Beyond self-reports and perception-based outcomes, most extant studies that I’m aware of have found decreases in real output. For example, a randomized controlled trial published by the NBER found that productivity of employees randomly assigned to work from home was 18% lower than employees randomly assigned to work in the office: