CFPB Implementation of Parcels Provision in HMDA Under Dodd-Frank
Land ownership is the foundation of the financial, legal, and real estate systems in our society
Parcel Data vs. Census Data + Census data is inaccurate at the parcel level + Census level data is updated every ten years + Parcels change every day + Foreclosures happen per parcel + Parcels provide the true gauge of anything happening at a property level Decisions and policies should be made using the proper granularity of data
Parcel Data Street address Owner information Sales price Appraised value Land value Improvements value Lot size Land Use Deeded acreage Property type Mortgage information Etc.
Authoritative Source Local governments process real estate information on a daily basis from recorded documents, assessment data, and parcel maps + Most current information + Most accurate information + Baseline of value + Best indication of economic activity + Common denominator of land + It is the public record
State of Maryland + 2.3 million parcels + Statewide Assessments and Taxation - State Department of Assessment & Taxation - assessments and billing - Maryland Department of Planning - map maintenance + Statewide parcel points (MDP) + Statewide raster versions of tax maps + MD Property View (from MDP) - Property, sales, CAMA, data extracts, and other layers - Licensed products (multiple tools and data products) - Funding supports maintenance + Statewide Web Map Service (MDiMAP) - Parcel boundaries w/link to SDAT datasheet + County vectorized parcel boundary datasets complete coverage - Variable content, design, timeliness, distribution policies + State and Counties working on Integrated Parcel Update Process
Using Parcels to Predict 2005 High Cost Loans + Analysis Set up national models and perform what if scenarios + Identify Easily identify what areas are likely to be affected next + Act Predictions lead to early action which leads to prevention + Visualization Like the weather, if you monitor it, you can learn to predict it 2006 Foreclosures
Using Parcels to Prevent If you are tracking and monitoring specific items and events you will have an early warning system to respond to clusters before they become national epidemics Possible Red Flags: + Changes in National Trend Lines + Home Values vs. GNP + Overlays with additional datasets + Mortgage resets, defaults, etc.
Question #1 Sufficiency or insufficiency of address. What is the value of parcel ID? + Address data alone is insufficient + Lack of standards + Many county assessors maintain owner address rather than situs address + Parcel ID is the best unique identifier for a parcel + Parcel ID is used to link to additional real property databases + The FIPS code could be used to create nationwide Parcel ID
Question #2 What is the burden of collecting parcel ID or address from originators? + Does not burden the originators because the Parcel ID is ALREADY included in the documentation used to process a mortgage + Just needs to be added to the HMDA Report to be useful
Question #3 What are the existing standards? + There is a lack of nationwide standards + Approximately 20% - 25% of parcel are still not digital (mostly rural) + A database design and crosswalk would be developed to create a standard + Non-digital parcel would need to be converted to standard format
Question #4 What are the existing databases? + A lot of various databases exist at a county level (i.e. ESRI, CAD, Access, AS400,etc..) + A standard data model and crosswalk takes care of the issue + There are several commercial national databases that cover most of the digital parcels
Question #5 How do different approaches deal with one to many and many to one relationships? + Non-graphic databases already account for these situations with each parcel and condo having a unique identifier + Graphic data can be handled different ways such as stacked parcels
Question #6 How do different approaches deal with latency issues? + Latency is certainly an issue when dealing with a dataset that changes daily + This is also an issue when dealing with commercial national datasets as the update cycle can be 2-3 years for many counties + A maintenance and update plan would need to be identified and the feasibility of that plan assessed
Question #7 What are the privacy concerns and implications? + Real Property data maintained by local governments is already public record and in many cases freely available online + The primary privacy issue is linking the parcel ID to loan information and credit score. + This could be addressed in a variety of ways such as the use of additional codes in the database
Next Steps + CFPB join FGDC + Hire a land records consultant + Identify states/counties for pilot projects + Review existing datasets and perform a gap analysis and cost analysis + Draft database design for a national dataset with appropriate attributes defined + Identify tasks and costs of data conversion for the remaining 25%