The main reasoning for this is that it is hard for employers to distinguish between graduates if everyone has a 2.1 grade. But it is possible for employers to ask for a full transcript of individual grades, though this is not nearly as common in Britain as you might expect. The stronger point (which you might have already picked up on) is that the existing system can be difficult to interpret internationally. Adopting the GPA system would be helpful to undergraduates wishing to study or work abroad.
I think this might be missing a trick. My experience of the 1st/2.1/2.2 system is that it has a very strong effect on students’ work effort. For weaker students, either those of lower natural ability or the more workshy, fear of the notorious "Desmond" (cockney rhyming slang after the eponymous archbishop) is the ultimate motivator. Many attractive careers simply advertise the minimum requirement of a 2.1, and therefore getting the lower grade can be quite a handicap in the job market.
For stronger students, the aspiration of a first, the only true distinguisher in the system, is also a strong incentive. The risk is that working quite hard could leave you with only a high 2.1, largely indistinguishable from all other 2.1’s. The crudeness of the grading system drags everyone up.
An interesting paper by Pradeep Dubey and John Geanakoplos of the Cowles foundation at Yale Univeristy makes the same point. They write:
Suppose that the professor judges each student’s performance exactly, though the performance itself may depend on random factors, in addition to ability and effort. Suppose also that the professor is motivated solely by a desire to induce his students to work hard. Third and most importantly, suppose that the students care about their relative rank in the class, that is, about their status. We show that, in this scenario, coarse grading often motivates the student to work harder.
One might think that finer hierarchies generate more incentives. But this is often not the case. Coarse hierarchies can paradoxically create more competition for status, and thus better incentives for work.
They give a simple example. Suppose there are two students, Brainy and Dumbo, with disparate abilities. Brainy achieves a uniformly higher score even when he shirks and Dumbo works. Suppose, for example, that Dumbo scores between 40 and 50 if he shirks, and between 50 and 60 if he works, while Brainy scores between 70 and 80 if he shirks and 80 and 90 if he works. With perfectly fine grading, Brainy will come ahead of Dumbo regardless of their effort levels. But since they only care about rank, both will shirk.
But, by assigning a grade A to scores above 85, B to scores between 50 and 85, and C to below 50, the professor can inspire Dumbo to work, for then Dumbo stands a chance to acquire the same status B as Brainy, even when Brainy is working. This in turn generates the competition which in fact spurs Brainy to work, so that with luck he can distinguish himself from Dumbo. He doesn’t want to be mislabelled. With finer grading everyone gets their own label so this effect disappears.
The corollary to this in my example is that if the brainy student knows that even when slacking off he will still do measurably better than most students he may decide that he can still get a very good job with 70 to 80. There may be students who score 80 to 90 with superior credentials but academic performance is only part of the hiring criteria. If he can signal himself as a brainy student he might think this is enough.
However, critical to all this is that all exams are taken together, as they are at Oxford or Cambridge universities, usually at the end of the degree in a consecutive-day marathon. The trend in other British universities has been to examine various courses throughout the degree. The result is that those in the middle of the ability range can work very hard at the beginning, bank a 2.1 and then slack off in the remaining years. It is partly for this reason that those universities pushing hardest for the changes have exams split across years. Oxford and Cambridge are less keen.