Canadian tax law includes a concept known as "arm's length". It is intended to make cheating or gaming the tax system more difficult.
The tax laws of this great country assume that two people who are not dealing at arm's length have coaligned interests. This presumably increases the possiblility that these people will try to take inappropriate advantage of some feature of the tax system, and therefore transactions between non-arm's-length actors are more carefully scrutinized by the tireless public servants at the Canada Customs and Revenue Agency (CCRA).
The definition of arm's length is stupid. Two persons are not at arm's length if they are related, or if there is a chain of related persons. What's related? A parent and child are related. Siblings are related. Spouses are related. A corporation is related to its controlling owners.
The upshot? Our company is related through a chain including blood and control to another company, which buys some of our time. Therefore we're not at arm's length with them.
The work we do is eligible for a special refundable credit under Canada's bizarre research & development tax rules. But since we are a non-arm's-length actor, the amount that is eligible for this refundable credit is less than the amount that they pay us -- significantly less, actually. The approximate equation as set out in Canada's charming tax law is A = B * C * D where A is the eligible amount, B is the amount paid, C is the fraction of our revenue that came from them, and D is the ratio of expenses to revenue. Expenses in this case includes regular salary (up to a cap) but not bonuses dependent on profit.
So in order to maximize their credits under this rule, they would like us to make C and D be as close to 1 as possible. That is, that we do all of our work for them (and thus derive all of our revenue from them, so C is 1). And that all of our revenue is expensed, which in our case means paid out as regular salary: no retained earnings, and no year-end bonus payments.
Both of these are against our interests, of course. It's nice to have a steady client, but if they need to cut someone from the payroll, contractors are always first in line -- blood relation or no blood relation. The standard contractor defense for this is to always have something else going, even if it's just a small side project.
Similarly with the salaries. We set our salary low deliberately, so that we only pay payroll and income taxes on the money we need to pay the mortgage, eat, and have a little fun. We try to retain money in the company (now that we're exempt from evil foreign investor tax) both so we can pay salaries even if there's a dry spell in revenue, and because we have this crazy idea of saving some money and investing it, y'know, maybe creating some jobs in the Canadian economy, creating some more value, building a business. Call me a foolish idealist.
So basically we have a situation where, because we are deemed to be not-at-arm's-length, our client suddenly cares deeply about how we run our company. Because small changes in our behavior can mean that they lose real money, they put pressure on us to behave in a particular way. Since we want to behave in a different way, we have arguments where we and they try to make some compromies between our competing interests. In summary: the tax code says our interests are coaligned so therefore we must be treated in this way; and the difference between our and our client's responses to this treatment shows very clearly that our interests are at odds.
The correct answer, in my opinion, is to eliminate the R&D tax credit. If this boondoggle of a credit is so widely abused that people who file for it must be presumptively treated like tax cheaters, get rid of it!
Oh yes -- there's another solution. Payments to employees up to 2.5 times the maximum pensionable salary are fully eligible for the credit. But we don't want to be employees.
And a double bonus: 2.5 times the 2003 maximum pensionable salary is about $300,000 CAD per year, or about $200,000 USD per year. So any company with a "key employee" who is worth more than that is not able to take the credit on the employee's full salary. The net effect will be to tend to encourage businesses to keep salaries for research personnel at or below that number. And the people who are worth more can go south -- brain drain, anyone?
Just had a whole week of nasty downtime, coming from a variety of directions.
It was mostly my fault: I hadn't been keeping the name servers in my whois records up-to-date when my own IP addresses changed. Tuesday, when a client's server (which just happened to be hosting my backup DNS) had its IP address changed by their upstream provider, that was the end -- because it wasn't just my backup DNS, it was all my DNS, as far as the whois database was concerned.
It took some hassle to fix that -- finding my whois password, reconfiguring the name servers on my current IP addresses. I had to get a new IP address for one of my upstreams too. Then I had to wait two days while the name server changes propagated through the top-level DNS system.
Finally mail works again, and I'm hit with a metric shitstorm of bounced spam messages forged from our work domain. Great. Then, as that subsides, I start getting postmaster messages -- double bounces -- from the client's machine. Apparently they're receiving -- more to the point, they're accepting -- e-mails from some site in Korea.
At least I have someone to complain to.