1) FHFA Brief Housing & Mortgage Policy Topics 14-02 July 31, 2014 FIRST-TIME HOMEBUYER SHARE AND HOUSE PRICE GROWTH Saty Patrabansh, Andrew Leventis, and Nayantara Hensel 1 Introduction In 2013, the Federal Housing Finance Agency (FHFA) published Mortgage Market Note 13-01, A Study of First-Time Homebuyers, which provided estimates of first-time homebuyer shares for the last 20 years. Although the Note provided a detailed analysis of credit and borrower characteristics for first-time homebuyers, it did not study variations in first-time homebuyer shares for different geographic areas. While first-time homebuyer shares are undoubtedly affected by many factors, understanding geographic variation meaningfully expands the current body of knowledge on how first-time homebuyers are affected by economic and housing market conditions. This Brief provides statistics from 1996 to 2013 using a large mortgage dataset assembled by FHFA and shows that first-time homebuyer shares differ a great deal across states. Also, the Brief discusses the relationship between trends in the share of first-time homebuyers and trends in house prices in various geographic areas. This exploratory analysis suggests that the first-time homebuyer share decreases as house price growth increases (or it increases when house price growth decreases). In other words, the first-time homebuyer share is negatively correlated with the change in house price growth. First-Time Homebuyer Share by State and Year In general, estimates of the share of first-time homebuyers in each state are not readily available. Statistics derived from the American Housing Survey (AHS) 2 and information from the National Association of Realtors (NAR) unfortunately rely on the data samples of relatively limited size. However, as discussed in Mortgage Market Note 13-01, FHFA has assembled a relatively large mortgage dataset comprised of loans that are guaranteed by either Fannie Mae or Freddie Mac (“the Enterprises”) and loans that are endorsed by the Federal Housing Administration (FHA). This dataset has been used for calculating first-time homebuyer shares. 1 The authors are Senior Economist, Principal Economist and Associate Director for Policy Research and Analysis respectively. They can be contacted at saty.patrabansh@fhfha.gov, andrew.leventis@fhfa.gov, and nayantara.hensel@fhfa.gov. 2 The AHS is a bi-annual survey of homeowner and home characteristics conducted by the U.S. Census Bureau and sponsored by the U.S. Department of Housing and Urban Development (HUD). Statistics derived from the AHS are a common information resource for housing researchers. 

2) FHFA Brief 14-02 Housing & Mortgage Policy Topics Mortgage Market Note 13-01 showed that, until 2007, the estimates from the Enterprise and FHA mortgage data tracked the aggregate U.S. estimates from the AHS and NAR very closely. It also showed that the overall market share of the Enterprises and FHA for purchase-money mortgages 3 has been 60 to 70 percent since 2008, which makes the combined mortgage dataset an excellent source for first-time homebuyer estimates over the past several years. The aforementioned Note also showed that first-time homebuyer shares were relatively stable until 2000. For the U.S., the share declined modestly from 46 percent in 1996 to 44 percent in 2000. 4 The first-time homebuyer share declined after 2000 and fell to 37 percent in 2003, remained fairly flat over the next few years, then significantly increased between 2006 and 2007, reaching 47 percent in 2007, 54 percent in 2008 and 63 percent in 2009, when a federal first-time homebuyer tax credit program was active. During the period of decline and flatness in the firsttime homebuyer share (2001-2006), the market shares of the Enterprises and FHA were also declining. The gentle decline in first-time homebuyer share during this period is also evident in the NAR and AHS estimates, which suggest that the decline is not entirely driven by the decreasing Enterprise and FHA market shares. After spiking in 2009-2010 as a result of tax credit programs, the first-time homebuyer share again drifted downward, reaching 56 percent by 2013. Exhibit 1 presents the share of first-time homebuyers in each state, computed from the loan-level data of Enterprise-FHA mortgages. The share is calculated as the number of purchase-money mortgages taken out by first-time homebuyers divided by the total number of purchase-money mortgages for primary homes. The vertical axis categorizes states by Census Division starting from the Pacific on the top to New England on the bottom. The horizontal axis shows the year of purchase from 1996 on the left to 2013 on the right. The table shows the first-time homebuyer shares for each state by year both as values and as a heat map, with larger shares shown in darker shades of green. States with the highest first-time homebuyer shares in recent years are: California, Nevada, the District of Columbia, Maryland, New York, New Jersey, Connecticut, Massachusetts and Rhode Island. These states are generally associated with high-cost metropolitan areas where job growth and worker mobility are likely to be higher and the share of first-time homebuyers in each of these states has exceeded 60 percent since 2011. In contrast, states with the lowest first-time homebuyer shares in recent years include: Montana, Wyoming, Iowa, Kansas, Wisconsin, Oklahoma, Arkansas, Kentucky, North Carolina, South Carolina, Vermont and Maine. Since 2011, shares in these states have been either at or below 50 percent, which is lower than the national share of about 56 percent. These states are generally associated with more non-urban areas. The share of first-time homebuyers in the remaining states has hovered around 50 to 60 percent in recent years. 3 “Purchase-money” mortgages are used for financing the purchase of homes. By contrast, refinance loans (which are not addressed in this paper) replace existing mortgages for borrowers who already own the collateral property. 4 Mortgage Market Note 13-01 addresses how these shares were calculated. As detailed in that paper, alternative metrics use different data sources and methodologies. First-Time Homebuyer Share and House Price Growth Page 2 

3) FHFA Brief 14-02 Housing & Mortgage Policy Topics New England Mid Atlantic South Atlantic East South West South Central Central East North Central West North Central Mountain Pacific State Census Division Exhibit 1. First-Time Homebuyer Share by State and Year HI AK OR WA CA NV UT AZ NM CO WY ID MT ND SD NE KS MO IA MN WI IL IN MI OH TX LA OK AR KY TN MS AL FL GA SC NC WV VA DC MD DE PA NJ NY CT RI MA VT NH ME Year 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 55 40 39 40 57 44 48 39 43 37 44 38 41 53 48 50 40 44 41 45 41 48 43 41 40 44 49 44 45 40 46 48 40 44 43 40 40 42 52 70 62 50 52 51 55 62 51 53 37 46 50 59 45 39 42 60 44 48 40 44 44 46 44 46 55 51 53 42 45 44 48 42 50 46 43 43 47 52 48 51 40 48 50 39 46 46 39 42 39 52 70 63 49 53 55 56 59 55 51 35 45 47 56 46 35 38 53 43 45 38 44 40 40 41 42 50 45 47 42 40 39 43 36 42 41 37 38 46 50 45 43 37 44 45 37 41 42 33 38 38 47 65 58 46 47 50 51 51 49 43 31 38 42 55 45 38 39 53 46 47 40 47 42 41 42 41 51 45 48 43 43 40 43 35 43 43 37 39 47 52 45 44 39 45 44 39 42 45 32 38 39 49 67 59 46 46 49 50 50 50 42 31 37 41 52 48 39 40 52 45 51 42 47 41 39 43 43 51 43 44 43 43 41 42 34 41 44 37 40 46 50 44 47 39 44 46 42 41 45 32 38 36 49 67 58 47 47 48 49 48 49 43 29 36 39 44 43 36 37 46 41 50 39 45 39 34 42 40 46 41 42 41 39 36 39 31 36 43 35 39 44 45 40 44 37 41 43 37 36 42 30 36 34 43 57 51 43 44 42 45 43 43 40 26 31 35 42 41 35 39 42 38 47 37 44 42 35 43 41 46 41 42 41 38 39 38 31 36 43 37 39 45 45 43 44 38 41 44 39 34 42 30 36 35 40 51 48 40 42 41 41 43 40 39 27 31 34 36 41 31 35 35 32 41 33 39 40 33 39 39 43 39 41 38 35 39 36 29 34 42 36 37 42 42 39 40 36 38 41 35 32 40 30 36 33 37 48 43 35 39 38 39 40 34 35 27 28 31 36 44 34 38 35 32 44 34 41 41 34 39 40 46 40 44 42 39 43 37 34 39 42 41 39 46 47 39 42 37 39 43 37 33 43 34 37 37 37 54 43 37 43 47 47 44 37 41 30 32 36 34 43 29 35 32 31 41 27 40 37 37 36 41 47 38 45 43 40 43 37 34 39 43 41 38 44 43 40 42 38 38 41 37 31 39 32 35 36 34 52 39 38 45 45 50 42 35 40 32 34 35 38 40 31 37 37 33 43 31 40 38 37 38 40 49 36 49 47 44 48 42 40 43 46 46 44 44 43 43 42 41 38 43 41 36 40 34 35 39 37 56 41 43 48 47 53 45 38 46 36 39 39 46 45 38 43 49 45 45 43 42 41 40 40 41 49 36 52 50 47 50 46 44 48 49 49 47 48 49 45 47 44 43 50 47 46 46 39 38 40 43 63 48 48 51 53 57 50 42 51 40 42 46 48 50 47 50 61 60 54 55 52 49 47 52 46 52 45 54 52 52 51 54 48 54 54 57 54 54 49 49 50 49 50 50 51 55 56 46 47 47 54 66 57 55 56 58 61 56 57 57 43 50 50 60 52 59 60 71 71 64 64 60 59 53 65 54 56 52 62 56 60 55 62 54 61 60 62 63 60 57 56 57 56 59 57 58 65 64 56 55 55 61 70 65 61 63 66 69 64 68 62 52 59 56 60 58 57 58 71 71 61 60 59 55 51 59 51 49 49 54 53 54 49 56 52 59 55 60 59 59 56 52 52 52 56 56 56 64 63 53 52 55 59 70 64 59 61 66 68 63 65 61 49 55 52 58 50 54 56 67 67 60 55 57 52 48 54 48 47 45 51 49 52 45 54 50 57 53 56 57 53 52 48 49 48 51 52 49 60 59 49 49 51 56 67 62 58 60 64 66 61 64 59 47 55 47 56 50 54 57 66 65 59 55 57 50 47 52 47 49 46 51 48 52 45 53 49 58 53 54 56 54 53 49 47 47 51 51 49 58 59 49 49 52 57 66 63 58 59 63 65 61 65 59 41 54 48 55 52 50 56 63 62 55 54 57 49 49 49 46 47 46 50 48 51 45 52 49 58 53 54 56 53 55 47 47 47 50 54 51 56 56 49 48 52 57 68 63 58 59 63 65 61 67 61 40 53 47 Source: FHFA calculations from the Enterprise and FHA mortgage data of purchase-money mortgages for primary homes. First-Time Homebuyer Share and House Price Growth Page 3 

4) FHFA Brief 14-02 Housing & Mortgage Policy Topics Since 1996, some states have seen a significant increase in the share of first-time homebuyers, with more than one-third experiencing an increase of over ten percentage points between 1996 and 2013. The following five states have witnessed an increase of over fifteen percentage points: Nevada, Rhode Island, Washington, Ohio, and Arizona. In sharp contrast to those areas, North Dakota, South Dakota, and the District of Columbia experienced declines in first-time homebuyer shares over the same period. Interestingly, Exhibit 1 reveals that some of the states that had the highest first-time homebuyer shares in 1996 (for example, California and Maryland) saw relatively little change over the 1996-2013 period. First-Time Homebuyer Share and House Price Growth On the surface, it is not clear whether the share of first-time homebuyers should be either directly or inversely related to changes in home values. Economic intuition suggests that either could be true. On the one hand, increases in home values could motivate potential first-time buyers to “get off the sidelines” and pursue the positive financial returns that can be derived from homeownership. On the other hand, increasing house prices clearly leads to decreases in affordability—a critical issue for first-time homebuyers who often are just getting started professionally and are still trying to save for a down payment. When prices are rising too quickly, potential first-time homebuyers may be priced out of the market until either their income rises, or house prices decrease to more affordable levels, or some combination of the two. This Brief examines whether the data support either one or the other of these two hypotheses using the simplest form of statistical analysis. To do so, two calculations were made for each state and each year since 1997. First, the change in the first-time homebuyer share was calculated for each state. Second, the FHFA House Price Index (HPI) was used to calculate year over year fourth-quarter house price growth for every state. 5 Exhibit 2 shows the scatterplot of the two variables and a simple linear model of change in firsttime homebuyer share (in the y-axis) and the house price growth (in the x-axis). The change in the first-time homebuyer share is negatively correlated with the change in house price growth. A simple regression model fitted to the data 6 suggests that a one percent increase in house prices tends to decrease the first-time homebuyer share by roughly one-quarter percent. 7 To further clarify the source of the negative correlation between the change in first-time homebuyer share and house price growth, correlations were calculated for each state. Exhibit 3 presents the simple correlation coefficient between the annual change in first-time homebuyer share and annual house price growth for each state. The results are sorted from the largest negative correlation to the largest positive correlation. 5 The FHFA HPI has been published since 1996 and tracks house price appreciation throughout the US. The — newest data can be downloaded at: http://www.fhfa.gov/DataTools/Downloads/Pages/House-Price-Index.aspx. 6 A regression model of this type is the simplest way of assessing the general relationship between two variables reflected in a scatterplot. 7 A change in first-time homebuyer share does not necessarily indicate a change in the number of first-time homebuyers. First-time homebuyer share can change even when the number of first-time homebuyers remains constant but the number of repeat buyers changes. Repeat buyers are purchasers who are looking to move from their currently owned homes. First-Time Homebuyer Share and House Price Growth Page 4 

5) FHFA Brief 14-02 Housing & Mortgage Policy Topics Annual Change in First-Time Homebuyer Share (Percentage Points) Exhibit 2. First-Time Homebuyer Share and Fourth Quarter House Price Index Annual State Level Changes, 1997-2013 20 15 10 5 0 -5 y = -0.2542x + 1.3645 R² = 0.1806 -10 -15 -40 -30 -20 -10 0 10 20 30 40 Annual Fourth Quarter House Price Index Growth (Percent) Source: FHFA calculations. Almost 90 percent of states exhibit a negative correlation, suggesting that first-time homebuyer share decreases as house price growth increases. Comparing the results in Exhibit 3 to the level of the homebuyer shares in the respective states, one finds the correlation coefficient is the most negative in several states with the highest first-time homebuyer shares. These include: California, Maryland, Nevada, Rhode Island and Massachusetts. Only 11 percent of states have weak positive correlations. States with weak positive correlations of over 0.1 are Alabama and Kansas. From 1996-2013, almost one third of the states experienced significant house price growth 8 at or more than 100 percent. The states with the most significant house price growth over this period were the District of Columbia (324 percent growth in the HPI Index), California (140 percent), Hawaii (133 percent), and Massachusetts (121 percent). Indeed, of the 13 states which had negative correlations exceeding -0.5 9, seven of them had a percent change in the HPI either at or more than 100 percent between 1996 and 2013. Only two of the states with negative correlations exceeding -0.5 had changes in the HPI index of less than 80 percent: Michigan (25 percent) and Nevada (41 percent). Several states with the most negative correlations had the most dramatic swings in house prices during the recent housing boom and bust. For example, Arizona saw prices rise by more than 150 percent between 1996 and 2006, and then saw prices recede by almost 50 percent in the early years of the housing bust. Nevada and California—two other states with large negative correlations reflected in Exhibit 3—also saw large swings in house prices. 8 9 Measured by the percent change in the HPI Index. Negative correlations exceeding -0.5 as measured in absolute value. First-Time Homebuyer Share and House Price Growth Page 5 

6) FHFA Brief 14-02 Housing & Mortgage Policy Topics Exhibit 3. Correlation of First-Time Homebuyer Share Change and Annual Fourth Quarter House Price Growth by State 1 0.8 Correlation Coefficient 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -1 CA MD NV FL RI AZ HI MI MA VA OR MN NH CT GA DC UT DE WA NM OH IL ME WI TN IN ID NJ NY CO MS SD NC AK LA SC TX MT WY MO PA KY ND WV OK AR VT IA NE AL KS -0.8 State Source: FHFA calculations. Given variations in the health of local labor markets, one might wonder whether the states with the strongest negative correlations had either significantly different unemployment or labor force participation rates of prime-aged home-buying adults (ages 25-44) compared to other states. Of the 13 states with negative correlations exceeding -0.5, only 5 of them had average prime-age unemployment rates during 1996-2013 exceeding 6 percent. Michigan had the highest average prime-age unemployment rate during this period (7.2 percent), while New Hampshire and Virginia had the lowest average prime-age unemployment rates (3.6 percent and 3.8 percent respectively). Moreover, about half of these states had a change in the prime-age unemployment rate of over three percentage points between 1996 and 2013, while half had a change in the prime-age unemployment rate of less than three percentage points. This suggests that high prime-age unemployment rates, and changes in the prime-age unemployment rates, were not necessarily the primary factors in the decline in first-time homebuyer shares as house price growth increased in some states. Prime-age labor force participation rates also did not show significant patterns among these states: over three quarters of the states had average prime-age youth labor force participation rates exceeding 83 percent. Nevertheless, further studies are needed on the interaction of labor markets and the relationship of first-time homebuyers and house price growth. 10 10 This analysis also examined whether either high unemployment rates or low labor force participation rates impacted first-time homebuyer shares. Though a positive correlation between the change in youth unemployment rate and the change in first-time homebuyer share was found, the correlation was very weak. There was no observable relationship between the change in youth labor force participation and the change in first-time homebuyer shares. First-Time Homebuyer Share and House Price Growth Page 6 

7) FHFA Brief 14-02 Housing & Mortgage Policy Topics Exhibit 4. Correlation of First-Time Homebuyer Share Change and Annual Fourth Quarter House Price Growth by Year 1 0.8 Correlation Coefficient 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 -1 Year Source: FHFA calculations. Between 1997 and 2013, the share of Enterprise and FHA mortgages relative to each other, and also relative to private mortgages, changed from year to year. Exhibit 4 examines cross-sectional state data by year in order to investigate whether the changing composition of the EnterpriseFHA sample could explain much of the observed negative correlation between house prices and first-time homebuyer shares over time. For each year, it shows the cross-sectional correlation between the annual change in first-time homebuyer share and annual house price growth across the 50 states. If the sole reason for the negative relationship between first-time homebuyer shares and price appreciation was that the combined Enterprise-FHA data sample captured a larger share of such borrowers in the overall market, then the year-specific correlations would not be systematically negative. Further, Exhibit 4 reveals that the negative correlations exist in nearly all years except 2010, the final year of the two-year federal first-time homebuyer tax credit program. Even during the height of the housing boom in either 2005 or 2006, the correlation coefficients are significantly negative: -0.56 and -0.60 respectively. In fact, in nearly all years in the 2001-2008 period 11 and also in 2013, the negative correlation coefficient was larger than -0.50 in absolute value. 12 This suggests that the negative relationship between the change in first-time homebuyer share and house price growth is robust and unaffected by the changing composition of the Enterprise-FHA sample over time. 11 Except 2004. The strongest negative correlations were in 2001 (-0.72), 2002 (-0.65), and in 2007 and 2008 (-0.69 for both years). 12 First-Time Homebuyer Share and House Price Growth Page 7 

8) FHFA Brief 14-02 Housing & Mortgage Policy Topics Conclusion The tendency of the first-time homebuyer share to decline as house price growth increases, as measured by the negative correlation between the change in first-time homebuyer share and house price growth, is evident in the aggregate correlation nationwide, as well as in the correlations across most states and in most years between 1996 and 2013. This provides evidence for the second hypothesis that increasing house prices may price some would-be firsttime homebuyers out of the market. Periods of declining house prices (i.e., increases in affordability), by contrast, apparently tend to induce relatively robust home purchasing volume by first-time homebuyers. Undoubtedly, additional research in this area is required. Future research could include further exploration of labor market outcomes such as employment and income, and household finance outcomes, such as student loan debt. First-Time Homebuyer Share and House Price Growth Page 8