I ran across another short analysis that I agree with. The analysis repeats my points about the authors being anti-regulation economists and the paper not being peer-reviewed, and it goes more in-depth about a problem I only hinted at, which is data selection.
In the paper itself, they use definitions in unhelpful and misleading ways. For example, they define "lockdown" as "the imposition of at least one compulsory, non-pharmaceutical intervention." So while you might think of Lockdown as governors ordering non-essential businesses closed for awhile starting in March 2020, this definition also includes public transit closures or even mask mandates (though not recommendations). Also, there's no nuance for how "compulsory" an intervention actually is: just because masks are required doesn't mean everyone's actually going to wear them, and just because businesses are supposed to close doesn't mean the police will force them to. But no, if there is a public authority declaring something compulsory, it counts as a lockdown. (In fairness, they don't always abuse this definition as much as they can in this study, but it does lead to other problems).
Next, the data. This study is a meta-analysis, meaning they combined the data from other studies, and excluded ones they deemed unsuitable. There's a lot of room for mischief here. A major point is that they ignored studies which relied on simulations to make claims of death averted by lockdowns. So even though COVID was following epidemiological predictions pre-Lockdown, and we all saw the death rate drop right after Lockdowns were put in place, they ignore that data because, well, there was no guarantee it would continue increasing exponentially without a lockdown. They justify this by saying that other factors may have led to the sudden decrease, like changes in the weather prompting changes in behavior. You can't really use this as an argument against Lockdowns though, because the point of the Lockdown is to affect behavior, and you can't count on the weather to change our behavior whenever there's a COVID spike.
So what studies do they include? Of the 34 studies, 12 of them were, themselves, Working Papers (i.e. not peer-reviewed). 14 studies were in Economics, while only 1 was in Epidemiology. Of the papers, half are from 2020, and none are more recent than June 2021 (so nothing on Omicron, and little on Delta). Within the 34 studies, they give more weight to studies looking at lockdowns after May 2020 over studies looking at lockdowns before June 2020 (basically ignoring the timeframe of the major Lockdown). They also distinguish between studies which look at mortality within 14 days of lockdown and those that look at later mortality, arguing that it should take longer than that for people to go from infection to death unless something else was already wrong with them.
Finally, there's this galling statement: "While it is true that epidemiologists and researchers in natural sciences should, in principle, know much more about COVID-19 and how it spreads than social scientists, social scientists are, in principle, experts in evaluating the effect of various policy interventions. Thus, we distinguish between studies published by scholars in social sciences and by scholars from other fields of research. We perceive the former as being better suited for examining the effects of lockdowns on mortality." Wow.
