There is an argument that elections can be understood as a referendum on the previous 4 years. E.g. voting for the incumbent candidate/party means more of the same.

Yes but how would that say anything specifically about a particular topic unless it was a defining and undeniable circumstance that happened during the presidency like a war or major scandal - and even then I think it's hard to confidently say "this is why people voted for X" without proper polling (and even that we know how accurate it is - but at least it'd be data).

Because the single largest thing people vote for is how much better they're doing now than 4 years ago. That's just been true forever.

Exit polling asking voters their reasons for voting the way they did.

There was a poll asking "what was your motivation to vote the way you did", and "the state of the economy" ranked pretty high among Trump voters (don't remember where I saw it though, sorry).

As a general rule, it is very hard to win a popular vote (again) when the economy is weak for middle class and below. By definition, this would mean that more than 50% of people are not benefitting enough from current economic growth.

> How does the election tell you anything about the economy?

He didnt say it said anything about the economy, just the perception of it.

Now to answer that question, there is the simple old adage of "its the economy, stupid" that won Clinton his campaign. But the reality is that economy perception is always asked in polls, all year round. So trends can be established wtih decades of data. Secondly, voting attitudes can be understood through economic perception, when people perceive the economy is thriving incumbents do better seeing Biden's poll numbers anyone could tell that the economy perception was not good.