It may be that it's just not obvious what the individual rent values are in the model, and that they do go down when the supply goes up. The only numbers visibly shown in the model are the number of units, the tax revenue, and the number of "affordable units"; the model doesn't display the expected distribution of rents, or the actual threshold for "affordable". As a result, it's hard to see how the parameters actually affect individual rents.

For the primary target audience (city planners), this makes sense.

Oh, sorry, I thought you meant my (price) model, not the Berkeley (production) model!

Unrelated: Why is the wages coefficient negative and large in your regression?

I think you're reading it backwards. The coefficients are the exponents that follow the parameters. The large negative one is the housing supply, because prices go down as more housing exists.

Thanks