They never actually measured the amount of work hours being done remotely. They never measured the number of employees working remotely. They just had some guy make up "remote-ability" numbers for each sector and then carefully pretended to be doing science. Meanwhile all these sectors have seen unrelated changes to operating environment that have not been controlled for. It's BS on BS on BS.
Overall I am not even convinced their remote work metric would correlate with hours worked remotely.
Getting access logs for whether people logged in from home or the office is really not a crazy stretch.
Measuring the dependent variable "productivity" is actually much harder since firm-level productivity (aka earnings) has a whole host of other things that impact it besides worker-level productivity and we all know how hard it is to measure developer productivity.