In the first iteration of Circles the basic income would be stable of n units per time per person. To counter unlimited new creation of coins there would be a demurrage factor in the range of 5-7% per year. This would eventually lead to a stable ratio total money supply (per person in the system) at (for 5%) 20x the yearly basic income. This is because at this total money supply 5% of it (that is destroyed in demurrage) equal to the amount of basic income (new creation of money).
In the current version of Circles the x% demurrage is replaced with a x% growth factor of the basic income. Instead of having a constant issuance of n units the issuance in is increased by x% per year or simply: n * (1+x)^years.
This will result in the same kind of stability in that sense that the ratio of basic income per person to money in existence per person will stabilize at a rate of 1:(1/(x%)) = 1:20 in the case of x=5.
Now wether money increases or decreases in purchasing power (inflation or deflation) depends on many other factors. (What is the total amount of goods/trust that can be bought with the money) But giving everything else equal I would make the claim that a scheme with demurrage and stable money supply (and no inflation) and a scheme with increasing total money supply that will cause inflation are equal.
The reason we decided for the second approach is simply because it is closer to what people are used to today.