be introduced. Chapter 9 evaluates the forecasting performance
of the various models presented earlier and
compares the models within sample, so the problems of
forecasting the independent variables do not arise.
The perplexing "nding of Chapter 9 relates back to
earlier parts of the book. The time varying parameter
model performs best on the example data set. This of
course does not mean that it would always perform best
on other data. Indeed the failure to compare on more
than one data set (arrivals to South Korea from the UK
and USA from 1962 to 1994), and with a wider range of
independent variables, and possibly with disaggregation
into travel types, could be considered a weakness in this
chapter. More interestingly, the &no change' model also
performs well. The perplexing issue is that this book must
conclude that whilst overcoming stationarity does
improve forecasts, the more complex models do not
perform well against a simple no change forecast. This
"nding continues to support the "ndings in current literature
(as stated earlier in the book) that causal models
tend to be less accurate than might be expected. Also
from a practitioner's point of view, beyond sample forecasting,
where the independent variables must also be
separately forecast can only increase error and further
reduce accuracy, and hence further favour the no
change model. The classical answer to this problem by
economists is to suggest that either the correct causes
PII: S 0 2 6 1 - 5 1 7 7 ( 0 1 ) 0 0 0 1 8 - 8
have not been found (readers can then refer to Chapter
8 and the use of Panel data) or the model is not correctly
speci"ed (hence the more complex modelling techniques).
However, it seems that despite the increased complexity
of the models, and their capacity to include more
variables, the causal models still fail to be particularly
accurate.
Overall this book is very important in raising the level
of technical expertise required to achieve accurate
causal model forecasts, and explaining why this is
necessary. As a text, the book pushes the understanding
of tourism forecasting methodology out to currently accepted
econometric standards, and must become a standard
research reference work for economic forecasting of
tourism series. Furthermore, the research "ndings of the
book suggest we still have some way to go in formulating
an accurate causal demand model, so watch this
space.
Finally, the book is perhaps highly priced at US$75,
but is high quality, hard cover and very well presented.