I’ve been rereading some of the essays in Ben Bernanke’s Essays on the Great Depression, which of course is self-recommending. I thought this passage summed up some relevant truths:
Our [with Martin Parkinson] own view is that the New Deal is better characterized as having "cleared the way" for a natural recovery (for example, by ending deflation and rehabilitating the financial system), rather than as being the engine of recovery itself.
Bernanke notes that there were "remarkably strong" productivity gains throughout much of the 1930s, even though there was no capital deepening. This is a central puzzle which any account of the New Deal, or New Deal recovery, must incorporate. These gains seem to span more sectors than could be accounted for by New Deal policy alone, and note that most government interventions, even good ones, don’t bring productivity gains over such a short time horizon and in such a regular and sustained fashion.
Bernanke does suggest that some of the gains came from forced unionization and "efficiency wage" effects and yes that would credit the New Deal. But I doubt that is the best hypothesis and of course it contradicts the traditional account of profit-seeking behavior from businesses (why weren’t they paying the higher wages in the first place?). Rick Szostak’s work suggests that the New Deal saw lots of labor-saving, process innovations, which meant both high productivity gains and pressure on labor markets at the same time. In my view most of these gains were simply the result of working through the implications of the earlier fundamental breakthroughs of the preceding twenty years.
Whatever is the case (and we genuinely don’t know), these productivity gains are central to the story of New Deal recovery. Roosevelt may deserve credit for some of them, or for allowing them to proceed, but don’t assume that the New Deal caused such gains just because you see them in the gross data.
You can find different drafts of the relevant Bernanke-Parkinson paper here, with various forms of gating.