Forecasts add to investment confidence
When it comes to spending money, businesses and governments are turning to forecasts to improve their investment decisions. Econometrician Dr George Athanasopoulos is leading research to improve the accuracy of these forecasts for the Australian tourism sector. He is also developing more complex predictive models to help governments analyse how policy changes will affect the national economy.
George says forecasting is essentially about improving business and investment confidence. In
the case of tourism, the domestic tourism sector is the largest single financial contributor
to the national economy. Tourism Australia provides 10-year tourism activity forecasts to
help the industry plan investment and promotional initiatives.
In 2005, domestic tourism generated an estimated $55.5 billion for the economy. Forecasts at the time indicated continuing growth. But George says when Tourism Australia asked him to refine its modelling he discovered errors in the original model that resulted in an overly optimistic outlook. The forecasts actually masked the beginning of a downturn.
'The new model identified that domestic tourism was on the decline. All forecasts were revised downwards,' he says. Tourism Australia was quick to make use of George's new forecasts and the industry has been able to respond with new domestic marketing campaigns.
George has continued to work with Tourism Australia through its research arm, Tourism Research Australia. He has been developing a model that incorporates hierarchies for local, regional, state, national and international tourism. This will allow for better planning to respond to changes in different market sectors.
'The key innovation of this method is the efficient reconciliation of forecasts from each level,' George says. 'Tourism Research Australia uses this method to generate accurate forecasts of international arrivals at the national, state and regional level - it makes sure the numbers add up. It also uses this for domestic tourism at a much more detailed level.'
On a larger scale, George is also developing highly complex Vector AutoRegressive Model Average (VARMA) models for economic forecasting and policy analysis.
Economists already use several other models to identify how government actions and policies filter through the economy.
'These models work okay, but we know the VARMA models will be more accurate. However, they are much more difficult to use because they are much more complex.'
Factors being modelled include interest rates, currency exchange rates and inflation. Because these have such significant flow-on effects in the economy, even small gains in the accuracy of modelling are important. More accurate modelling, for instance, may provide earlier warning of potential international economic crises.
Improving the usability of VARMA modelling is a major aim of George's research. He is working on an automated VARMA model that will allow economists and others without high-level expertise to use it.
He is also enthusiastic about sharing his love of numbers. To this end he is collaborating with Monash colleague Professor Rob Hyndman to produce an online textbook titled Forecasting: Principles and Practice (http://robjhndman.com/fpp/) .
The book will include practical examples and will be regularly updated as research refines the modelling. George believes that with free, open access via the internet, it will become the most widely used text in the field for both teaching and practical use.