-
Notifications
You must be signed in to change notification settings - Fork 5
Market research model
A market research workflow would involve one or more steps in an UrbanSim simulation to provide useful input to the real estate pro forma analysis. Some of the benefits of such a workflow would be:
- Provide an endogenous feedback loop between real estate feasibility analysis, and conditions in the market (e.g. absorption, vacancy, prices). This would help provide a more organic, market-driven process within the simulation.
- Add an element of speculative behavior to simulated real estate developers; this element is, we think, realistic but is not really modeled
- A mechanism for filtering out inputs to the feasibility model, to (potentially drastically) improve speed and enable simulation at smaller time steps
Which buildings get built in each simulation year is currently determined by two models:
- The pro forma step calculates profitability for all parcels in the region. The primary endogenous inputs are 1) the current rents of the parcels, and 2) the typical rents for each use for each parcel. This does not currently use information about housing supply other than indirectly through prices.
- The developer step selects parcels for development. Parcels are assigned a development probability based on results from pro forma. A target number of units to build is assigned. This is either an input, or calculated using the number of existing agents, number of existing units, and a target vacancy. Finally, enough buildings are built to match that unit target.
- No feedback about demand/supply match other than through prices
- All parcels are assessed for feasibility in every year (computationally wasteful)
This model workflow is called a workflow because it involves several different steps at different points in the simulation.
We have to record information at certain time steps in the model to perform later market research calculations. This might include, for instance, information about vacancy or prices at the building scale at a point in time. We can save these as Orca columns to be used in later steps.
For example:
-
For each building in the buildings table (to calculate after hedonic regression):
- Residential price per square foot (this and each column below tagged with the simulation year)
- Office price per square foot
- Industrial price per square foot
- Retail price per square foot
-
For each building in the buildings table (to calculate after location choice models):
- Used residential space (this and each column below tagged with the simulation year)
- Total residential space
- Used non-residential space
- Total non-residential space
These are simple, and prices are already calculated, but not saved with each simulation year. Saving these results in this way allows us to calculate trends later.
These are the core models, and could all be housed within a market research model class, accessed through an Orca step right before feasibility calculations in each year. Specific models could include:
- Absorption
- Absorption trends over time
- Price trends over time
- Land residual
- "Hot market" identification (based on actual development behavior)
Each of these needs to be further specified.
One advantage of using a preliminary market research model is that we can smartly filter out parcels from the feasibility step that have little to no chance of seeing new development. This part of the model should return a reduced table of development projects: only those that have a reasonable chance of getting developed.
This output needs to be carefully designed for each market research model.
Some of the market research outputs should also serve as inputs to the feasibility model to actually factor into the profit calculation for each development site. For example, lower absorption should lower the revenue for a potential building, changing its profitability.
This output needs to be carefully designed for each market research model.
The number of developments that get built are determined in the developer model. Currently, the rule is based on an exogenous target vacancy number, or an outright target unit number:
- If the number of target units (derived from target vacancy) is 0, don't build anything
- If the number of target units is greater than the number of profitable buildings, build everything
- If the number of target units is > 0 and < the number of profitable buildings, select buildings by probability (based on profitability)
This needs to be adjusted to get closer to the assumption that, development caps notwithstanding, profitable buildings will get developed.
A general timeline should be:
- Simulation year 2020
- Transition, hedonic models, LCMs
- Data collection: vacancy, prices for all buildings, tag as year 2020
-
Market research
- Calculate absorption for buildings built in 2019
- Calculate absorption trends between 2016 and 2019
- Calculate "hot" neighborhood trends based on development activity between 2017 and 2019
- Calculate price trends between 2017 and 2020
- Determine which buildings are candidates for feasibility analysis (send to pro forma)
- Send model results to pro forma
- Pro forma
- Only runs on candidates
- Takes into account market research model results
- Developer
- Picks buildings based on new rules
- Pipeline
- Tag year new buildings were built