Fiscal Impact Methodology

Fiscal Impact Methodology

Data Sources

­2019 ACS 5-Year PUMS (data/docs) data and metadata were directly sourced from census.gov. Some pre-processing was required to split the Data Dictionary into Housing and Person files (with appropriate headers), and all data was processed directly (i.e. without being imported into a spreadsheet) to preserve data integrity (e.g. retain leading zeroes on State and PUMA codes).

Housing Multipliers

Housing Estimates

  1. Filter housing data for YBL_NUMERIC (ALL) and MV_MONTHS_NUMERIC (<=48)
  2. Compute the sum of the Housing Unit Weight (WGTP) and each of the housing record-replicate weights (WGTP1-WGTP80) for each unique combination of ST, PUMA (excluded in State-level aggregation), class_single_or_multi_family, class_attached_or_detached, class_owner_or_rental, class_unit_bedroom_type
  3. Compute the difference between the Housing Unit Weight (WGTP) and each replicate weight (WGTP1-WGTP80) and square the result for each.
  4. Sum all of the squared differences and compute:
    • Variance (WGTP_VAR) = Sum of Squared Differences * (4/80)
    • Standard Error (WGTP_SE) = Square root of Variance (WGTP_VAR)
    • 90% Confidence Margin of Error (WGTP_ME90) = 1.645 * Standard Error (WGTP_SE)
    • Error Margin as a Percentage (WGTP_EMP) = 90% Confidence Margin of Error / Sum of Housing Unit Weight (WGTP)

Person Estimates

Person estimates and statistics are computed using the same mechanism, however person records are only counted in the different cohorts when the requisite flag for each person record = 1, resulting in a set of outputs as classified below. Each cohort code has a corresponding set of statistics.

Multiplier Computations

Multipliers are then computed by dividing the sum of Person Estimates (PWGTP_[cohort_code]) by the Housing Estimates (WGTP) for each combination of dimensions.

Multiplier Margins of Error

Multiplier Margins of Error and Error Margin Percentage are then computed as per the ACS documentation (U.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data: What Researchers Need to Know, Appendix 3, pp. A-18, U.S. Government Printing Office, Washington, DC, 2009.

The Margin of Error is computed using the following equation:

C10 ‚Äď Ratio derives a ratio from two estimates and its measures of precision with the numerator not being part of the denominator. The MOE for a ratio is calculated as:

where:

  • MOEnum of the MOE of the numerator.
  • MOEden is the MOE of the denominator.
  • R = Xnum/Xden is the derived ratio.
  • Xnum is the estimate used as the numerator of the derived ratio.
  • Xden is the estimate used as the denominator of the derived ratio.

In our case this is the equivalent for each cohort of:

  • SQRT(person_me90^2¬†¬†+¬†((person_weight/housing_weight)^2¬†*¬†housing_me90^2))¬†/¬†housing_weight

Assumptions

Years

  • Year Built last: Converted to numeric YBP numeric based on lower year of range (where range given). 1939 or earlier mapped to 1900.
  • Mover (When moved into this house or apartment): Numeric value assumed to upper boundary of each band.

Housing Classifications

  • Single/Multifamily: Base variable: BLD_output_value (Units in Structure)
  • Attached/Detached: Base variable: BLD_output_value (Units in Structure)
  • Tenure: Base variable: TEN_output_value (Tenure)
  • Units: Base variable: BDSP_output_value (Number of bedrooms)

Person Classifications

  • Students by Grade: Base variabl: school_aged_flag by Age (Between 5 and 17 years)