One piece of analysis I'd be interested to see is whether the transition probabilities among the 4 categories along with the implied 5th category of "not homele

One piece of analysis I'd be interested to see is whether the transition probabilities among the 4 categories along with the implied 5th category of "not homeless" have changed. This is harder to get good data for because it would require tracking individuals across time. The payoff is that it would likely reveal evidence of structural changes if such changes exist. At the population level it is hard to tell whether and where changes are happening. Getting the transition probabilities would allow for more effective targeting of interventions (though most of the folks who work on this already know which transitions are the hardest to recover from).