VoxEU Column Financial Markets Macroeconomic policy

Evidence of a housing ATM effect? The destination of home equity extracted by US households during the recent housing boom

In this column, Federal Reserve Bank of Boston economist Daniel Cooper presents new evidence suggesting that the spending impact of equity extraction during the recent US housing boom was relatively small compared with the household balance sheet changes and residential investment. This finding contrasts with recent findings claiming that households consumed the vast majority of the money they extracted.

Economists are interested in studying the borrowing response of households to rising house prices and increasing home equity during the recent house price boom for two main reasons. First, it addresses the question of whether rising house prices in the US fuelled consumption through equity-extraction-financed spending. Second, analysing the destination of funds extracted when house prices rose provides insight into what parts of the macroeconomy might suffer now that house prices have dropped and a large amount of homeowners’ equity has evaporated.

Work by Kennedy and Greenspan (2007) shows that households’ net equity extraction from their homes averaged nearly 6% of disposable income between 2001 and 2005. Figure 1 below shows how extracted equity relative to income increased sharply along with house prices in the US during the housing boom.

There are many reasons why households might extract equity from their homes: to finance household expenditures, to make major home repairs or improvements (so-called residential investment), to consolidate more costly debt, such as credit card debt, to invest in personal businesses or other forms of entrepreneurship, to help finance the purchase of other real estate, or to engage in a form of investment arbitrage, or a mixture of the above. Recently in the US, home equity credit has been one of the cheapest forms of borrowing, with tax advantaged treatment and rates below those on credit card debt.

Economists took an interest in the destination of households’ extracted funds as early as 2002. Canner et al. (2002) used survey data to examine the destination of extracted funds. They found that on average roughly 16% of the dollars borrowed went toward consumer purchases, 35% went toward home improvements, 26% went to repaying other debt, and the remaining portion went to stock market and business investments. The bottom line is that in the early 2000s households reported spending very little of the money they borrowed against their homes on current consumption and instead used the money to repay debt and make residential and other household investments.

More recently, Mian and Sufi (2009) used loan-level data to investigate households’ equity extraction behaviour. They found little evidence that households’ borrowing in response to house price increases was used for home purchases, home improvements, or debt repayment. They argue by process of elimination, since their dataset lacks household spending data, that households must spend the vast majority of their extracted funds on current consumption. Disney and Gathergood (2009) examined the relationship between changes in house prices and changes in households’ secured debt holdings in both the US and the UK. They found that equity extraction due to rising house prices was much more prevalent in the US. They further observed that UK households extracted equity primarily to repay unsecured debt such as credit card debt, while, in contrast, US households appear to have spent most of the money they borrowed on current consumption.

Table 1 shows the destination of extracted equity based on independent analysis of households’ consumption, saving, and residential investment behaviour. I use data from the Panel Study of Income Dynamics (PSID) for this analysis. The PSID is advantageous for this study because it contains detailed information on households’ housing debt, financial and nonfinancial assets, income, consumption, and a variety of household characteristics. More specific details of the analysis can be found in Cooper (2010).

New insight: Equity extraction has a small consumption effect in the US

My results suggest that equity extraction has a small consumption effect in the US. Unlike the Mian and Sufi or Disney and Gathergood papers, I use actual household spending data to track households’ equity extraction behaviour rather than inferring their consumption response to equity extraction based on other, non-spending results. In particular, a $1 increase in equity extraction between 2003 and 2007 led to a 10-cent to 20-cent increase in overall non-housing expenditures for homeowners who stayed in their original home. This effect was strongest in the two-year period between 2003 and 2005 that preceded the downturn in US house prices in late 2006. Equity extraction had a much smaller impact on household spending in late 1999 and early 2001.

My paper also explores the relationship between household spending and equity extraction across different groups and finds heterogeneity in households’ behaviour (results not reported here). In particular, the consumption of 35- to 50-year-olds is essentially unaffected by equity extraction. When I control for households in that age range with bad income shocks, however, I find that such households increase their consumption by 10 cents to 15 cents when they extract equity. This finding is broadly consistent with consumption smoothing through home equity borrowing by potentially financially constrained households in response to negative income shocks. Being able to control for households’ idiosyncratic income changes is a further advantage of using PSID data to analyse equity extraction.

Turning to the saving and residential investment results, it is clear that households also extracted equity to fund home improvements and other household investment needs. During the 2003 to 2005 and 2005 to 2007 periods, a $1 increase in equity extraction led to a roughly 20-cent increase in capital spending on home additions and improvements for households that made such improvements. Household saving increased by a similar amount over those time intervals. Households’ saving response was primarily concentrated in investment in financial assets, personal businesses, and other real estate. The combined residential and personal investment impact of equity extraction is roughly double the consumption effect during these periods. These findings are more in line with the early work of Canner et al. (2002) and less consistent with the more recent equity extraction analysis.

The estimated effects in Table 1 do not sum to one. In other words, I cannot explain the full destination of each dollar households extract. One potential reason is measurement error. Household level data is inherently noisy so I am unlikely to be able to explain the entire destination of each dollar extracted. In addition, I do not observe household behaviour at the exact moment the equity extraction occurs. For example, households could extract equity to pay off credit card debt, but then run that debt back up over my sample period. Such behaviour is consistent with the fact that I observe no debt repayment effect. I also cannot fully capture households that extract home equity to help finance the purchase of a new home. I can, however, force the coefficients in Table 1 to add to one as part of the estimation process. To do this I use a technique called “seemingly unrelated regression analysis”. These results are presented in Table 2. Even under this restriction, the consumption impact of equity extraction is still relatively small. The impact is largest by far in the 2005 to 2007 period, when roughly 50 cents of each dollar extracted went to household spending.

Contrast with recent studies

My results suggest that equity extraction had a limited impact on household spending especially relative to household residential and other investments and balance sheet reshuffling. My findings contrast with the findings of Mian and Sufi (2009), which suggest that the vast majority of extracted equity went toward financing household consumption expenditures. The magnitudes of my findings are more similar to Canner et al. (2002), although our approaches are very different. Those authors use survey data that asks households how they spent their extracted equity. They observe only whether households say they spent some of the money on consumption, not the actual amount. As a result, Canner et al. capture average effects rather than the marginal dollar effects that I report.

Overall, there is little doubt that US households dramatically increased their housing-related debt during the recent housing boom. The evidence regarding what they did with the money they extracted from their homes is mixed. Based on my work, I would argue that much of the money went toward residential and other household investment, and not toward consumption. The policy implication of this finding is that the drag on consumption resulting from households’ loss of home equity available to finance expenditures should be relatively small, especially compared with previous estimates

Disclaimer: The views expressed are those of the author and do not necessarily reflect the view of the Federal Reserve Bank of Boston or the Federal Reserve Board of Governors.


Canner, Glenn, Karen Dynan, and Wayne Passmore (2002), “Mortgage Refinancing in 2001 and Early 2002”, Federal Reserve Bulletin:469-481, December.

Cooper, Daniel (2009), “Did Easy Credit Lead to Overspending? Home Equity Borrowing and Household Behaviour in the Early 2000s”, Federal Reserve Bank of Boston Public Policy Discussion Paper Series, no. 09-7.

Disney, Richard and John Gathergood (2009), “House Price Volatility and Household Indebtedness in the US and the UK”, Centre for Finance and Credit Markets, Working Paper, 09/02.

Mian, Atif and Amir Sufi (2009), “Home Prices, Home Equity-Based Borrowing, and the US Household Leverage Crisis”, NBER Working Paper 15283.

1,472 Reads