That theory is obviously not compatible with the excess mortality graphs moving in lockstep with the Covid death graphs. If the virus has no correlation with deaths and it is all a mis-attribution, why would the pattern of deaths match so well with the pattern of excess mortality?
Excess mortality being a model rather than ground truth is irrelevant too. It still doesn't explain the correlation.
The claim about an inert virus was an extreme example to make the point clearer. Imagine such a virus did exist (which it doesn't). Then by the methodology used to compute COVID deaths, there would still be a large number of deaths attributed to the virus even as excess deaths didn't move at all, even though that should be a statistical impossibility.
The question is therefore not "is SARS-CoV-2 entirely inert", obviously that would be ridiculous. The question is "to what extent are the reported numbers confounded by age and co-morbidity effects". The Economist is arguing that excess death figures show the confounding doesn't exist or must be so small it hardly matters. I'm arguing that by their own data (which seems to be continuously updated even long after the article is published?) there are countries where either the numbers are over-counted in this way, or the mortality models are seriously wrong, because COVID deaths > excess deaths.
The Economist tables show this. Britain reports 10,000 more COVID deaths than excess deaths, and the over-count is surely much higher than that given that the healthcare system now has huge backlogs with people dying on waiting lists for non-COVID related diseases. Denmark also reports far more COVID deaths than excess deaths. Some of those are created by policy, and the proportion will only increase over time given the vaccines + generally screwed up health system.
Then there's the question of how excess is calculated. It's excess vs a model. What matters here is not really what's true but what China can argue is true. Modelling is discredited in the west now, in the unlikely event they cared to respond at all, they would certainly argue that any calculations of damages were based on motivated reasoning and bad modelling (and then present their own bad modelling in response).