***** #9109985 Sidney Wong, Ph.D., Community Data Analytics Daniel Miles, Ph.D., Econsult Solutions, Inc. National Planning Conference May 7, 2017 New York City
Introduction Changes since 2000 Local Variations Improvements to Multipliers Q & A and Discussion
Demographic Multipliers Definition, Examples, History, Theory, Evolution
What is a Demographic Multiplier? An average ratio of various populations or demographic measures per household Residential - households or occupied housing units Nonresidential industrial, retail and offices development Cohort differentiation age or grade groups
Common Examples of Number of school-age children (SAC) Average household size (AVHH) Public school attendees Number of occupants or SAC at specific age cohorts Number of workers per 1,000 sq. ft. of gross floor area
Origins of Formalized in 1978 in The Fiscal Impact Handbook, Ch. 13-15 Records from 1970 Census Public Use Sample (later the PUMS) Recently built housing units 9 regional subdivisions covering the US Source: http://www.transactionpub.com/title/the- Complete-Illustrated-Book-of-Development- Definitions-978-1-4128-5504-4.html
Traditional Before 2006 1978 The Fiscal Impact Handbook 1985 The New Practitioner s Guide to Fiscal Impact Analysis 1994 Development Impact: Assessment Handbook 2006 derived from 2000 Census Public Use Microdata Sample (PUMS) records Fannie Mae Series 50 States & DC AVHH, SAC, Public School-Age Children Who Lives in NJ Housing? 3 sub-state regions & NJ AVHH, SAC, Public School Children More elaborate
PUBLIC USE MICRODATA SAMPLE American Community Survey (ACS) raw dataset Un-tabulated records about Individuals households, and housing units 1-year, 3-year, and 5-year samples 5 percent for 5-year Most recent 5-Year release: 2011-2015 on January 19, 2017
Household Characteristics Multipliers are specific to the sample drawn from PUMS as defined by characteristics of households or by the type of development they reside: Households living in recently built units Subsample for those living in specialized housing Age-Restricted Community Affordable Housing Transit Oriented Development
Housing Configurations A housing typology defined by - Size (number of bedroom) - Tenure - Structure type - Value groups Fiscal Impact Handbook, derived from 1970 Public Use Sample Bedroom Studio Single-Family Homes Garden Apartment High-Rise Apartments Town Houses Mobile Homes 1 2 3 Blended 19 configurations
Housing Configurations Fannie Mae Series, derived from 2000 PUMS Bedroom Single-Family Detached Single-Family Attached 5+ Units Own 5+ Units Rent 1 2-4 Unit Mobile 2 3 4 5 Each is subdivided into four housing value groups. 76 configurations, only large states have full coverage.
Factors Affecting Demographic Size Tenure Multipliers 0-1 Bedroom rental less intensive 2+ Bedroom rental more intensive Structure type Single-family detached dwellings are more population intensive Townhomes, after controlling for size, recently become more intensive then detached units. Value or rent level higher value less intensive Types of development Local geography reflecting its housing and labor market
Types, Forecast, Multipliers
Types of School Fiscal Economic Traffic Environment Social Political Cumulative Others
Is A Development Good or Bad for my Community? It depends The types and magnitude of impacts according to the attributes of the proposed development. The constraints: capacity and community preferences. Development generates a host of new costs and benefits. It is important to assess ahead of time if the benefits outweigh the costs.
Applications of Impact Studies Planning Urban design Land use policies Rezoning & subdivision Annexations/Acquisitions Redevelopment Comprehensive planning Infrastructure planning Forecast of added population & enrollment Estimation of pollution and traffic impacts Budget & Finance Annual budget preparation Formulation of development charges Revenue & expenditures forecasting Capital Improvement Programming Inputs to bond financing Fiscal planning Municipal staffing Level of service changes
Critical Information to Estimate Impacts and Fiscal Costs Occupants Age School-Age Children Public School Attendees Household Income Number of Cars Available Year of Moving In Other Information Housing Units Structure Types Number of Bedrooms Rental o r Owned Year Structure Built Other information School Traffic
The Role of Multipliers in Impact Studies Demographics have the largest impact and greatest influence on fiscal impacts. The most important demographics include: Household size Number of school children Income levels can shed light on who will be calling a project or development home.
Example Total Residents Unit Type Number of Bedrooms Number of Units Residents / Housing Unit Total Residents Single Family Age Targeted 3 51 1.12 57 Single Family 3 18 2.24 40 Single Family 4 100 3.61 361 Townhouse - Owner Occupied 3 112 2.24 251 Townhouse - Rental 3 23 2.90 67 Mixed Use Condominium 3 46 2.24 103 Apartments 1 135 1.24 167 Apartments 2 135 1.86 251 Total Housing Units 620 1,298
Example School Children Unit Type Number of Bedrooms Number of Units Public School Students/Housing Unit Total Public School Students Single Family Age Targeted 3 51 0.15 7 Single Family 3 18 0.29 5 Single Family 4 100 0.95 95 Townhouse - Owner Occupied 3 112 0.29 32 Townhouse - Rental 3 23 0.64 15 Mixed Use Condominium 3 46 0.29 13 Apartments 1 135 0.07 9 Apartments 2 135 0.33 45 Total Housing Units 620 222
Changes Since 2000 Demographics, Economy, Housing Market, Preferences
Changes Since 2000 Source: http://www.beingthereseniorcare.com/wpcontent/uploads/2014/10/babyboomerexpo.jpg https://securitytoday.com/articles/2015/04/01/~/media/sec/security%20product s/images/2015/09/changing_role_of_security_high_end_residential.png http://www.slate.com/content/dam/slate/articles/double_x/doublex/2012/12/1 21212_DX_OldParentsBaby.jpg.CROP.rectangle3-large.jpg http://www.foreclosure-support.com/images/foreclosure-house-up-for-sale.jpg http://20some.20some.netdna-cdn.com/wpcontent/uploads/2015/07/heroimage_1439x960_teamwork-e1436796970958.jpg https://static.pexels.com/photos/2324/skyline-buildings-new-york-skyscrapersmedium.jpg http://cache4.assetcache.net/xd/500830364.jpg?v=1&c=iwsasset&k=2&d=df8d445051b40c742b43 7D38280FE185057966828B4ABE84D21EEF1769CC03D78196B548624031B3 http://2020iscoming.info/else/moving-truck-pictures.coming
Changes the 2000 PUMS Misses Baby Boomers Age-restricted communities Post-housing crash reconfiguration Appeal of city life to Millennials, but emerging evidence of possible reversal Recent slow down of twenty-somethings entering the labor market Affordability problem Multifamily development Households choosing to delay parenthood
Average Household Size Declines in Most States Only 5 states experienced an increase: Texas, Delaware, Florida, California, Nevada 2.8 2.7 11 States with Largest Absolute Decline 2000 2010 All Occupied Units 2.6 2.5 2.4 2.3 2.2 US US ME ND VT MT SD WI NH MI LA NM AK Sources: Table H12, 2000 and 2010 Census SF1
Declining SAC in Selected States 0.6 2000 2015 All Occupied Units c 0.5 0.4 0.3 AZ CO MD NJ OH PA VA WA 2000 & 2015 Newly Moved in Household Sample Sources: Community Data Analytics, 2017
SAC Changes, 2000-2015 2-Bedroom Multifamily Units, Owned and Rented Decline in 39 states. Median change: -7% decline. Wide range of changes. Maximum decline: -8% (District of Columbia). Maximum increase: +22% (South Dakota). 2000 & 2015 Newly Moved-In Household Sample Source: Community Data Analytics, 2017
SAC Changes in Pennsylvania 1.2 1.0 0.8 0.6 0.4 0.2 2000 2014 Various Housing Types 0.0 Single-Family ffsfd ffsfd Detached ffsfd TownTownhome Town 2-4 Multifamily 2-4 2-4 2-4 5+ Rent MF 5+ 5+ Rent Rented 5+ Rent 2-B 3-B 4-B 2-B 3-B 4-B 1-B 2-B 3-B 1-B 2-B 3-B 2000 & 2014 Recently Built Unit Sample Sources: Community Data Analytics, 2016 & Fannie Mae Foundation, 2006
School Impact of 3-Bedroom Units The most popular housing choice Many Different 3-Bedroom Units Single-family detached Duplex, quadplex Townhomes Condos Mid-rise and high-rise units Row houses Rental apartments Age-restricted units (55+) Affordable units Did SAC for 3-Bedroom units decline like the average household size in the past decade? Yes, as shown in Pennsylvania; but.
SAC Changes in Maryland 1.2 1.0 2000 2014 3-Bedroom Units 0.8 0.6 0.4 0.2 SF Detached Townhome MF 2-4 MF 5+ All Units All Owned All Rented 2000 & 2014 Newly Moved in Household Sample Sources: Community Data Analytics, 2016
SAC Changes in New Jersey 1.1 0.9 2000 2014 3-Bedroom Units 0.7 0.5 0.3 0.1 SF Detach Townhome MF2-4 MF5+ MF5+ Owned MF5+ Rented All Owned All Rented 2000 & 2014 Recently Built Unit Sample Sources: Community Data Analytics, 2016 & Fannie Mae Foundation, 2006
SAC Changes, 2000-2015 3-Bedroom Single-Family Units, Attached and Detached Decline in 49 states Median 2000-2015 change: -9% decline Maximum decline: -38% (District of Columbia) Maximum increase: +2% (Nebraska) 2000 & 2015 Newly Moved-In Household Sample Source: Community Data Analytics, 2017
Illustration of Estimation Biases in 2000 Multipliers Compare 2015 and 2000 School-Age Children Ratios For every 1,000 3-Bedroom units in New Jersey 3-Bedroom Units Overestimation Underestimation All Owner-Occupied 156 All Renter-Occupied 54 Single-Family Detached 64 Townhome 27 Multifamily 2-4 Unit Structure 35 Multifamily 5+ Unit Structure 152 2000 & 2015 Newly Moved-In Household Sample Source: Community Data Analytics, 2017
Recent Townhome Crowding In the 1990s, at the 2-Bedroom level, the SAC ratio for single family detached units generally exceeded that of townhomes. 2015 School-Age Children for 2-Bedroom Units SF Detached Townhome 0.3 0.2 0.1 AZ CO MD NJ OH PA VA WA 2015 Newly Moved-In Household Sample Source: Community Data Analytics, 2017
Multipliers Need to be Updated Significant changes between 2000 and 2015 Widespread decline in multipliers Large variations among states Decline or increase specific to housing configurations Recent crowding of townhomes The World has changed since 2000 and so should demographic multipliers.
Variations across Geographies Biases of State-level Multipliers http://cmsny.org/wp-content/uploads/wongfigure1.png
Local Variations Matter How can the same ratio be applied to Albany, Buffalo or New York City? Should the multipliers for Dallas-Fort Worth differ from those in San Antonio? What is the smallest geographically specific data from PUMS?
PUMA Public Use Microdata Area A geographic unit of about 100,000 to 190,000 people The smallest geography for releasing PUMS records Redrawn about a year after each decennial census
Large Variations in Ohio SAC 3-Bedroom Single-Family Units, Owned and Rented Distribution of 93 PUMAs by SAC Value Groups 18 Ohio SAC = 0.625 15 12 9 6 3 0 <0.40 0.40-0.449 0.45-0.499 0.50-0.549 0.55-0.599 0.60-0.649 0.65-0.699 0.70-0.749 0.75-0.799 0.80-0.849 >0.85 2015 Newly Moved-In Household Sample Source: Community Data Analytics, 2017
Large Variations in Ohio SAC 2-Bedroom Multifamily Units, Owned and Rented 93 PUMAs in OH OH SAC: 0.276 Highest SAC: 0.556 Lowest SAC: 0.056 Wide range of variations Variance ranges from -80% to +102% Median Variance: -6% 2015 Newly Moved-In Household Sample Source: Community Data Analytics, 2017-79.66% - -41.9% -41.89% - -18.93% -79.66% - -41.9% -18.92% - 5.07% -41.89% - -18.93% -18.92% - 5.07% 5.08% - 36.62% 5.08% - 36.62% 36.63% - 101.57% 36.63% - 101.57%
SAC Variations in Other States 2-Bedroom Multifamily Units, Owned and Rented District of Columbia and Maryland, 2014 0.5 0.0 101 102 103 104 105 1001 1002 1003 1004 1005 1006 1007 1101 1102 1103 1104 1105 1106 1107 MD District of Columbia Montgomery County Prince George s County 0.5 New Jersey, 2015 0.0 101 304 306 601 701 901 906 1003 1103 1205 1302 1404 1700 2002 2301 2303 2500 NJ Newly Moved-In Household Sample Source: Community Data Analytics, 2016 & 2017
SAC Variation in Other States 2-Bedroom Multifamily Units, Owned and Rented Pennsylvania, 2014 0.5 NW North-Central Pittsburgh Region Allentown Philadelphia & SE Region Central 0.0 102 701 702 900 1701 1807 2002 2803 2901 3002 3003 3103 3209 3401 3501 3502 PA Newly Moved-In Household Sample Source: Community Data Analytics, 2016 Location matters - inner city, urban area, suburb, exurb, rural area, etc. each has its distinctive housing market. State multipliers cannot reflect local conditions but state average.
Improving Demographic Multipliers Up-to-date, geographically specific, wider scope, more reliable estimates
Better Multipliers Multipliers play a crucial role in the estimation of development impacts. Reflecting Recent Trends Fannie Mae multipliers have not been updated. Frequent Updating ACS-PUMS is now released annually. Specific to Local Geography State multipliers cannot reflect local conditions. Clearer Definition and Wider Scope
Local Multipliers Insufficient Sample Size Housing construction governs the number of observations in recently built unit samples adopted by traditional multipliers. Highly differentiated housing configurations usually lack sufficient sample size. Even small states face this problem for less common housing types.
Finding an Alternative Sample for Local Multipliers A larger sample increases the chance for reliable local multipliers. The alternative sample should cover likely occupants of the subject development. Movers A sample of households newly moved into an old or recently built unit is a reasonable alternative. Such samples consist of the recently built unit sample.
Newly Moved-in Household Sample 3 to 7 times larger than the recently built unit sample. Allows reliable multipliers for broad-based blended housing configurations at the PUMA level. Also provides reliable multipliers for most common housing configurations at aggregate PUMA levels. Less affected by housing activities. Reflects long term effects.
Comparing 2 Samples 5 4 3 2 1 New Jersey Pearson R = 0.988 N = 72 AVHH Scatterplot, 0 to 4 Bedroom Units Recently Built Units Sample 1 1 2 3 4 5 1 2 3 4 5 Newly Moved-In Household Sample 5 4 3 2 Pennsylvania Pearson R = 0.973 N = 72 Source: Community Data Analytics, 2017 based on 2010-2014 5-Year ACS-PUMS
More Multipliers PUMS records reveal many household attributes that can be used to develop more multipliers: Number of workers per household Number of workers who drive to work Number of cars accessible Number of commuters who use public transit Number of non-public school students Average household income Per capita income Number of occupants on welfare
Specialized Samples PUMS records can identify such samples: Households living in condominiums Households who have retirees Low and moderate Income households Households with transit commuters Households who are double income without children Millennial households Other custom household or development types
Clearer Concept and Refinement of Definitions New generation multipliers would be more precise: The definition of each type of samples and multipliers The cut-off of sufficient sample size The relationship between SAC and School Attendees Distinction between grade groups and age cohorts
Handouts and Poster on
Previous Work of Community Data Analytics What is a Demographic Multiplier and Why Does it Matter?
Upcoming Community Data Analytics Research CDA will give a technical presentation on multiplier methodology in the upcoming 2017 ACS Data Users Conference in Alexandria, VA. A more detailed methodological discussion, Residential Demographic Multipliers: Using PUMS Records to Estimate Housing, will appear in HUD Office of PD&R s Cityscape later this year. If you are interested, please leave your contact information.
SESSION CONCLUSIONS Use current data instead of 2000-PUMSderived multipliers. Use locally specific multipliers, not state multipliers, to avoid state average effects.
Q & A Discussion Community Data Analytics 1435 Walnut Street, 4 th Floor Philadelphia, PA 19102 215.717.2777 Sidney Wong, Ph.D. wong@econsultsolutions.com cda-esi.com econsultsolutions.com
Suggestions in Using Multipliers Focus on how to best estimate different aspects of development consequences. Decompose the development by housing configurations and other attributes Statewide Public School Children ratio is biased because variations in public school participation rates. Sample size problem intensifies Geographical specificity Highly differentiated housing configurations Multipliers by cohorts are less reliable.
Steps of Fiscal Impact Analysis