I agree this sounds like bad policy, but what's the logic for doing this with actual capital goods then? Doesn't that have exactly the same problem of limiting corporate investment?
The reasoning is that the deductions for expenses should be applied for the same year as the income they bring in. For expenses that will cover multiple years, they are spread out over those years.
The only logic was to make the Trump 2017 tax cuts look "revenue neutral". They were cooking the books so the CBO would give the tax cuts a passing grade.
https://americansfortaxfairness.org/ways-means-trump-tax-law... Quote: Corporations have traditionally been allowed to deduct all of their research expenses in the year incurred, even though a lot of research pays off slowly so its costs should similarly be written off over time. Adopting this position, and as a way to partially pay for its big corporate-rate cut, the Trump-GOP tax law decreed that starting in 2022 companies would have to write off research and experimentation expenses gradually: over five years for domestic research, 15 years for foreign. This requirement to “amortize” the expense over time reduces the value of the deduction, increasing corporations’ taxable income and requiring them to pay more in income taxes upfront. The Ways & Means legislation proposes to retroactively reverse this provision.
Accounting likes to recognize expenses with revenues. If an asset will be producing revenue for five years, its cost is recognized over that same time span.
So...don't you get value from the newly created software for ~5 years?
That's going to heavily depend on the type of software and whether it's sold as a shrinkwrap product or a subscription.
For example: your average AAA game will likely produce the vast majority of its value inside the first year upon its release.
At the extreme opposite end you have things like enterprise software using subscriptions, whereby the software continues to produce value year over year, but you're generally also paying developers' salaries to maintain and enhance the softare year over year. That, too, makes little sense in spreading out salaries as a cost over 5 years.
I can't really think of any cases where a piece of software is sold as shrinkwrap software, requires no ongoing maintenance/updates, and is expected to continue earning revenue for many years afterward. That just isn't the industry we live in.
> At the extreme opposite end you have things like enterprise software using subscriptions, whereby the software continues to produce value year over year, but you're generally also paying developers' salaries to maintain and enhance the softare year over year. That, too, makes little sense in spreading out salaries as a cost over 5 years.
There's also a ship-of-theseus problem here. How much change has to happen to a codebase before it's not the same software anymore?
The answer to that question can only be reliably answered in 5 years.
Actual honest valuation of software is something that requires actual evidence.
Software returns have extremely high variance. From a lot, to none, to high negative (For projects that don't complete, or worse, deploy to negative effect.)
Yep, some software systems become money pits. You end up having to pay more people to keep them running.
Only now if those people you have to pay to keep them running are software developers, you have to act like the money you’re spending on them is helping make new value, not merely paying interest on technical debt. Fun!