Freddie Mac Valuation Update

Transcription

1 Freddie Mac Valuation Update CoreLogic Mortgage Fraud & Valuation Consortium November 3, 2016

2 Freddie Mac s Twin Goals A Better Freddie Mac and a better housing finance system For families, customers and taxpayers by...innovating to improve the liquidity, stability and affordability of mortgage markets...competing to earn their business...reducing their exposure to mortgage risks Better Business Together every day in every way Deliver better solutions to make more loans, lower costs and drive servicing excellence Provide better support to strengthen business opportunities Look for better ways to provide greater certainty Develop a better infrastructure for the future Freddie Mac 2

3 Better Solutions: The Loan Advisor Suite Helping our customers succeed through an innovative suite of solutions designed to deliver certainty, efficiency, reliability, and usability Certainty Give you confidence that the loans you originate meet the requirements for delivery and sale to Freddie Mac Efficiency Lower the cost to originate via automated data validation and focus your attention where it needs to be Reliability Instill trust that our tools are available when you need them Usability Provide intuitive and easy-to-use tools, with clear and actionable feedback Freddie Mac 3

4 Loan Advisor Suite Component Overview Application Processing / Underwriting Pre-Closing Closing Loan Delivery Servicing Loan Advisor Suite Loan Product Advisor Loan Collateral Advisor Loan Quality Advisor Loan Closing Advisor Selling System Loan Coverage Advisor Business Intelligence The Loan Advisor Suite builds the manufacturing quality story required for greater purchase certainty. Freddie Mac 4

5 Loan Advisor Suite Interacting with Our Products Application Processing / Underwriting Closing Loan Delivery LOS Loan Officer Processor / UW Closer Post Close / Delivery DATA LP / LPA CR DATA LP / LPA INCOME DATA ASSET DATA UAD UCD ULDD ULDD LAS Loan Product Advisor Loan Collateral Advisor Loan Quality Advisor Loan Closing Advisor Feedback Purchase Eligibility Risk Assessment Underwriting Guidance Income Verification Asset Verification Appraisal Quality Risk Valuation Risk Representation & Warranty Relief Eligibility Spanning 3 Cs LPA Data Compare Risk Analysis Purchase Eligibility UCD Completeness & Integrity Closing Accuracy Freddie Mac 5

6 Loan Collateral Advisor What is Loan Collateral Advisor SM? Analyzes appraisal reports submitted to the UCDP Provides Freddie Mac s view of appraisal quality and valuation risk Provides real-time feedback Customer Benefits: Increases operational efficiencies and improved workflow Reduces risk of repurchases related to significantly deficient or inaccurate appraisals Loan Collateral Advisor results can be flexibly implemented within Sellers current processes! Freddie Mac 6

7 Loan Collateral Advisor Freddie Mac has worked with development partners to ensure ease of use and efficiency gains. Freddie Mac 7

8 Loan Collateral Advisor What do the risk scores indicate? Valuation Score» Assesses the accuracy of the appraised value relative to Home Value Explorer, our automated valuation model» Considers both overvaluation and undervaluation risk Appraisal Quality Score» Assesses multiple components of the appraisal quality, data accuracy and completeness» Validates that appraisals adhere to our standards and guidelines Sellers can use the Valuation Score and the Appraisal Quality Score as factors in determining risk tolerances. Freddie Mac 8

9 Loan Collateral Advisor LCA Findings Messages assist with analysis of the appraisal report and in managing workflow processes. High risk valuation score indicates a significant misalignment between the appraisal value and the HVE value. The review required may be different if the message indicates over valuation as compared to when the message indicated under valuation. High risk Appraisal Quality Score may indicate possible data inaccuracies. The review required may be directed to these elements to determine whether an update or correction to the appraisal is needed. Sales information may help determine the relevance of the comparable sales selected to the pool of available sales. Sellers are leveraging scores and messages to direct workflow and allocate resources! Freddie Mac 9

10 A Case Study The Current Appraisal Process Freddie Mac 10

11 Rethinking Valuation Practices Our challenge then is to reconsider the appraisal the process of developing the opinion of value and the content of the appraisal report to serve the future underwriting, pricing, and management of mortgage credit risk. If credit investors can obtain the same or equivalent information from other sources we live in a big data world today, after all then what is the value proposition of an appraisal? From: Rethinking Appraisals in a Modernized Housing Finance System, Prepared for Delivery to the Collateral Risk Network s Third Quarter Meeting, Baltimore, MD, June 29, 2015 by Edward J. DeMarco, Senior Fellow in Residence Milken Institute Center for Financial Markets. Freddie Mac 11

12 Investors: Price or Value? Investors want to know whether the collateral will cover the loan amount at some point in time in the future should the borrower default. Even in a competitive market, I might not want to underwrite the loan using the current price, unless the market is also in equilibrium. Current appraisal practices do not provide Investors with all the information needed to make informed decisions. New Analytics: Fitch s model and Market Condition Indicators. Freddie Mac 12

13 Researching the Current Appraisal Process The current appraisal process performs four critical functions: Inspect the Property 2 nd base Protect Against Fraud 3 rd base 1st base Estimate Value Home Loan! Mitigate Repurchase Risk Freddie Mac 13

14 A Case Study Thinking, Fast and Slow Freddie Mac 14

15 Appraisers Suffer from Anchoring Bias Group A: Chesterfield MO is a nice suburb of St. Louis. Group B: Chesterfield MO is a nice suburb of St. Louis. Do you think the median sale price in Chesterfield is more or less than $900,000? Group B s estimate was ~ 40 to 52% higher! Freddie Mac 15

16 Weird Density of Estimates of Value Loan Application Data Source: CoreLogic LoanApp data, purchase only, originated in or after 2008, origination LTV between 10 and 100, and origination credit score between 620 and 850. Notes: 1. Percentage_diff=0 if abs(appraised value-contract price)<= $100. Only -0.15<=percentage_diff<=0.15 are shown. 2. Other filters are applied to ensure data robustness Freddie Mac 16

17 Failed Metrics Calibrator Period: Jan 1, 2015 April 15, Total Number of Appraisers: 56,653 HVE Calibrator model output ranges from If score <= 600, appraisal is at risk of being overvalued by at least 10% and is referred to as appraisal values that failed HVE Calibrator. % of Loans that failed AVM check # of properties appraised by a given appraiser Freddie Mac 17

18 How Good are Appraisers at Making their Adjustments? Study using Appraisals for Relocation Loans What features get adjusted? what-adjustments-have-the-most-influence-on-appraisal-reports.aspx#.vyejoe3vzx4 Freddie Mac 18

19 Thought Experiment 103 Main Street Sale Date: Yesterday Sale Price: $200,000 Attributes: 2,000 Sq. Ft. ¼ acre lot 3 bedrooms Using adjustments recommended by a model, the adjustments come to $25,000. What to do now? Freddie Mac 19

20 Thought Experiment (continued) Freddie Mac 20

21 How Can Big Data and Advanced Analytics Help? Text mining can help with both valuation and condition. Data mining tools can help protect against fraud. Market Neighborhood Property Freddie Mac 21

22 Data on the Uniform Collateral Data Portal (UCDP) Source: Freddie Mac C-Ratings: describe the property condition, maintenance and/or improvements C1 = new or no physical depreciation C6 = substantial damage / neglect Freddie Mac 22

23 Knowledge Graphs Knowledge Graph is a knowledge base that connects millions of pieces information to enable smarter search capabilities. Google Knowledge Graph was introduced in 2012 using many information sources such as Search Clicks, Linked Pages, etc. to extrapolate the edges (relations) between the nodes (entities). Facebook has the Social Graph whose nodes and edges are constructed from the social network s millions of users and their data. Freddie Mac 23

24 Application of Knowledge Graphs to Mortgages! We can use Graph clustering algorithms to map out neighborhoods programmatically and determine the right neighborhood boundaries. Freddie Mac 24

25 Home Knowledge Graph Imagine estimating the value of a subject home (yellow outline) with three comparable homes. We can use the Home Knowledge Graph to find homes that are similar to the three comps and increase the number of properties used to model the subject home s value, decreasing the noise in the system. Freddie Mac 25

26 Application of Knowledge Graphs to Appraisals Map out neighborhoods and determine the right neighborhood boundaries Increase the number of properties used to estimate the subject home s value Similar Neighborhood Estimation Freddie Mac 26

27 A Cautionary Tale The math-powered applications powering the data economy were based on choices made by fallible human beings.many of these models encoded human prejudices, misunderstanding, and bias into the software systems that increasingly managed our lives * Examples: - Models used to evaluate teachers - Model that ranks universities and departments - For-profit universities - Parole models * From: Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy, by Cathy O Neil, Crown Publishing, New York, N.Y., (2016). Freddie Mac 27

28 The Road Ahead Machine Learning Deep Learning Emulation Entity Model Resolution Fair Lending FCRA Regulatory Guidance How to Leverage in the Context of a Regulated Financial Institution? Recent paper by Google on Lattice Regression Freddie Mac 28