The report you link is not a meta-analysis (at least not one I recognize). Also, meta-analysis is a freaking minefield: garbage in, garbage out rules the day - i.e. the quality of any conclusions is limited by how honestly and competently it was performed, and even then, errors in the underlying papers can cause significant undetected problems - and worst of all, because a meta-analysis is a big blender, it makes it harder to find real problems distorting the conclusions; there's just more to slog through. If you're really scrupulous and there's no other way, I get that it's a nice way to eke out a bit more statistical margin, but it's such a easily misused tool - if the topic is at all controversial, I personally wouldn't trust em for pretty much anything. But hey, this report doesn't seem to be such an analysis.

That lockdown report makes exactly the same kind of errors your earlier reasoning did (i.e. attribute costs to lockdowns, rather than the pandemic, and ignoring possible benefits); I think the author is biased and decided not to follow the evidence, but rather to collect whatever arguments he can to arrive at his forgone conclusion. Notably, he spends lots of time trying to argue that the benefits of lockdowns were actually due to voluntary social distancing, but fails to do the same for the costs. The article smells like motivated reasoning - and given a topic that's so utterly impossible to study in isolation of course you can always find legions of reasonable sounding arguments either way. Also, the author is fairly blase in dismissing epidemiologists, which he is notably not - probably because expert opinion doesn't agree with him. And of course - if you define away the benefits to zero and assign all the costs to this account - hey, then the cost/benefit looks bad. But the devil in is the details, an that report makes no case as to why his particular way of accounting is any better than anybody else's.

I meant to leave it at that, but that paper is so bad, I can't help myself: Like, what's up with Table 2: that's a classic case is intentionally confusing correlation and causation (and it's so egregious the paper's author tries to come up with some flimsy excuse for it) - countries that have worse outbreaks will choose to have greater stringency, but of course he literally makes the argument that because this is true, lockdowns don't work. Say what? What a hack. If the author were honestly trying to tease out subtle causes and effects rather than merely ranting along his preconceived lines, he would never have included BS like that as part of his argument.

And if you google the guy - he's clearly been willing to lie with statistics before, including in a court case about same-sex marriages. This paper: same old, same old.