The map puts two homes in the same neighborhood. They are a few blocks apart and similar in size. An analyst may be tempted to treat one sale as evidence for the other. A buyer can see why that is too quick. One address has a direct walk to the station. From the other, the walk crosses a road that makes the trip feel twice as long. The addresses are hypothetical; the problem is not.
Property markets are spatial, but space is not a neat set of colored shapes. Buyers search along school routes, commute lines, family ties and budgets. Insurers and lenders may draw their own boundaries. A new apartment building can change nearby supply. A flood map can alter the perceived risk of one side of a street before a neighborhood average moves at all.
The wrong kind of nearby
A spatial microsimulation keeps those distinctions inside the mechanism. It represents homes and buyers in places, then lets search, comparison and transactions unfold across them. A buyer who fails to win one house may move two streets over, or may leave the area entirely. A price change can pull demand away from another listing. The model can ask which homes were actually in the same buyer's choice set rather than assuming every property inside an administrative boundary competed equally.
Some boundaries behave like walls for one buyer and slopes for another. A family set on a particular school zone may exclude an address across the line regardless of a modest price cut. A commuter may dislike crossing a busy road but accept the trip for a better house. The same road may be irrelevant to someone who works from home. A map that gives every buyer the same response to every line is not a model of local choice; it is a map with people painted over it.
Researchers have used spatial agent-based modeling for housing in Greater Sydney, representing different buyer groups and local areas. The point of citing that research is methodological. It does not give another city a ready-made map or a calibrated set of buyer preferences. A model must learn the local market from local evidence and test whether its spatial rules reproduce observed patterns.
Imagine that the hypothetical homes sit near a school boundary. The boundary shifts. One property now falls inside a zone prized by some buyers, while the other does not. A citywide average might barely move. For these homes, the set of plausible buyers can change sharply. A spatial model could simulate how that demand might redistribute among available listings, how sellers might respond and which effects disappear when the first motivated buyers have already purchased.
That is more informative than adding a blanket premium to every home inside a polygon. Buyers may care about the boundary differently. Some have no school-age children. Some cannot afford the home even after the boundary helps. Some will choose a smaller property closer to work. The spatial effect depends on people and alternatives, not just a line on a map.
The practical unit is often a choice set. Which homes would the same buyer seriously consider this month? Two addresses separated by a railway line may be farther apart in that sense than homes in different official districts connected by a direct train. A model can build choice sets from observed search and transaction behavior, travel times, budgets and property attributes. It must also admit where search data are missing or biased toward the people whose online behavior is visible.
Competition follows those choice sets too. The closest rival to a home beside a station might be several stops away, not the house on the next block. A new listing along that line could draw away its likely buyers while leaving the family market inside a school zone untouched. One neighborhood inventory figure can therefore blend a tight market and a crowded one. A seller needs to know which inventory her own prospective buyers can see.