysis.” Regression analysis assumes a constant growth trend. The trend may be linear or exponential. The essential assumption of regression analysis is that the trend remains steady over time. Given this, projection of future sales based on past sales becomes a straight-forward exercise. However, should the underlying trend change, regression analysis fails.
A different approach centers on “component analysis.” This assumes that the factors that determine terminal sales can be identified and measured. Once this is done, the impact of each factor can be estimated. The forecast rests on the estimated cumulative effect of the factors. The challenge of component analysis is to identify the factors and to quantify their effects.
Two key components determine terminal sales. On the one hand are terminals that serve new subscribers. On the other hand are replacement terminals, which serve established subscribers. The former is measured by the annual increase in subscribers. The latter is measured by the cumulative base of subscribers. We approximate both. So doing enables us to apply component analysis to forecasting total terminal sales.
World subscriber growth
Figure 1 reports our estimates of total world mobile subscribers as of years-end 1991 through 1998, annual subscriber growth during 1991 through 1998, and our forecasts for 1999 and 2000. During 1991, subscribers increased by 4.5 million to reach a total of 16.1 million. During 1998, they increased by 102 million to reach a total of 308 million.
We estimate that during 1999, world subscribers will grow by 151 million, reaching a total of 459 million. During 2000, they will grow by 155 million, reaching a total of 614 million. We believe that 2000 will mark the peak in subscriber gains. During 2001, the absolute increase in subscribers will begin to decline.
We are certain that the peak and subsequent decline in subscriber growth will take place. We are unsure of the precise years in which that will occur. This notwithstanding, because replacements have become so important, the decline in absolute subscriber growth will have relatively little impact on total terminal sales.
The terminal replacement rate
Figure 2, page 56, estimates and forecasts the world terminal replacement rate from 1994 through 2000. We have derived these rates from our analyses of total subscribers, annual subscriber growth and total terminal sales.
This figure reveals a remarkable pattern. Between 1994 and 1996, the replacement rate was relatively stable-increasing but slightly from 17.9 percent in 1994 to 19.7 percent in 1996. This meant that for every 100 subscribers at the beginning of 1994, 17.9 replaced their terminals during the course of that year. For every 100 subscribers at the beginning of 1996, 19.7 replaced their terminals during the course of that year.
However, during the next two years, the replacement rate escalated. During 1997, it increased to 26.6 percent. During 1998, it grew to 32 percent. We surmise that it will level off at about 35 percent in 1999 and remain there into the future. With this increase in the replacement rate during this brief period, one can understand why forecasts of terminal sales proved uniformly too low.
Why the sudden increase in the terminal replacement rate? We believe that four factors likely explain it. In many ways, these are tied to the transition from analog to digital technologies. First is the decreasing weight and size of terminals. The U.S. market provides an example. In 1990, the average handheld terminal weighed 422 grams (14.9 ounces). By 1998, the average weight had dropped to 183 grams (6.54 ounces). Less weight and smaller size makes terminals more convenient to carry.
Second is increasing functionality of terminals, such as short messaging service. This makes terminals more useful.
Third is improved circuitry and battery chemistries. These extend battery life. Due to these, battery talk times of several hours and standby times of more than a week are common. Again, this makes terminals more useful.
Fourth is the declining prices of terminals. These declines have been remarkable. Again, using the U.S. market as an example, at year-end 1993, the average low-priced digital terminal retailed for $489. By year-end 1998, this had fallen to $110.
In sum, these four factors make terminal replacement more attractive and more affordable to established mobile subscribers. It is for these reasons that the replacement rate has accelerated since 1996.
Forecast terminal sales
With both empirical experience and an explanation of it, we are able to forecast annual terminal sales. This forecast is based on the three component factors which we discussed earlier: (1) annual subscriber growth, (2) the cumulative base of subscribers and (3) the terminal replacement rate. We do not include the step-by-step calculations.
Figure 3 forecasts world terminal sales for 1999 and 2000. The left column repeats our previous forecasts of annual subscriber increases. Again, we anticipate 151 million new subscribers in 1999 and 155 million in 2000. The middle column reports terminal replacement sales. Assuming a 35 percent terminal replacement rate, this will lead to replacement sales of 108 million in 1999, increasing to 161 million in 2000. The right column sums the previous two. As such, it reports total terminal sales for each year. These will approximate 259 million in 1999 and 316 million in 2000.
These values are far greater than any so far imagined. They reflect the ever-greater functionality and greater value manufacturers are designing into their terminals at ever lower prices.
The competitive implications of unanticipated terminal sales
We can adduce, at least, three competitive implications of these unanticipated large terminal sales.
First, the component shortage which has
nagged the industry since late 1998 will become more severe. To the extent possible, forward-looking terminal manufacturer
s will take every step possible to assure component supplies. Such steps may include financing expansion for suppliers where needed. However, even with such steps, the industry can expect a growing component shortage during fourth-quarter 1999. This will increase in severity into 2000.
Second, the shortage will compel suppliers to allocate components. The largest terminal manufacturers will receive preferential allocations. This will be especially so if they are funding production expansion for the suppliers.
Third, the preferential allocations of components will place smaller terminal manufacturers at disadvantage, severely constraining their production. This will assure the larger manufacturers of even greater market share and competitive advantage than otherwise would have been the case.
The changing definition of terminal
Mobile terminals as presently conceived will not describe the range of terminals future subscribers will be buying. Many such terminals will take the forms of personal digital assistants, which provide a separate functionality from that of mobile communications. The Palm Pilots of today are precursors of such devices.
This serves as a warning for Nokia Corp., Motorola Inc., L.M. Ericsson, and other conventional terminal manufacturers. The computer industry will lead the way in producing such devices.
By 2003, we foresee terminal sales falling into three categories. First, and still the largest, will be second-generation digital (primarily Global System for Mobile communications, Time Division Multiple Access Interim Standard-136, and Code Division Multiple Access IS-95). We also include in this group, Japanese PDC, integrated Dispatch Enhanced Network/E-SMR, and residual analog.
Second will be the beginning introduction of commercial third-generation mobile terminals.
Third, and most important, will be the evolution of RF independent network access devices. These will include PDAs, handheld computers, and laptop computers, among others.
Unlike current terminals, such access devices will not be registered onto networks. Rather, using either hard or wireless connections (such as infrared and Bluetooth), they will access networks through other RF devices. In most cases these other RF devices will be conventional voice terminals. They may, however, be separate stand-alone RF links. This will mark a fundamental shift in terminal market dynamics.
Dr. Herschel Shosteck is president and chief executive officer of Herschel Shosteck Associates Ltd. Located in suburban Washington, D.C., the firm specializes in analyzing and forecasting market and technology trends of the mobile communications industry. The data used in this analysis are reported in “Shosteck E-STATS,” the electronic database developed by the company. The full forecasts described in this article extend to 2003. For further information contact Jane Zweig, executive vice president, (301) 589-2259 or jzweig@shosteck.com