Approaching with math and large amounts of data are actually really great because you can start to account for variability and the flaws of traditional theoretical economics. Hard part now is determining who is twisting analysis for ideological/power reasons vs. staying true to the data/reality.
The problem is if you look at say home ownership as a balance between the costs of rent vs. loans that works at a very high level. But, second order effects such as decreased earnings from immobility vs pride in home ownership etc get really complicated if you want high fidelity models.
To use an analogy, classical physics is great for building walls, but it does not help you build 5nm transistors.
