In order to create a comprehensive picture of the evolving housing needs in Massachusetts, EOHLC partnered with researchers from the University of Massachusetts Donahue Institute (UMDI) and the Metropolitan Area Planning Council (MAPC.) With input from the Housing Advisory Council, EOHLC and the research team created multiple scenarios of population change and housing needs, based on different assumptions about the future. These scenarios illustrate the range of possible futures for the state and provide insight into the housing needs of the coming decade.
EOHLC chose to work with the research team of UMDI and MAPC since they have an established partnership and track record of developing projections for Massachusetts and its regions. For more than a decade, this same team has been contracted by MassDOT to support the detailed forecasts needed for regional transportation plans developed every four years. In this partnership, UMDI provides regional population projections and MAPC creates household and housing demand forecasts for each population scenario. To create consistency and comparability with regional transportation planning efforts, EOHLC also chose to use the 13 regional planning agency areas as the geography for projections.
UMDI’s population projections use a cohort-component method that applies recently observed survival, fertility, and migration rates by age and sex to Census 2020 base populations to project population by 5-year age groups in 5-year increments to 2050 for all Massachusetts regions. Survival and fertility (deaths and births) are based on recent rates published by the Massachusetts Department of Public Health applied to the forecasted population for each age group. Migration into and out of each region is estimated for each age group at a variety of geographic scales: within Massachusetts, to and from other New England states, to and from the rest of Massachusetts, and international immigration. By changing the assumptions about migration at each geography, UMDI created three distinct population scenarios for the future decades.
Once the population scenarios were created, MAPC formulated household projections for each region. These projections use age- and region-specific rates derived from detailed census data to estimate the number, type, size, and income of households likely to form for each population scenario and forecast year. As part of this effort, MAPC adjusted assumptions about household formation and other attributes to create various alternative scenarios for housing needs. The most significant such assumption applies year 2000 household formation rates to certain age groups in order to estimate latent demand. MAPC also estimated the number of overcrowded and doubled-up families in each region and adjusted the headship rates to simulate each family forming their own household instead of doubling up.
In order to estimate the housing needs associated with the current shelter population, EOHLC examined administrative data from the Emergency Shelter program. This analysis produced estimates of the number, size, age, and nativity of families in the system. This information was then provided as inputs to the population and household models so that the needs of homeless families could be incorporated into the housing demand forecasts and the recent influx of international migrants could be incorporated into certain population scenarios.
MAPC and EOHLC compared current vacancy rates (by tenure) in each region to the target vacancy rates to estimate the current shortage of units for sale or for rent. Each housing scenario also accounts for the additional units needed to maintain a healthy vacancy rate as the housing stock grows. When calculating the vacancy rates, the research team excluded homes that were considered unavailable—those that are vacant for seasonal use, repair, or other reasons. Therefore vacancy rates are calculated as the number of units for sale or for rent, divided by the number of occupied units plus those for sale or for rent.
To estimate the number of homes lost to seasonal conversion between 2010 - 2019, EOHLC estimated how many households in 2019 were living in homes built since 2010, and then subtracted that from the net change in year-round housing units over that same period. For example, Cape Cod added 8,800 units but saw the number of year-round homes decline by 3,400 even as there were 2,400 households living in newly built year-round homes. These figures indicate that 5,800 homes were converted from year-round to seasonal use over that period. Similar patterns are observed, to a lesser extent, in three other regions: Berkshire County, Pioneer Valley, and Franklin County. In all four of these regions, the number of homes converted to seasonal use was converted to a percentage and then applied to the 2020 housing stock to estimate seasonal conversion over the coming decade.
In total, EOHLC and the research team created four different population and housing demand scenarios:
- Crisis as Usual: UMDI’s low scenario population projections combined with baseline headship rates and no changes to vacancy rates
- A Sense of Normalcy: UMDI’s low scenario population projections combined with headship rates that account for latent demand, doubled up households, shelter population, and return to healthy vacancy rate.
- Competing and Growing: UMDI’s middle scenario population projections combined with headship rates that account for latent demand, doubled up households, shelter population, and return to healthy vacancy rate.
- Ready for What Comes: UMDI’s high scenario population projections combined with headship rates that account for latent demand, doubled up households, shelter population, and return to healthy vacancy rate.
More details on the population and housing demand projections can be found in the methodology documents prepared by UMDI and MAPC, which can be downloaded using the links below: