Illuminating the Real Estate Landscape Expert & Dedicated Minds Research and Forecasting Philosophy The Economy and Demographic Trends. Since we believe that it is essential to understand regional economic and demographic conditions in order to properly forecast the real estate markets, we devote a good deal of energy to our economic and demographic model. We have developed a true MSA-level model to drive our forecasts of local economic conditions. The model takes national drivers as inputs, including GDP, employment growth, various price indices and interest rates. Our model then estimates econometrically the relationships between the MSA variables, the national variables, and past values at the MSA level. Employment is a key driver in this process. Births and deaths at the MSA level are modeled using birth and death rates from local statistical sources. These rates move slowly, so we have a high degree of confidence in our projected rates, which trend with economic and age structure conditions. This process gives us the natural increase in the population. We then model net migration as related (econometrically) to economic variables (employment growth and unemployment rate). This equation works particularly well. The sum of the natural increase and net migration gives us total population growth. The population projections are done by age cohort, and the age cohort household forecasts are made by using headship rates applied to population. Like natural population increase, headship rates move slowly over time, so they are reasonably well projected into the future. Employment growth is linked to the U.S. employment growth rate and past values of the growth rate at the MSA level. In addition, historic variations in the growth of particular metropolitan areas with respect to the national growth rate are captured by using intercept shifts. The unemployment rate is modeled simultaneously with the population variables. The unemployment rate is reported from data derived from the household survey. On the other hand, the payroll employment data are from the survey of establishments. It is not a simple matter to mechanically produce an unemployment rate from the payroll employment data and the labor force (demographic data). As a result, we model the unemployment rate based on payroll employment growth and the population growth rate (as a proxy for labor force growth). The unemployment rate needs to be simultaneously determined with the population growth rate because employment growth and the unemployment rate enter the migration equation, and migration (through the population growth rate) enters the unemployment rate equation. This model provides important outputs for each MSA including employment by major sector, unemployment, personal income, consumer price index, population by age cohort, and single family and multifamily housing market forecasts. Generally, the model works well in "normal" times when there is no structural change underway. We intensely review the output from the model, and make adjustments to account for factors that we know are happening which the model cannot readily take into account. A major restructuring at the industry level, the electric utility industry, for example, sometimes has had to be accounted for in an add-factor process for particular employment subsectors. We assess the top down national and regional forecasts against the forecasting results for the 77 metropolitan areas as a crosscheck. And of course we regularly perform back-tests to see how accurate our forecasts are.