Feature Article: Econometrics Predictions for the Application Software Market

by Mike Evans

Since the 1970's, users' expenditure on application software has risen inexorably. While Cambashi predict that the current recession will bite a "V" shape into this curve, we also predict that, in the medium term, the upwards trend will continue. For software providers, it is not the end of the feast.

Cambashi's predictions are based on econometric market models. These are quite different to the conceptual models found in books like Crossing the Chasm, by Geoffrey A. Moore. Our models relate, mathematically, growth in application software demand to economic data such as growth in GDP.

These models exploit our understanding of the forces that cause software investment. Just like any other investment, the basic reason to invest in application software is to obtain a return. Application software is a form of automation, replacing labour following a process, just as a machine might replace an assembly line worker. If the rules are right, and the programme implements those rules, then the process will be executed more efficiently and reliably. The ability of computer based processes to remember exceptions and take into account many more input parameters than any human process also increases the effectiveness of the process. And so we increase the return on the original investment.

In the late 1980's, inspired by work of others, Cambashi started to develop a series of econometric models to help explain and predict the evolution of application software markets. We developed sets of equations that modelled the relationship between investment growths, value added growths and application software growths in a number of countries. At first the results were disappointing but when we restated values in local currencies and introduced time lags, we found relationships. We found that investment growth is a more important input parameter than GDP growth, confirming that software is just another type of investment.

The actual outcome is not only a factor of demand, there must also be the supply capacity in terms of sales and deployment skills. Modelling the market from the product perspective, our forecasts take into account seasonality and factors like recurring revenues versus more volatile new name business. We publish a compromise between our two forecasts, usually at the European level. We test our models against historical data to see how well they fit against estimates of market size. Apart from the dot boom and dot crash era at the millennium, they have shown a good fit.

In the present credit crunch, we are seeing negative GDP growth and even steeper declines in investment growth. However, the forecasts we buy from economic analysts running whole economy models show that investment growth will return quite quickly. When we run these numbers through our models, we conclude that while GDP may decline for a considerable period, application software will recover relatively quickly.

We don't guarantee this outcome, after all governments are revising today reports they published a year ago about what happened two years ago! However, in the past economies returned to balance and we think that will happen again. Even if we have a discontinuity, it will probably be a matter of timing. There is a chance of a singularity, otherwise known as the end of capitalism, in which case all bets are off. As one of the Christmas cards sent to us said, "I do hope Keynes was right."

The effect of exchange rate fluctuations is something we have often commented on. A reasonable 2008 over 2007 revenue growth in UK pounds, at late 2008 exchange rates, turned out as a decline in dollars or Euros. When rates are volatile, we reckon the only solution for our clients is to make sales and marketing plans in local currencies and avoid comparisons between one country's growth and another.

For a number of years, Eurostat has published value added growth and investment growth for some 50 or so industry sectors across Europe. While this was not comprehensive, and there were some issues due to late and missing data from some nations, it allowed the econometric modellers who supply us to start to provide forecasts of future growth by industry sector. We have been able to use this data in custom projects to provide detailed industry forecasts to clients.

Recently, a new supplier entered the market. They supply comparable value added and investment forecasts for about 100 industry sectors across about 80 countries. We have bought this data and are working through the steps needed to extend our models to provide even more industry specific forecasts for more countries. We expect that as the economy recovers it will become crucial to allocate resources with better market planning by segment - and Cambashi intend to be ready to help.