Service lives of R D 1

Service lives of R&D 1

• Other fixed capital stocks measured directly or through PIM. • Direct measurement of value of stocks of R&D is much more difficult, since R&D is by definition unique (no market price), much is own account production, and it may not be registered as an asset by the enterprises. • Means that R&D stocks will have to be estimated using PIM, so that data on service lives are needed. 2

• The length of R&D service lives has been estimated in academic research in the past using various assumptions. • In more recent work, panel data detailed by industry were used, and differences in service lives over time and between industries were explored. • Diewert (2005) found most of these econometric models unsatisfactory, but considered it possible to estimate the depreciation patterns using improved models. 3

• Following a suggestion from Charles Aspden (OECD) the Central Bureau of Statistics in Israel has examined an alternative solution from the beginning of 2007: obtaining information on service lives directly from producers or users of R&D, similarly to what is done to estimate service lives for tangible assets. 4

The pilot survey - framework • Data from important R&D performers in main industries. • If imports of R&D are very common, collection of data should also be made in framework of a regular business survey. • New survey on exports and imports of services indicates that imports of R&D significant in some industries. 5

The pilot survey - framework • Many important R&D performers produce R&D used by businesses abroad. • Their data on service lives less relevant for domestic capital. But due to their experience may be used for expert opinions for international use. 6

Choice of contact persons • A VC funds expert: the preferred respondents Vice President of R&D, CTO (Chief Technical Officer), or if it is an R&D enterprise, Product Manager. Marketing people or financial managers may not have the relevant information. • A number of representatives for each enterprise were interviewed. Financial managers sometimes were reluctant to respond. But R&D managers understood the questions well, supplied all information, often added important information that could improve the survey. 7

2 stages of the pilot survey • 1. stage: survey using only personal interviews, one or two representatives from survey department and representative from NA unit. Long questionnaire to determine how to pose the shorter questions in a regular survey. • 2. stage: shorter questionnaire sent by mail. For first batch sent, follow-up with a personal interview – to get respondents opinion. The second batch only by mail. 8

Information collected Length of stages covered in questionnaire: • Gestation lag : time lag between start and completion of R&D projects. Longer than a year for almost all R&D. • Application lag: between end of development and beginning of use. Often short. • The stage when R&D is used in production 9

Inclusion of lengths of project and application lag needed • Major parts of R&D is own account production, and for practical reasons expenditure on R&D will be classified as capital formation from the start of R&D projects. • Length of R&D project and application lag also needed for enterprises that sell R&D produced. Before sales R&D will be work in progress. 10

The feasibility of obtaining data • All respondents interviewed were able to give estimates of average length of service lives of R&D. • When more than one enterprise in a certain industry responded - similar length. • But some explained that length has changed in recent years, and become shorter. Implies that data need to be collected every few years. 11

Types of R&D • In many cases: more than one kind of R&D, each with special length of service life. • Examples: Semiconductor industry applications for communication and for transportation. Pharmaceutical industry new medicine and generic medicine. • Means that important to collect data on the composition of R&D. • It is possible to obtain composition, mostly separate units. 12

Service life and duration of R&D projects • Length of service lives appears to be closely connected to the cost and duration of R&D projects. If R&D project has a very long duration or is expensive, then service lives are usually longer. • Data on the duration of R&D projects were easy to obtain – the enterprises had structured working programmes for R&D projects. 13

Successful and unsuccessful R&D • The fact that part of R&D is unsuccessful was taken into account in work programmes. • Usually a time schedule for decisions to abandon unsuccessful projects. • Respondents were aware of the rate of success. • Revenues on successful R&D usually cover all R&D, also unsuccessful attempts 14

Length of application lags • The length of the application lags often quite short. • Enterprises work simultaneously on R&D and on designs for use of R&D in production, so implementation almost immediately. • Enterprises have tight work programmes for a number of years ahead, and are well able to respond to questions about application periods 15

Reason for ceasing to use R&D • Reason given by the respondents: use of new R&D, which replaced and improved upon the former R&D. • In most cases the old R&D asset is entirely abandoned, when replacement introduced, but in some cases it may be used in production on a minor scale to reap some remaining benefits from it. 16

17

Conclusion • Seems possible to obtain relatively firm and consistent responses on length of service lives. • Similar lengths for similar types of R&D indicate that collection of expert opinions may be sufficient. • Lengths could even be identical at a global level, so that international figures could be used. • Important that additional countries collect similar information. 18