I feel like this should be obvious but wide roads don't imply sitting in traffic. If traffic is 1.5x but road width is 2x then you're less likely to be sitting in traffic.
That correlation analysis is not good, you've got N=1 there and an astronomical number of confounders. A better analysis would be to look at inter-city or inter-municipality variability in crime and correlate that with the cross-section of housing price changes. You'll have a rich N=50 to N=5000 dataset.
I don't agree with your assertion that homelessness isn't caused in part by house prices. That seems unintuitive to me. Yes some people will get roommates, but what about people that already had roommates and were barely meeting rent? Some of them will have the foresight to get up and move interstate before they get wiped out, but there's frictions and costs associated and not everyone will succeed.