30.04.2024
Hello my name is Andy Leventis and I'm. Supervisory Economist with the Federal. Housing Finance Agency. I'm in charge of. the production of FHFA's House Price. Index, or HPI. I'd like to show you a. little bit about how we construct the. HPI. We begin by assembling house price. data from a variety of sources, primarily. Fannie Mae and Freddie Mac. Each month. these Enterprises submit to us. loanlevel data for mortgages they have. bought or guaranteed since the late. 1970's. The mortgage data,. which goes into our database, includes. home values, that is sales prices and. appraisals for loans used to purchase. properties or refinance mortgages. For. some of our indexes, we use an expanded. set of information. In these cases, we. also incorporate sales prices reported. to County Recorder offices, sales prices. and appraisal values from the Federal.
Housing Administration, or FHA, and in one. case, we also use information from. mortgages that serve as collateral for. advances from the Federal Home Loan Bank. of New York. Once we assemble the data. from our various sources, the critical. next step is to search through the. information and find homes for which we. have two or more historical values.. We are frequently able to find homes. with two,three, or more historical. transactions. In the case of our purchase. only indexes, we are looking for. properties which have sold multiple. times. In other situations where our. indexes incorporate appraisal data, we. simply need to see multiple values for. the same property, whether they be sales. prices or appraisal values. Once we find. properties with multiple transactions we. assemble transaction pairs. These show.
The price change that was observed over. a specific interval for the same. property. For instance, if in our database. we see a property that sold three times . in this case we'll call it 123 Elm. Street Anytown, USA we would construct. two pairs. The first pair would show the. price change between the first and. second transactions so the first sale. was for a $100,000, the. second sale was for $250,000, which meant. that the price change was. +$150,000. The first time period was first. quarter of 2000, the second sale. was in the fourth quarter of 2006.. Then, the second pairing. would show the price change between the. second and third transaction. In this. case the first sale was for $250,000 the. second sale was for $200,000. which meant that there was a. price decline of $50,000. The first of. these two transactions was the fourth.
Quarter of 2006 and the second was in. second quarter of 2010. So, using all the. historical data at our disposal, we. construct millions of transaction pairs. just like these. Then, the final stage of. the process is to take this database of. pairs,. now shown with these red diamonds, and. apply a statistical technique known as. regression to approximate the market. trend. The specific regression model. known as the repeat transaction approach,. produces a set of historical measures, a. House Price Index that broadly mimics. the price changes that are observed in. all the pairs as a whole. The measure of. course will not exactly replicate the. price changes for every individual home,. however the index is calibrated to. largely reflect the overall market. trends. I hope that you've found this short. video to be valuable. If you're.
