It’ll be harder for software companies to receive R&D Tax Incentives from 2010

20 September 2009

My first engineering job was with a defence R&D organisation. To the scientists and academics there, research was for the creation of completely new work or ideas (undiscovered work).  The technical risk involved was that the proposal was wrong or impossible, and that alone could take years to prove.

Software companies, however, have been able to receive the R&D Tax Concession for work that largely involved “customisation and integration of existing systems”, mainly on the basis of spin around the technical risk involved.  In 2010 the Australian government will be releasing new R&D Tax Incentives that intentionally prevent these “questionable merit” cases, among several other improvements.

I agree with the department’s assessment that these cases were dubious. Customisation and integration can be technically challenging but the risk is predominantly schedule, financial or organisational (Ie. throw more money, time or the right people at it and it’ll get done). It’s not novel just because no one else has integrated the systems before.

The good news is that the incentives are more generous under the new scheme,  budgeted under the assumption that fewer companies will be eligible.  The changes intentionally favour small and medium sized businesses believed to be more responsive to fiscal incentives.

The main improvement is that the old tax deduction will be replaced with a tax credit. A tax credit is a reduction in the tax payable (independent of the company tax rate) and critically, it can be carried forward if the company doesn’t make a taxable income in the financial year.  That’s awesome.

Also of note to software companies is that activities supporting R&D are still permitted under the new scheme, but small R&D activities won’t be allowed to trigger large supporting activities.  The department isn’t quite sure how this will be determined yet, perhaps by proportion, sole-use test, a lower rate of assistance or the like.  I saw this feature exploited by a lot of software companies under the old scheme.

The explicit rules around software R&D haven’t been written yet but there’s likely to be new tests for eligibility. The consultation paper suggests a model used in the UK that explicitly defines the software likely and not likely to be considered (eg. search engine algorithm is good, creation of website is not). I suspect that’ll quickly become antiquated as its based on today’s software focus.