Alternative to forecast sequential regressions

Question

Seeking advice on how I could forecast the below better. Ultimately I am looking to build a 5-year rent growth forecast, but as part of the forecasting process, I will need other variables.

Currently, I am forecasting by using a sequence of linear / OLS forecasts. The reason this path was taken is because it is simple and follows industry intuition. I can also explain the rental forecasts based on the space available. Also, given construction pipelines are available ahead of time, we can get a good sense of how much markets maybe oversupplied in upcoming years.

The steps below is also done for varying markets with varying number of observations. Some markets have a lot of data, others, not so much.

Step 1

I forecast demand by using external macroeconomic variables. For these I am using a third party data provider and take these variables and forecasts as given.

Y_demand = Beta * X_macro

Step 2

Then using the demand forecast feeds into a calculation for forecasted availability (space that is currently not occupied). I forecast supply by looking at the delivery dates of projects currently under construction.

Space Available = supply - y_demand

Step 3

Finally, rent growth is forecasted by looking at the available space minus its LT Avg. This was done to have the space available data be roughly normally distributed.

Y_rent = Beta * (space available – long-term average) 

Why I am looking to change. This process is incredibly time-consuming. Also, I believe by running multiple step regressions, I am introducing errors at each step. Finally, I am also getting requests to forecast multiple new markets with varying degrees of data quality. So I believe I need an efficient way to build my forecasts.

I was looking into utilizing Hierarchical Bayesian (HB) methodology, but I am unsure how to start or structure the forecasts. The reason I was looking to use HB was because of the ability to utilize information where data is available, to help with forecasts for markets where the number of observations is lower. I am unsure if I can forecast simultaneously the demand and rent components or if they will need to be sequential as well. Any help would be truly appreciated.

NOTE: I also was thinking about SVARS. While this may help the problem of forecasting variables sequentially, it does not help when I am forecasting markets with low number of observations.