Elsevier

Research in Economics

Volume 74, Issue 4, December 2020, Pages 301-322
Research in Economics

Research Paper
The effects of the monetary policy on the U.S. housing boom from 2001 to 2006

https://doi.org/10.1016/j.rie.2020.10.001Get rights and content

Highlights

  • This is a micro-founded DSGE model that features risk averse mortgage borrowers in the U.S. housing market.

  • Calibrated with the actual U.S. data, the model shows that Fed's monetary policy during 2001 and 2006.contributed significantly to the U.S. housing price boom.

  • Financial innovations alone, such as massive issuance of mortgage backed securities or relaxation of lending standards, had limit impact on house price.

  • Careful inspection of data and stylized facts confirms that monetary policy is more powerful in generating asset price inflation and driving financial innovations.

Abstract

This paper presents a DSGE model to test the relative significance of monetary policy and financial market innovations in creating the U.S. housing boom between 2001 and 2006. The model generates a trajectory of house price that mimics the Case–Shiller index well when actual Federal Fund rates are taken as inputs. It fails to do so when the monetary policy follows the Taylor rule even if MBS are introduced. We identify several transmission mechanisms of monetary policy with an emphasis on the financial accelerator. The model predicts that banks’ lending standards will go down with the benchmark interest rate.

Introduction

What caused the financial crisis of 2008? There are two schools of thought. One, represented by the erstwhile Chairman of the Federal Reserve, Ben Bernanke, blames financial deregulation and the consequent excessive risk taking in the form of financial innovations (Bernanke, 2010). The other, advanced by John Taylor (Taylor, 2008, 2012) and others, holds monetary policy as the primary contributing factor for the unprecedented U.S. housing boom that eventually led to the crisis. Bernanke defends the Fed's policy prior to the crisis as nothing unusual, whereas Taylor notes considerable deviations from the dictates of the Taylor rule (Taylor, 1993) between 2001 and 2005.

Dokko et al. (2009) provide the most comprehensive defense of the Fed's monetary policy to date. The authors run simulations of the Federal Reserve Board/U.S. (FRB/U.S.) model with actual federal fund rates as the inputs. They compare the simulation results with those when the interest rates are set according to the Taylor rule. The level of housing market activities (measured by residential investment/GDP) under actual policy rates is only slightly higher than that under the Taylor rule (see Edge et al. (2009) for the FBR/EDO model and similar conclusions). They also find an insignificant correlation between monetary policy and housing price. Based on these findings, the authors suggest that the major causes of the housing boom should be found elsewhere, for example, in unconventional financial products and financial deregulation.

However, both FRB/U.S. and FRB/EDO models feature no financial frictions. By contrast, we develop a model that explicitly features household mortgage borrowing and re-investigate the relative importance of the two sets of factors, that is, monetary policy and financial market conditions, in fostering the housing boom.

Many features of our model draw from the financial accelerator models of BGG (1999) and Carlstrom and Fuerst (2000), and housing model of Iacoviello and Neri (2010). We differ from the latter in that we do not rely on an exogenous borrowing constraint but rather derive it endogenously to capture the mortgage delinquencies. We differ from the traditional financial accelerator model in that the borrowers in our model are risk-averse households rather than risk neutral entrepreneurs, as we focus on the U.S. residential housing boom.

Our model features two types of households, Savers and Borrowers. Using houses as collateral, borrowers make one-period mortgages and use their own net worth to finance their consumption and acquisition of houses. A representative bank draws the saving household's deposits and pays the household a risk-free interest rate (the benchmark or policy rate) set by the central bank. It then makes a mortgage loan to every borrower according to an otherwise classic mortgage contract that specifies the loan quantity and interest rate, which are determined by the borrower's net worth, among other things.

Our model simulations show that persistently low interest rates alone can produce housing price boom of the scale comparable to that in the U.S. prior to the 2008 crisis. When the central bank cuts the benchmark interest rate, it affects the housing price through several channels. As savings with lower interest rate on deposits become less attractive, the saving household spends more on housing service and consumer goods, and the house price goes up immediately. In response to the rise in demand for consumer goods, the firm will increase production by employing more labor. Real wage rises, and it will lead to even greater demand for houses. Besides this traditional transmission mechanism of monetary policy, asset price is further reinforced by credit expansion of the bank. Borrowers respond to lower interest rates of mortgages by leveraging up. More importantly, with the windfall in net worth due to the appreciation of house value, borrowers have more to put up as collateral, and hence, they are able to borrow a greater amount. Meanwhile, as the expected default rate falls with improved investment returns in a bull market, the bank will offer more favorable terms, such as lower interest rates and less collateral, or equivalently, higher loan-to-value ratio (LTV). Monetary shocks are thus enlarged by the positive feedback between housing price and bank credit via the financial accelerator. As a result, a significant housing price boom emerges in our model.

We calibrate our model with the U.S. data and feed actual Federal Fund rates into the model for simulations. The model generates a housing price trajectory that mimics the movement of the Case–Shiller index fairly well. In contrast, when the monetary policy follows the Taylor rule, no dramatic asset price inflation can be observed.

We also conduct experiments to gage the significance of financial innovations, as compared to monetary policy, in creating housing boom. Massive use of mortgage-backed securities (henceforth, MBS) turn illiquid mortgages into liquid assets and allow banks to expand credit (Salmon 2010; U.S. Financial Crisis Inquiry Report 2011; Loutskina, 2011; Justiniano et al., 2015). The improvement in banks’ liquidity is captured in our model by an increase in supply of loanable funds for the bank or a positive shock to the loan-to-deposit ratio (henceforth, LTD ratio). To quantify the liquidity shocks, we divide the actual value of MBS issuance by U.S. household savings for the period 1997 to 2008. The extra liquidity provided by MBS was around 4% of household savings before 2001. It rose sharply in 2001q3 and continued the upward trend through 2003q3 to reach a peak of 14% (see Fig. 6 in Section 4). We model the liquidity effect of MBS to enlarge the household saving pool accordingly.

Model simulations under the MBS-induced liquidity shocks yield a housing price boom substantially smaller than that under the monetary shocks. The intuitions behind this result are not difficult to understand. The additional liquidity available does not affect the general taste of both borrowers and savers. Monetary easing, on the other hand, not only affects banks’ willingness to lend, but also affects the households’ willingness to borrow. A cut in benchmark policy rate directly lowers the interest rates on mortgage, and it also causes the household to substitute savings for housing and consumption, which adds to the upward pressure on property price. This substitution effect does not exist when MBS are used.

Having argued so, we warn against rushing to conclusions on the relative importance of monetary policy and MBS in creating asset bubbles. Restricted by the simple structure, we cannot include all effects of the two sets of factors into our model. For instance, MBS and other financial products may change risk preferences of market participants (Bies, 2004; Mian and Sufi, 2009; Nadauld and Sherlund, 2009; Schwartz, 2009). This issue is dealt with primitively in this paper by varying the Loan-to-Value (LTV) ratio exogenously. We treat more risk taking as equivalent to a higher LTV ratio of the bank or lower requirement for down payment (Duca et al., 2010). In doing so, we replace individual loan contracts in the baseline model with an aggregate and exogenous LTV ratio similar to Kiyotaki and Moore (1997), Iacoviello (2005), and Iacoviello and Neri (2010). The actual LTV ratio of the U.S. banking system was, by and large, stable between 1998 and 2001, with an average of around 0.87. It began to rise in 2002 and reached a peak of 0.97 in 2006. Had this ratio been applied to mortgages, it would mean nearly zero down payment, a practice often seen in the pre-crisis years. We plug the time series of real LTV ratio into our model and run simulations with interest rates set in line with the Taylor rule. The increases in the LTV ratio turn out to have no material impacts on housing price. The main reason lies on the demand side, where the Borrowers’ willingness to borrow is self-restrained because getting deeper into debt today would undermine their position to borrow in the near future.

We provide empirical and anecdotal evidence in Section 5 of this paper to support our theoretical results. Following a study of the Euro zone, we run a regression analysis for 11 Asian economies for the period 1990 to 1996, and we obtain a significant positive correlation between property price inflation and policy rate deviations from the Taylor rule. Housing booms in these economies have little to do with sophisticated financial products, and thus, they can be attributed mainly to monetary policies.

When interpreting the U.S. data, we refer to the literature to explore possible causalities between monetary policy and changes in the behavior of financial market players. The long period of low policy rate was associated with bank asset expansion and shifting toward riskier assets (Rajan, 2005, 2010; Ziadeh-Mikati, 2013). The Fed's own bulletin in 2010 admitted that, “Greenspan slashed interest rates and kept them too low for too long. Banks and shadow banks leveraged themselves to the hilt, loaning out money as if risk had been banished.” Loans were granted to riskier borrowers (subprime mortgages) at interest rates lower than what they could afford (McDonald and Stokes, 2013). Easy credit fueled the housing boom, and rising prices led to even looser standards of lending (Krugman 2009). Consistent with these arguments, MBS and sub-prime MBS data anomalies in the U.S. occurred after the Fed's interest rate policy began to diverge from the Taylor rule in the early 2000s. We are inclined to explaining, at least partially, these unusual phenomena as rational responses of financial institutions to monetary policies (See Zhang and Xu (2019) for a theoretical investigation of this argument).

The remainder of this paper is organized as follows. Section 2 briefly reviews the literature on the debate about this topic. Section 3 describes and calibrates the general equilibrium model in details. Section 4 reports simulation results, based on which we discuss the relative importance of monetary policy and financial innovations. Section 5 presents some empirical and anecdotal evidence to aid our theoretical analysis of the previous section. Section 6 concludes the paper.

Section snippets

Related literature on the debate

The investigation of housing price abnormality is still going on, and so far no consensus has been reached. There are abundant researches that study the role of relaxation of lending standards. Based on Kiyotaki and Moore (1997)’s exogenous borrowing constraint framework, where the maximum allowed credit is bounded by a share of the collateral value (the LTV ratio), these studies investigate how much the increasing LTV ratio contributes to the housing boom (Favilukis et al., 2015;

The model

The benchmark model to be constructed is a dynamic stochastic general equilibrium (DSGE) model with sticky price (Barsky et al., 2007) and risk averse borrowers under financial rigidities. Iacoviello (2005) uses an exogenous borrowing constraint model (also see Kiyotaki and Moore, 1997; Iacoviello and Neri, 2010) to study the housing market. We modify that model as later an extension to study the effect of financial innovation.

Consider an economy in an infinite time horizon, with total

Simulation and effects of monetary policy

In this section, we simulate our baseline model to gage the importance of monetary policy relative to financial innovations in creating asset bubbles. To clearly see the extent to which each set of factors affect house price, we mute all the other external shocks, including technological and preferential changes, in our model simulations. We first conduct experiments with monetary shocks, that is, deviations of the benchmark interest rate from the Taylor rule, but we do so without financial

Stylized facts

In this section, we present stylized facts about monetary policy, financial innovations, and their relationship with house price. The most challenging task of an empirical study on this subject is to separate the effect of monetary policy from all other effects. Given the limitations of data and econometric methods, we follow previous research to choose a sample in which the financial market can be considered to be in steady-state equilibrium.

Using data for a period without noticeable growth of

Conclusions and discussion

We constructed a DSGE model with an embedded mortgage lending to gage the impact of monetary policy on house price as compared to the liquidity effect of MBS. Our model simulations show that monetary policy is more powerful in generating asset price inflation than the use of liquidity-enhancing financial products. In the discussions following the model simulations, we identified multiple transmission mechanisms of monetary policy, namely substitution between consumption and saving, net worth,

Declaration of Competing Interest

There is no conflict of interest.

Acknowledgment

I thank Xiaonian Xu, Bala Ramasamy, John Taylor, Harald Uhlig, Bin Xu, Tian Zhu, and participants of seminars in CEIBS, and Fudan University for helpful comments.

References (54)

  • R.B. Barsky et al.

    Sticky-price models and durable goods

    Am. Econ. Rev.

    (2007)
  • A.B. Bernard et al.

    Plants and productivity in international trade

    Am. Econ. Rev.

    (2003)
  • B.S. Bernanke
  • B.S. Bernanke

    Monetary policy and the housing bubble

  • S. Bies

    “Innovation in financial markets and banking relationships

  • A.S. Blinder
  • C.T. Carlstrom et al.

    Agency costs, net worth, and business fluctuations: a computable general equilibrium analysis

    Am. Econ. Rev.

    (1997)
  • Chionsini, G., A. Foglia, and P.M. Reedtz, 2003, “Bank mergers, diversification and risk,” Working paper, bank of...
  • J.C. Coffee

    Analyzing the credit crisis: was the SEC missing in action?

    N. Y. Law J.

    (2008)
  • K. Dam et al.

    On the relationship between market concentration and bank risk taking

    Work. Pap.

    (2005)
  • G. Dell’ Ariccia et al.

    Lending booms and lending standards

    J. Financ.

    (2006)
  • J. Dokko et al.

    Monetary policy and the housing bubble

  • J.V. Duca et al.

    Shifting credit standards and the boom and bust in U.S. house prices

    Fed. Reserv. Dallas, Work. Pap.

    (2010)
  • Favilukis, J., S.C. Ludvigson, and S.V. Nieuwerburgh, 2015, “The macroeconomic effects of housing wealth, housing...
  • M. Friedman

    The Role of Monetary Policy

    Am. Econ. Rev.

    (1968)
  • S.A. Gabriel et al.

    Secondary markets, risk, and access to credit: evidence from the mortgage market

    Work. Pap.

    (2007)
  • D. Gale et al.

    Incentive compatible debt contract: the one-period problem

    Rev. Econ. Stud.

    (1985)
  • Cited by (2)

    View full text