| 12 |
While the methodology here largely follows the U.S. Bureau
of Economic Analysis (BEA), a few key differences between the
two should be borne in mind when making comparisons. First,
the effect of treating software as investment appears to have
had a more substantial effect on GDP in the U.S. than in Canada.
This stems in good part from differences in how software (purchased
separately from hardware but reported as capital spending in
organizations’ accounting records) was treated in the
U.S. and Canada prior to its inclusion in GDP as investment.
It was explicitly reclassified as a current expense by the U.S.
BEA while, at Statistics Canada, it was implicitly classified
as hardware investment. Thus, in order to include software as
investment in GDP, all business and government spending on software
was added to GDP as ‘new investment’ by the U.S.
BEA while, for Canada, only the non-capital spending is added.
Second, trade in software services (as measured in Balance of
Payments (BOP)) isn’t taken into account in the U.S. work,
while it’s taken into account here. The BEA plans however
to expand its coverage of software trade in the future. Third,
the adjustment for custom-design software embedded in hardware
and/or sold is more substantial for Canada than in the U.S.
The adjustment here covers a range of industries and leads to
a 10% reduction of custom software investment. In the U.S. case,
the adjustment is limited to custom software inputs to the computer
manufacturing industry only and has a negligible impact on custom
software investment. Last, while the methodology for own-account
investment is essentially the same, its total cost is estimated
at just over twice the labour cost in the U.S., but only 1.5
times the labour cost here in Canada. Details of the U.S. methodology
can be found in the Survey of Current Business, August 1998
and December 1999 issues, as well as the paper “Software
Prices and Real Output: Recent Developments at the Bureau of
Economic Analysis” (April 2000) by Robert Parker and Bruce
Grimm. For a more recent summary, including an outline of planned
improvements, see Carol Moylan, “Estimation of Software
in the U.S. National Accounts: New Developments,” for
OECD Meeting of National Accounts Experts, Paris, September
2001. |
