Each quarter, Veros Real Estate Solutions releases a new VeroFORECAST that predicts what average property values in hundreds of Metropolitan Statistical Areas across the nation will do over the following year. We send the report to subscribing individuals and businesses as well as, in an abbreviated overview, to the media.
Next month, using the most recent data generated through first quarter 2019, we will release appreciation predictions for changes in average property values in 349 metro areas (MSAs), 984 counties and 13,545 ZIP codes through March 31, 2020.
Forecasting changes in the housing market may sound like crystal ball-gazing, but, in fact, there is both a science – and an art – behind making reasonable predictions of real estate appreciation or depreciation. For well over a decade, the VeroFORECAST has been able to analyze various key indicators that will impact values with enough confidence to give lenders and other financial service providers, a snapshot of the market at four important future time horizons: six, 12, 18 and 24 months.
Veros, for example, has designed a system it believes will be the most meaningful for its clients and other lending-industry stakeholders, by generating valuation forecast models for the metropolitan, county, and zip code level. The company further stratifies results within those designations by providing separate valuation predictions for single-family residences and condominium and townhomes. Then, the results provide an additional layer of detail by breaking those property types into three price tiers at the county and zip code level: below the 25th percentile, the 25th through 75th percentiles, and above the 75th percentile.
HOW MODELS ARE DEVELOPED
The complexity of a process that seeks to provide accurate forecasting involves a large collection of variables. Forecasting for some market models is relatively straightforward, whereas other markets require more intricacy. Types of predictor variables considered in forecast models include unemployment rates, interest rates, inflation, population trends, seasonality, market inventory, affordability index, and the percentage of mortgages originated with high LTVs. Each model also recognizes that an individual predictor variable may not necessarily be the best variable to include in a model. Therefore, a model may also consider rates of change of variables from a previous time period or lags of variables from previous time periods... Read more on LinkedIn or download the PDF