I would have thought they might decrease, but after googling as suggested actually came up with this post from 2011http://homebrew.stackexchange.com/questions/5000/how-much-hop-bitterness-dissipates-with-time
which discusses this research paperhttp://www.scielo.br/scielo.php?pid=S0100-40422000000100019&script=sci_arttext
For discussion here is the post linked above. Seems a bit aggressive on the decrease, but also notes the numbers are not precise. I'd credit the author of the post if I knew who mdma from Norway was.
Paraphrasing from Fundamentals of Beer and Hop Chemistry (text below), 2/3 of the hop bitterness in wort has a half-life in excess of 5 years, and the remaining 1/3 as a half-life of 1 year. From that, I've made a table showing the amount remaining over time. The total column shows the percentage of bitterness compared to the total at the start:
months cis- trans- total
0 68.00% 32.00% 100.00%
3 65.68% 26.91% 92.59%
6 63.45% 22.63% 86.07%
9 61.29% 19.03% 80.31%
12 59.20% 16.00% 75.20%
15 57.18% 13.45% 70.64%
18 55.23% 11.31% 66.55%
21 53.35% 9.51% 62.87%
24 51.53% 8.00% 59.53%
30 48.08% 5.66% 53.74%
36 44.86% 4.00% 48.86%
48 39.06% 2.00% 41.06%
60 34.00% 1.00% 35.00%
So, if you leave a beer for 1 year, it's IBUs will have decreased to 75% of its original value. For example, an 80 IBU beer will have 60 IBUs after 1 year.
Here's the relevant part of the text:
The ratio of the isohumulones depends on the reaction conditions. In the wort medium it is normally 68:32 in favour of the cis-compounds. However, the cis-compounds are much more stable (half-life >> 5 years) than the trans-isomers (half-life of ca. 1 year) during the course of time. This affects, obviously, the cis:trans ratio and has significant consequences with respect to taste and flavour stability.
Of course, other variables do come into play, so these figures can't be taken as exact, But then, as we all know, that's how it is with brewing. Our malt isn't exactly the color on the label (usually a range of values), and hops aren't the bitterness on the pack since they degrade from the get go. So, although not exact, these figures should offer a good estimate.