Re-examining the decline in the US saving rate: The impact of mortgage equity withdrawal

https://doi.org/10.1016/j.intfin.2013.06.001Get rights and content

Highlights

  • Role of mortgage equity withdrawal in explaining decline in US saving rate.

  • VEC model with saving rate, asset prices, equity withdrawal and interest rates.

  • Mortgage equity withdrawal key determinant of savings.

Abstract

This paper investigates the effect of mortgage equity withdrawal on saving in the US over the period 1993–2011. A multivariate time series analysis based on a vector error correction model (VECM) is carried out. The saving rate, mortgage equity withdrawal, net wealth, interest rates and inflation are included in the empirical model. The results show that the equity withdrawal mechanism plays a relevant role in explaining the saving rate pattern.

Introduction

The US saving rate has been declining sharply since the 1990s. The personal saving rate dropped from an average of 8.6% in the period 1980–1990 to an average of 5.5 in the period 1990–2000. The average rate has fallen to 3.5% over the period 2000–2011. This decline is now considered a stylised fact and has attracted a lot of attention from academics and policy-makers. Greenwood and Jovanovic (1999) put forward the idea that recent advances in technology and in labour productivity have led US households to revise upwards their permanent income estimates. Lusardi et al. (2001) take the view that the appreciation of assets and the increase in medical care expenditure are the causes of the drop in the personal saving rate. Muellbauer (2008) argues that significant improvements in credit access have increased the ability of households to extract or borrow against their home equity, changing the saving behaviour in many countries.

In this paper, we focus on mortgage equity withdrawal (MEW – also known as home equity extraction) as a possibly important cause of the decline in the US saving rate. This is defined as the amount of equity that is extracted from the underlying asset when it appreciates. In general, when housing wealth increases (due to a rise in house prices) and mortgage rates are low, homeowners have an incentive to withdraw housing equity (see, e.g. Duca and Kumar, 2011, Paradiso et al., 2012, Paradiso, 2013) and this may increase consumption expenditure. Smith and Searle (2008) argue that housing equity withdrawal plays a role of a buffer against unexpected events allowing households to support their consumption plans over the life-cycle; Greenspan and Kennedy, 2005, Greenspan and Kennedy, 2008, and Hatzius (2006) also think that MEW has played a crucial role in determining private consumption expenditure. Empirical studies for the US show that regressions of consumption on mortgage equity withdrawal yield coefficients ranging from zero to as high as 0.62 for the long-run propensity to consume (Catte et al., 2004, Hatzius, 2006, Klyuev and Mills, 2010, Girouard, 2010). Specifically, Catte et al. (2004) find that MEW drives consumption with a marginal propensity to consume equal to 0.2 for the US when an error correction model including consumption, disposable income, net financial wealth, net housing wealth and MEW variables is estimated. Using a single equation error correction model, Hatzius (2006) finds that each dollar of MEW generates 62 cents of extra consumer spending when the consumption ratio, net wealth, interest rate and MEW are included in the analysis. Klyuev and Mills (2010) study the role of MEW in explaining the decline in the saving rate for different countries. Their empirical results for the US indicate that MEW is not statistically significant in a single equation error correction model with the saving rate, net wealth, interest rates and inflation. Girouard (2010) investigates the effects of housing wealth on the marginal propensity to consume in the US and other OECD countries and shows that they are stronger where mortgage markets are “most complete”, in particular where they provide opportunities for MEW.

This paper aims to contribute to the current literature on the decline of the US saving rate over the period 1993–2011 by focusing on the role of MEW in a multivariate time series framework. In particular, the analysis improves on the earlier studies discussed above in two respects. First, a VECM model is estimated instead of a single equation error correction model. This is important since the assumption of exogeneity implicitly made in a single equation model for the right-hand side variables (see Urbain, 1992, Ericsson and MacKinnon, 2002) may not be a valid one for MEW and housing wealth (see Mishkin, 2007, Iacoviello, 2011, among others). By contrast, in the Johansen's (1988) approach used here all variables are jointly modelled in a complete closed form model, full information analysis can be carried out and the number of cointegrating vectors can be determined performing appropriate cointegration tests. Second, the estimation of a multivariate model instead of a single equation one allows to investigate the dynamic linkages between the variables using impulse response analysis, a valuable tool in cointegrated systems (see Lütkepohl and Reimers, 1992). Since in such a framework the deviations from equilibrium are stationary, they will eventually revert to equilibrium, and their time paths provide useful insights into the short-run and long-run relationships between the variables of the system. Therefore, our approach enables us to investigate both the short- and long-run impact of mortgage equity withdrawal on the saving rate.

In the empirical analysis, we consider two specifications. First, we estimate a VECM with five variables typically used in the empirical literature (see Hatzius, 2006, Klyuev and Mills, 2010), namely the saving rate, net wealth, real mortgage interest rates, inflation and MEW. Second, a VECM with disaggregate net wealth, housing and non-housing wealth, is estimated. A partition of net wealth is here considered because housing wealth is often viewed as the main determinant of consumption expenditure (Poterba, 2000, Kishor, 2007) and, after the housing bubbles of recent years, its relative weight has increased further (see Donihue and Avramenko, 2007, Iacoviello, 2011).

The empirical results show that the signs of the estimated long-run coefficients on housing wealth, real mortgage interest rates, and inflation are not significant. Since the restrictions on these coefficients and corresponding factor loadings are found to hold, we estimate two three-variate VECMs excluding housing wealth, inflation and real interest rates in the six-variate VAR and inflation and real interest rates in the five-variate VAR. The impulse-response analysis conducted on the two VECMs with three variables indicates that a positive shock to mortgage equity withdrawal has a significant negative effect on the saving rate, showing that mortgage equity withdrawal is an important driver of the saving pattern over the last 20 years.

The paper is organised as follows. Section 2 describes the data. Section 3 presents the empirical results. Section 4 offers some concluding remarks.

Section snippets

Data description

For the empirical analysis, we use quarterly data over the period 1993:Q1–2011:Q1. The series are: the saving rate, total net wealth, housing and non-housing net wealth, the real mortgage rate, inflation and mortgage equity withdrawal. The saving rate is the personal saving rate and the data have been obtained from the Bureau of Economic Analysis (BEA). Total net wealth, housing and non-housing wealth are constructed from the flow-of-funds accounts of the Boards of Governors of the Federal

Empirical results

In the empirical analysis we consider two VECMs. The first specification includes five variables: the saving rate (sr), net wealth (nw), the real mortgage rate (imor), the inflation rate (inf) and active MEW (amew); the second includes sr, imor, inf, amew, housing (hw) and non-housing (nhw) wealth.

As a preliminary step, we investigate the unit root properties of the variables using the ADF and DF-GLS tests. The results are reported in Table 1. The null hypothesis of a unit root cannot be

Conclusions

This paper contributes to the current literature on the behaviour of the US saving rate by focusing on the role of mortgage equity withdrawal. Whilst previous studies have analysed the relationship between the saving rate, mortgage equity withdrawal, net wealth and interest rates in a single equation error correction model, the present one estimates a vector error correction model since the assumption of exogeneity implicitly made for the right-hand side variables of a single equation model may

References (43)

  • H.D. Gibson et al.

    The Greek financial crisis: growing imbalances and sovereign spreads

    Journal of International Money and Finance

    (2012)
  • H. Lütkepohl et al.

    Impulse response analysis of cointegrated systems

    Journal of Economic Dynamics and Co

    (1992)
  • A. Paradiso et al.

    US inflation and consumption: a long-term perspective with a level shift

    Economic Modelling

    (2012)
  • G.S. Bakshi et al.

    Inflation, asset prices, and the term structure of interest rates in monetary economies

    Review of Financial Studies

    (1996)
  • G.B. Canner et al.

    Mortgage refinancing in 2001 and early 2002

    Federal Reserve Bulletin

    (2002)
  • P. Catte et al.

    Housing markets, wealth and the business cycle

    (2004)
  • B.Z. Cynamon et al.

    Household debt in the consumer age: source of growth-risk of collapse

    Capitalism and Society

    (2008)
  • P.E. Davis et al.

    Bank lending and commercial property cycles: some cross-country evidence

    Journal of International Money and Finance

    (2011)
  • R. Disney et al.

    House price volatility and household indebtedness in the US and the UK

    (2009)
  • M. Donihue et al.

    Decomposing consumer wealth effects: evidence on the role of real estate assets

    B.E. Journal of Macroeconomics (Topics)

    (2007)
  • J.A. Doornik

    Approximations to the asymptotic distributions of cointegration tests

    Journal of Economic Surveys

    (1998)
  • J.V. Duca et al.

    Financial literacy and mortgage equity withdrawals

    (2011)
  • G. Elliot et al.

    Efficient tests for an autoregressive unit root

    Econometrica

    (1996)
  • N.R. Ericsson et al.

    Distributions of error correction tests for cointegration

    Econometrics Journal

    (2002)
  • N. Girouard

    Housing and mortgage markets: an OCED perspective

  • A. Greenspan et al.

    Estimates of home mortgage originations, repayments, and debt on one-to-four family residences

    (2005)
  • A. Greenspan et al.

    Sources and uses of equity extracted from homes

    Oxford Review of Economic Policy

    (2008)
  • J. Greenwood et al.

    The information-technology revolution and the stock market

    American Economic Review

    (1999)
  • H. Hansen et al.

    Some tests for parameter constancy in cointegrated VAR-models

    Econometrics Journal

    (1999)
  • J. Hatzius

    Housing holds the key to fed policy

    (2006)
  • M. Iacoviello

    Housing wealth and consumption

    (2011)
  • Cited by (7)

    View all citing articles on Scopus
    View full text