Billions in support slipping away from our most promising companies

The single biggest government program supporting fast growing young tech businesses is under threat - and entrepreneurs are paying a heavy price.

The Research and Development Tax Incentive is the single biggest government program supporting startups in Australia. The program as a whole accounts for around $3 billion in Federal Government expenditure each year, of which about two-thirds is spent on companies with less than $20m in annual turnover. Support for startups under the R&D Tax Incentive dwarfs expenditure on any other startup programs. The program is open to any business undertaking eligible research and development (R&D). For startups, who are intently focused on R&D, it is a core part of business infrastructure. Since its introduction in 2011 it has been a principal reason for startups to employ Australians to undertake R&D, and a major driver for companies to keep R&D efforts based in Australia even when the business expands overseas.

Taking a lead from companies like Atlassian and 99designs, an increasing number of young Australian tech companies are choosing to ‘build local, sell global’ by keeping R&D teams in Australia - even when the business has shifted its headquarters overseas to meet the expectations of customers or investors. The R&D Tax Incentive is a major reason why this model is attractive. Federal political leaders have, for some time, been rightly proud of the program’s ability to support the rapid growth of new, high value employers in cutting edge technology areas.

HOW THE SCHEME HAS BEEN USED

The R&D Tax Incentive is set up in such a way that, despite its importance to startups, it has always been awkward for software firms to meet the requirements. Under the scheme, eligible R&D must follow a strict and extensively-documented scientific model (called the ‘Frascati model’) of hypothesis, experimentation, observation and evaluation, and logical conclusions. The results of the experiment must not be able to be predetermined, and the process must be carried out with the intention of generating ‘new knowledge’ (though it’s worth noting that the legislation specifically identifies developing new or improved products or services as ‘new knowledge’).

This process is highly suitable for lab experiments and scientific research (the ‘R’ in ‘R&D’). It is not so suitable for software development or other kinds of technology commercialisation. Nevertheless, the current iteration of the R&D Tax Incentive was designed specifically to include most software development as eligible R&D. When introducing the current legislation in 2010, the Minister said in his second reading speech to Parliament:

Recognising the pervasive nature of information technology in a modern economy, the new R&D Tax Incentive will ensure most software R&D is treated consistently with R&D occurring in other sectors.

As a result, it became established industry practice for software companies to claim activities related to the development of new, innovative products under the scheme. That practice was encouraged by specialist R&D tax advisers from reputable global accounting firms (alongside a growing number of boutique R&D tax firms). Politicians spoke openly about the positive impact the scheme was having helping exciting young technology companies grow. Implementing agencies also clearly supported the idea that software development was a core part of the scheme - in 2012, the year after the scheme was introduced, AusIndustry released formal advice to the ICT sector which said:

As a key enabler of business and innovation, ICT R&D is central to services, processes and products across virtually all sectors of the economy. Based on experiences gained from the R&D Tax Concession, AusIndustry anticipates ICT involvement in the new R&D Tax Incentive will be high.

Consistent with this advice, claims made by software companies were paid consistently for many years. The OECD updated the Frascati Manual in 2015 to explicitly acknowledge the validity of software development as a form of R&D.

Although the process outlined in the R&D Tax Incentive legislation doesn’t seem particularly well-suited to software development, it was intended to cover software as a core component of the scheme. This gap between the strict letter of the law and its practical effect is not uncommon.

Government programs are enshrined in legislation, which is hard to change, but the rest of the economy moves quickly. Interpretation of the legislation, either by the courts or by the agencies implementing the law, is often at least as important as the legislation itself. Clauses that seem narrow can often take on a broad meaning as the program evolves to fulfil its purpose in changing circumstances. One of the key things courts and agencies look to when deciding how to interpret legislation is the intent of the lawmakers in designing it - and one of the key pieces of evidence for that is the Minister’s second reading speech.

As the program evolved, technology companies still needed to conform to the basic requirements of the Frascati model, but there was a broad acceptance that this ‘development’ work was an important part of the scheme and the Act should be interpreted to take account of that. Claims weren’t put under the microscope. Compliance testing in relation to the scheme was low, as part of the overall strategy to allow companies doing development to participate even though it was difficult for them to conform strictly to Frascati in the way researchers could.

COSTS ON THE RISE

The R&D Tax Incentive has been the focus of some political unrest over the last few years. The cost of the scheme has been rising, leaving policy makers concerned that it could spiral out of control. The biggest growth in cost has been among small companies (less than $20m in turnover), which includes almost all startups. Partly in response to this concern, the Turnbull Government commissioned a review of the program, which submitted its findings to government in April 2016. The review made six recommendations to help strengthen the scheme’s ability to encourage ‘additionality’ (R&D that wouldn’t have been undertaken if the scheme didn’t exist) and to reduce cost by increasing integrity.

The Turnbull Government did not formally respond to the 2016 review until more than two years later, when it introduced changes to the R&D Tax Incentive as part of the Federal Budget in May 2018.

TIGHTENING IN THE PROGRAM: SOFTWARE CLAIMS GET SQUEEZED OUT

In the meantime, concerns about the cost of the scheme continued, and the program came under pressure. In February 2017 the ATO and AusIndustry issued a Taxpayer Alert seeking to clarify the position on the eligibility of software claims under the R&D Tax Incentive. The Alert noted that the ATO and AusIndustry were ‘reviewing the arrangements’ for software companies claiming the R&D Tax Incentive. It referenced concerns held by the ATO and AusIndustry that software companies were not conforming to the ‘stringent requirements of the laws that govern the R&D Tax Incentive’.

StartupAUS cautioned vocally about this shift when the Alert was first issued. The Government responded by saying that nothing had changed: the law remained the same as it always had, and the Alert was simply meant to provide guidance on how claimants could conform with the law. Senior administrators were adamant on this point. Partly, this was true - the letter of the law hadn’t changed. But its interpretation certainly had, and startups began to feel the effects right away.

Eligibility of claims started to be called into question. Companies who had been claiming, on sound advice from seasoned R&D Tax Incentive advisers, similar activities for multiple years in a row began to be told their software development activities didn’t qualify. This is particularly concerning as audits can reach back up to seven years, retroactively disallowing multiple years of already-paid claims and putting companies under extreme financial pressure.And - perhaps most significantly at a system level - companies began to see R&D claims as risky, and started to self-censor. A common experience right across the sector is that R&D claims for legitimate software development are being pared back substantially to try to reduce the risk of facing a potentially catastrophic clawback.

Gradually, the R&D Tax Incentive has been slipping away from startups.

This has been compounded by an overall reduction in the scheme for all claimants. In September 2016 entitlements under the scheme were reduced by 1.5%, to 43.5% for refundable claimants and 38.5% for offset-eligible claimants. In the 2018 Budget, further constraints were proposed, including the introduction of a new ‘intensity threshold’, and a cap on claims at the refundable end of the scheme. Under these changes, R&D payments are would be tied to company tax rates - which means if the Government is successful in reducing small business tax rates to 25%, R&D payments will drop even further. For many loss-making startups (who don’t yet pay tax), that could amount to yet another reduction. These changes have now been shelved until after this year's election.

CRACKING DOWN

During his Budget speech in May 2018, Scott Morrison (then the Treasurer) announced that there would be a crackdown on R&D claims. Additional resources were allocated to allow more audits and reviews to be conducted:

We are cracking down to ensure that R&D tax incentives are used for their proper purpose, with enhanced integrity, enforcement and transparency arrangements, saving taxpayers $2 billion over the next four years.

Software companies are rightly concerned. The result has been that many legitimate software claimants have fallen out of the system, with founders worried by the increased risks associated with making a claim for software development.

A SMALLER PIECE OF A SHRINKING PIE

All of this amounts to a significant cut to the R&D Tax Incentive. In fact, we know the quantum: Morrison’s Budget speech notes that Treasury has estimated that the crackdown alone is worth about $2 billion over four years - reducing the overall value of the scheme by about 17%. But this reduction is not being framed as a cut, because the context is that Australia’s total spend on R&D is already well below the OECD average at around 1.87% of GDP (OECD average around 2.3%) and is falling. Australian businesses spent $16.7 billion on R&D in 2015–16 compared to $18.9 billion in 2013–14, a decrease of 12%. Instead, this tightening of the scheme is being framed as ‘boosting integrity,’ a process which the Government says will benefit claimants in the long run by keeping costs in check and allowing the scheme to stay alive.

In practice it is a way of reducing costs without making the case for it to the public - a political result without a political cost.

At a time when Australia needs to be thinking ambitiously about how to encourage more R&D, we are tightening the belt.

CHART: GROSS DOMESTIC SPENDING ON R&D (TOTAL, % OF GDP, 2000 – 2016) (OECD data)

STARTUPS ARE IMPACTED MORE THAN MOST

Startups - particularly software startups - find themselves in a particularly difficult position here. They’re engaging in a form of development that doesn’t always neatly fit the process outlined in the legislation. They’re also generally small and short on cash runway, which means even the threat of an ATO clawback can have very serious (often existential) business ramifications. Few are likely to hire armies of lawyers to fight any adverse ruling (something larger claimants would do as a matter of course). As a result, if they percieve that software is no longer welcome in the R&D Tax Incentive program, it’s reasonable to expect that lots of these businesses will drop out of the scheme entirely, or substantially reduce their claims regardless of their merits.

We are already beginning to see this occur. To many large corporate claimants, whose R&D claims might form a small percentage of annual revenue, talk of a ‘crackdown’ may be run of the mill. But cash-strapped startups are often counting on these payments, with many even borrowing against their future R&D Tax Incentive earnings to cover immediate cash shortages. The consequences of an adverse audit would, in many cases, be dire: businesses could fail or be crippled by debt and the personal risk for founders is extremely high.

But despite the huge individual consequences of businesses failing because of adverse audit findings, it may be that system- wide under-claiming is even more problematic across the program as a whole. Huge sums of money, tens or hundreds of millions of dollars, that would otherwise be supporting the growth of Australia’s startup sector may be forgone. For a group of businesses that are always on a knife-edge financially, that puts a large number of other companies - those not subject to an audit but whose claims are diminished by the downwards pressure on the sector - at risk.

AT A TIME WHEN AUSTRALIA NEEDS TO BE THINKING AMBITIOUSLY ABOUT HOW TO ENCOURAGE MORE R&D, WE ARE TIGHTENING THE BELT.

THE COST: JOBS (AND GROWTH)

Reducing expenditure on software R&D is not just bad for startups - it a bad idea for the economy as a whole. The R&D Tax Incentive is a program used extensively by many of Australia’s most forward-thinking businesses to help fund new research and development programs that employ thousands of high skill workers.

Startups are no exception. In fact, they are some of the scheme’s highest-value claimants, representing genuine return on taxpayer dollars, particularly when it comes to jobs. The R&D Tax Incentive makes up a critical part of the funding landscape for Australia’s emerging software sector. StartupAUS data suggests almost 9 in 10 startups (89.2%) rate the program as either ‘very important’ or ‘critical’ to the success of their business.

CHART: HOW IMPORTANT IS THE R&D TAX INCENTIVE TO YOUR BUSINESS? (Source: StartupAUS survey data)

CHART: STARTUPS’ INTENTION TO USE R&D RECEIPTS (Source: StartupAUS survey data)

We also know that startups use the funding from the R&D Tax Incentive disproportionately to hire more staff. When asked in a 2016 survey how they would spend an increase in R&D Tax Incentive, founders overwhelmingly said ‘hire more staff doing product-related research and development’ (82.4%).

This should not come as a surprise. We know that young companies are by far the biggest drivers of job growth across the economy. Data from the Australian Department of Industry shows that new firms create substantially more jobs than established ones, and that in Australia firms up to three years old created 1.44m jobs over the six years to 2011 compared with a net loss of 400,000 jobs by established firms over the same period.

The report notes that ‘the bulk of this employment growth is driven by a relatively small number of high-growth- orientated startups’. It goes on to say that ‘while innovative entrepreneurship can disrupt competitive markets, it also has the potential to nurture business dynamism and economic growth. Like many OECD countries, Australia is in the midst of an economic transition. Australia’s situation is different in that it is not so much seeking recovery from a downturn as searching for new sources of growth to balance the relative decline in resources sector investment. The role of the entrepreneur is central to this process.’

The definition of startups referred to here is broader than software or even tech companies, and the numbers are correspondingly larger. But the broad point remains absolutely fundamental: young, high growth businesses create all the jobs. They are therefore some of the highest-value recipients of this sort of support.

This is not a uniquely Australian characteristic - it is true of economies across the developed world. A study by the Kauffman Foundation in 2015 found that 1.5m new jobs are added to the US economy each year by new firms, while over an extended period existing firms have been net job destroyers, losing a total of 1m jobs per year. The report emphasises that it is not small businesses that are the primary engine of job creation, but rather young businesses.

In fact, figures from across 15 OECD countries show that over an extended period, including during the global financial crisis, new businesses have consistently been net job creators whilst existing business have been net job destroyers.

CHART: NET JOB GROWTH - NEW VS ESTABLISHED FIRMS, 2001-11 - AVERAGE OVER 15 OECD COUNTRIES (Source)

BAD ACTORS

There remains a concern among some policy makers that bad actors - specifically fraudulent over-claimers - are a significant source of the growth in cost of the scheme. We’ve already seen that it’s clear from the history of the scheme, the intention of the government when it brought the scheme into operation, and the way it’s been implemented to date that software companies form a legitimate part of the program. But are there some bad apples ruining it for everyone?

The projection of savings based on the ‘crackdown’ is some $2 billion over four years, or roughly one sixth of the total value of the scheme. It is highly unlikely that genuine fraud accounts for such a large percentage of claims, especially when considering the barriers to entry (i.e. anyone attempting to defraud the scheme must first form a company, spend significantly on some activity, and often have that claim prepared by a professional R&D Tax Incentive adviser.) While there may be some bad actors - as there are in almost any scheme - the vast majority of software claimants are acting in good faith, on advice from qualified professionals, and in line with expectations set by government since the introduction of the scheme.

Even if genuine fraud were a significant factor, reducing expenditure on software R&D across the board is not the right way to fix this problem. It is a blunt-instrument response to a very specific problem which has the effect of undermining the scheme’s effectiveness as a whole.

THE RISK

There is a serious risk that a depleted R&D Tax Incentive could begin to cripple Australia’s burgeoning startup sector at a critical time. The program has become a vital piece of infrastructure in the startup and innovation landscape. Reduced access for startups is likely to result in:

• A reduction in working capital for high growth technology startups
• Lower growth rates for affected companies
• Lower hiring rates for software development firms
• Decreased R&D output as support diminishes
• Failure of some companies, including particularly R&D intensive businesses counting on a reliable source of R&D support
• A reduction in Australia- based R&D by global companies
• A contraction in the number and quality of Australia’s new technology businesses

THE FIX

There is no easy fix here. Administrators maintain that there has been no change in how the scheme operates, making it difficult to identify ways to halt the damage.

In the immediate term, there needs to be some assurance for vulnerable software companies. Companies with turnover of less than $20m that have been claiming the incentive, in good faith and on credible professional advice, need an assurance that they are not going to be subject to audit processes unless their claims are manifestly unreasonable or have had sharp unfounded increases. Such a moratorium should remain in place until the introduction of a clear legislative fix to the way the R&D Tax Incentive operates or a new scheme that directly supports software development is implemented. This would would help address uncertainty and reduce existential risk for good-faith claimants.

Specifically the moratorium should:

• Be applied to all software claimants with turnover less than $20m.
• Prevent new audits and halt existing audits as long as
• the claim in question is similar in scope to claims submitted in previous years (i.e. similar technology area);
• the claim is not more than 50% larger than the previous year; and
• the claim has been developed by a specialist R&D advisor (professional services firm).

Alongside this, a clear path to supporting technology companies doing genuine development work in Australia should be formulated. It has become apparent that the language of the R&D Tax Incentive, as it stands, cannot be relied upon to produce consistent results for software companies. New language in the R&D Tax Incentive legislation that caters to the specifics of the software development process would help revitalise the program as a vehicle for the sort of support emerging tech companies need. Failing that, an alternative program focused on the development of innovative technology and driven by job creation as a primary metric would help decouple the ‘research’ and ‘development’ aims of the scheme while supporting growing tech businesses to deliver the additive economic value of their development activities.

Alex McCauley is the CEO of StartupAUS, Australia's national startup advocacy organisation. This article originally appeared in StartupAUS's annual Crossroads Report in D, and was prepared in collaboration with a range of startups, VCs, and R&D advisors.