Measuring Investment in Software
Nadim Ahmad · OECD science, technology and industry working papers · 2003
The latest system of national accounts (SNA93) recommended that purchases of software (and any own-account production) should be capitalised as long as the acquisition satisfied conventional asset requirements.This change added about 1% to GDP in most OECD economies in the mid-1990s.However, the range of the revision has been significantly different across countries, leading many observers to question the comparability of these statistics.An OECD task force set up in October 2001 confirmed that differences in estimation procedures contributed significantly to the differences in software capitalisation rates, and a set of recommendations describing a harmonised method for estimating software were formulated.Most of these recommendations were approved at the OECD 2002 National Accounts Expert meeting.This paper reviews the recommendations in its annexes, giving more expansive explanations in some areas, and provides estimates of changes to GDP levels and growth that might be expected if the OECD recommendations were applied.Estimates of changes are also presented using an alternative harmonised method.Whichever harmonised method is applied, the impact on GDP levels is likely to be significant, and in some countries about 1%.1.This is not the only area of the national accounts where differences exist.A forthcoming OECD paper, "Comparing Growth and Productivity Performance in the OECD Area -The Role of Measurement", describes these other areas and provides qualitative assessments on the impact of these differences on the comparability of growth for OECD economies.