Fred
Phillips

The Economist reported that in early 2007, for the first time in history, more humans lived in cities than in the countryside. We are now a different species, in terms of the environmental niche we inhabit.  One thinks of Isaac Asimov’s Trantor, the planet that was completely covered by buildings.  Is Earth headed for a similar future?

2004 was the first year Amazon.com moved more dollar volume in consumer electronics than in
books.  Investors had to revise their image of the company and ask new questions of its management. 

When Toyota overtook GM in car sales in 2007, it raised questions of US competitiveness and Japanese market dominance that had not been asked since the early 1980s.

Crossovers of this kind are dramatic.  They are not surprising, because they happen in the middle of the well-known s-shaped life cycle curve (see the Figure), where the growth or decline of a behavior has become steady and predictable. Any astute analyst can see them coming, well
in advance.  Crossovers are no less thought-provoking, though, for all their individual predictability. 

And when many significant crossovers (in varied social and market domains) occur in fast succession, they say something important about our times, and about market opportunities.

Peruse the Table below.  I guarantee it will help you see recent decades in a new light,  lead you to an idea for a new product, or at least, give you pause for thought.

Table: A haphazard collection of crossovers.  Imagine – and, if you will, exploit – the trends they imply.

video game sales exceeded music sales (PricewaterhouseCoopers estimate).

In 2008, for the first time,

the U.S. overtook France as the world's leading wine-drinking nation.  Growth in U.S.
consumption was driven by Australian wines. 

These crossovers often highlight important trends.  In the cats/dogs case, it was the trend of households toward apartment-dwelling, with all that that means for appliance sales and space-saving storage devices (not to mention pet care products).

Each crossover involves a new behavior overtaking an old behavior. This might be because the incidence of the old behavior is declining (as in music sales through traditional channels), or because both are growing but the new behavior is growing faster than the old one. 1991 provided an example of the latter: Investments in IT and traditional capital goods were both growing, but IT was the star performer by far.

Many of the crossovers shown in the Table illustrate a new behavior substituting for an older one.  (That is, an individual engaging in more of the new behavior necessarily engages in less of the old one, and both behaviors deliver the same benefit.) Dial-up vs. broadband home Internet connections are a classic instance of technology substitution.  Let’s have some fun with that one before returning to the (really more interesting) question of why the others are not classic cases of substitution.

The crossover for home Internet connections happened in 2004.  With just a little extra data, we can fit the Fisher-Pry equation for technology substitution.  From CNET (http://news.com.com/2100-1034_3-5815756.html) we learn that “29 percent of North American households connected to the Net via 

broadband.” Home broadband penetration in Canada is higher than in the US, but let’s go ahead and apply the 29 percent figure to the US. Next, websiteoptimization.com/bw/0607/ tells us that between March 2005 and March 2006, the percentage of broadband households in the US grew from 34.97% to 44.45%.

Knowing that broadband and dial-up had equal market shares in 2004 (the crossover point), the total market for both types of Internet access was twice 29 percent, or 58 percent of US households.  In 2005, 34.97 percent penetration meant broadband had 34.97/.58 = 60 percent share of net access, and in 2006 broadband’s share was 44.45/.58, or 77 percent.

The Fisher-Pry equation is log (s_n-1 / s_n) = kt, where s_n-1 is the market share of the old technology, and s_n is the market share of the new technology.  k is a parameter, and t is time.  Fisher-Pry is a single-parameter model. The bad news is that makes it rather rigid, and inescapably symmetric.  The good news is that our three data points are more than sufficient for fitting this simple model.  Skipping some details, we arrive at the fit shown in the Figure.

According to CNET, Forrester Research predicts 62 percent broadband penetration by 2010, the newcomers attracted mainly by lower prices. How does our Fisher-Pry forecast square with Forrester’s prediction?  Our curve shows a 96.6 percent share for broadband in 2010.  Multiply by the full market size (at least this was the full market size in ’04, though it will have grown by 2010) of 58 percent of US households, and we arrive at a competing estimate of 56 percent penetration for 2010.

This is a bit less than Forrester’s forecast, and the difference seems partly due to the flattening-out of the s-curve as we approach the end of the decade.  Forrester also says declining prices will drive higher broadband penetration.  The Pew Trusts disagree.  Their 2004 survey of reasons for switching to broadband at home showed 36 percent of switchers were moved by “connection too slow or frustrating.”  Only three percent mentioned price as a motivator. (Pew Internet&American Life Project, Feb 2004, http://www.pewtrusts.com/pdf/pew_internet_broadband_0404.pdf)  As you form your own opinion (I said this was for fun), let’s look at some of the other crossovers.

Figure: Fitted trajectories of two modes of Internet access, illustrating their crossover.

No one would suggest that IT is a substitute for capital equipment.  They don’t deliver the same benefit; we still need manufactured goods that conventional computers cannot make for us.  Some US companies invest in more IT domestically as they offshore manufacturing to China.  This example illustrates the complexity that can make the simple curve of the Figure an inappropriate tool for analyzing crossovers.

When Japan’s pop culture exports passed its car exports, it said less about Japan’s industrial structure than about new competition for Hollywood and Bollywood.  It would be a mistake to view this as an instance of substitution.

It is usually said that a technology is “mature” when the rate of new patents in the area begins to level off.  When pornography loses its place as the killer app for a new communications technology – in 2007 social networking sites became the killer app of the World Wide Web – that might be a truer measure of the technology’s maturity. And, I guess, the customers’ maturity too.

In 2001, California’s population of children aged five years and under numbered just over three million.  Forty-eight percent of these children were Hispanic (http://www.ppic.org/content/pubs/jtf/JTF_YoungChildrenJTF.pdf).  The crossover has surely occurred by now, and it has immense implications for the future composition of the workforce, for educational programs, for protection of children (who are themselves US citizens) of illegal immigrants, and for many other socioeconomic issues.  Small wonder that writer-director Wayne Kramer’s 2007 movie Crossing Over is a “multi-character canvas about immigrants of different nationalities struggling to achieve legal status in Los Angeles” (imdb.com).

I first became interested in crossovers when MRCA, for whom I then worked, publicized the new ascendancy of cats over dogs as household pets. The event was intriguing not only for its own sake, but because the pet ownership checkbox on our demographic questionnaires was so tangential to MRCA’s core business of product purchase tracking. An alert (or bored) analyst had decided to tabulate these checkboxes, and accidentally discovered the crossover.  An equally alert manager issued a press release, which the media immediately picked up. Crossovers are newsworthy!

Keep your eyes open for crossovers, and understand the problems and opportunities they represent.  I keep my collection posted at www.generalinformatics.com/crossovers.html.  If you would like to suggest additions, send them to me at info@generalinformatics.com.