To me, the slight economic premium for the A shares always represented the value in the extra voting rights inherent in the A shares. For a long time those extra voting rights were not worth much because Buffett and Munger had such a large, outsized influence in the company that shareholder votes really did not matter all that much.
As time has gone on, Buffett has slowly been disposing of his shares and more and more As are being converted into Bs. Now Munger is gone. The extra voting rights of the A shares are becoming more valuable.

That's all logical, but to me it seems less than sufficient to explain the observed drift. Logic is rarely enough to make sense of the market!

The A:B premium was pretty steady for many years before the split at the time of the BNSF purchase. Around 0.3%, if I recall.

It dropped to essentially zero premium for a few years after the S&P 500 addition.
Other than occasional excursions, it was basically nothing at 1500:1 almost all of 2010-2020.

There was a bump in 2021 which faded, then it started climbing substantially and steadily in the last 18 months.
The recent levels haven't been seen before outside of chaotic bear markets when the B holders, the relative weak hands, seem to sell faster.

That just doesn't seem to match the pattern of buybacks and the concomitant increase in value of the voting premium.
To me, it seems that something new is going on. Maybe not something important, but it remains a bit of a mystery to me.

Maybe as simple as a bit of shift in marginal relative demand because of the ability to buy fractional A shares at some brokers?

Jim