It doesn't matter so much when comparing people who took exams around the same time. But it really DOES matter when comparing today's 20 year old against a 40 year old, both applying for the same job. Today's grade A was probably something like a grade C, 20 years ago, so the older applicant looks a lot worse on paper and may not even get through the initial screening process.
It also matters because a decade or two ago, someone with a string of As was really something special. Now it's commonplace. Employers need a way of selecting those with genuine ability. How do they do that when they get large numbers of applicants, all with A grades? It devalues the achievements of those at the top. Hence why every few years, there's either a new higher grade announced or a change in grading structure. Before long there'll be a grade 10 at GCSE or a 9 or 9+ to break up all those currently getting a 9. Likewise at A level, there'll be an A to break up those with As.
Furthermore, it makes it harder for Unis to select the most suitable applicants for their courses. It was almost unheard of for Unis to "require" an A* at A level, but now it's seen more and more as they get too many applicants with As that they need a new way of differentiating.
On a slightly different tangent, I also find it unhelpful that exams are graded according to a normal distribution curve, which means one year may need 90% to get an A*, the next year it may only need 85%, etc - so there's no comparison between years nor applicants who've done exams in different years. Yes, I know it's claimed to be due to "smooth out" variations in difficulty between years, but surely it would be better to set the questions properly so that each year is the same in terms of difficulty! It's how professional exams work - my accountancy exams had fixed pass marks year after year - the percentage required didn't change between years and strenuous efforts were put in place to ensure consistency of difficulty between years. Why can't that be done for GCSE and A Levels?