Economics Letters 119 (2013) 24–27

Contents lists available at SciVerse ScienceDirect

Economics Letters

journal homepage: www.elsevier.com/locate/ecolet

Asset prices, monetary policy, and aggregate fluctuations:An empirical investigationLichao Cheng a,1, Yi Jin b,∗

a China Life Center, Jinrong Street, Xicheng District, Beijing 100033, Chinab Department of Economics, Monash University, Caulfield East, VIC 3145, Australia

a r t i c l e i n f o

Article history:Received 17 November 2012Received in revised form2 January 2013Accepted 8 January 2013Available online 17 January 2013

JEL classification:E31E32E44E52

Keywords:House pricesStock pricesTerm structureMonetary policyStructural vector autoregressions

a b s t r a c t

This paper studies empirically the dynamic interactions between asset prices, monetary policy, andaggregate fluctuations in the U.S. during the Volcker–Greenspan period. Results from a simple structuralvector autoregression indicate that monetary policy reacts directly to the term spread and indirectly tostock prices and house prices via output and inflation, that there is an asymmetry in the interactionsbetween asset prices and aggregate activity, and that asset prices exhibit positive comovement.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

This paper employs a simple structural vector autoregression(SVAR) to investigate the dynamic interactions between assetprices, monetary policy, and aggregate fluctuations in the U.S. dur-ing the Volcker–Greenspan period. An innovative aspect of ourstudy is that it incorporates stock prices, house prices, and the termstructure of interest rates.2 Our specification extends Christianoet al. (1999, CEE henceforth) to incorporate asset prices. Specif-ically, we consider an SVAR which consists of six quarterly vari-ables, grouped in the vector Zt = [Yt , πt , Rt , TSt , qHt , qSt ]′, where

∗ Corresponding author. Tel.: +61 3 99032042; fax: +61 3 99031128.E-mail addresses: chenglc@clamc.com (L. Cheng), yi.jin@monash.edu (Y. Jin).

1 Tel.: +86 10 66221371.2 The paper is related to the empirical literature on the relationship between

monetary policy, the business cycle, and stock prices (Fama and French, 1989;Patelis, 1997; Thorbecke, 1997; Rigobon and Sack, 2003; DeStefano, 2004; Bernankeand Kuttner, 2005; Bordo et al., 2007, etc.). It is also related to the growing bodyof literature on the role of housing market in the macroeconomy (Iacoviello andMinetti, 2003, 2008; Jin and Zeng, 2004; Leung, 2004; Iacoviello, 2005; Chen andLeung, 2007; Del Negro and Otrok, 2007; Carrington and Madsen, 2011; Jaccard,2011; Chen et al., 2012; Chang et al., 2012; Punzi, forthcoming, among others).

0165-1765/$ – see front matter© 2013 Elsevier B.V. All rights reserved.doi:10.1016/j.econlet.2013.01.005

Yt represents the log real GDP,πt inflation,Rt the federal funds rate,TSt the term spread, qHt the log real house price, and qSt the log realstock price. Following Iacoviello (2005), we take the ConventionalMortgage Home Price Index (CMHPI) to be the measure of houseprice. The Standard&Poor’s 500 index (S&P500) is used as themea-sure of stock price. Both prices are divided by the GDP deflator. Theterm spread is the difference between the yields on the 10-yeargovernment bond and the 3-month T-bill. The inclusion of the termspread, which is inversely related to the long-term bond price, fol-lows the lead of Chang et al. (2011). The list of these six variables ismeant to represent in a concisemanner the broad picture ofmone-tary policy setting, macroeconomic performance, and the financialaspects of the economy.3 We also add the CRB/BLS Spot Price Indexas an exogenous variable to help resolve the ‘‘price puzzle’’.

We adopt a variant of CEE’s short-run recursiveness identifica-tion scheme. Although a fully fledged analysis would involve otheridentification strategies, such as the use of long-run restrictionsor sign restrictions, our SVAR has already delivered a rich set of

3 The data sources are as follows. Federal funds rate: Federal Reserve Board. Termspread: Bloomberg. CMHPI: Freddie Mac. Stock price index: Online Data RobertShiller. Real GDP and GDP deflator: Bureau of Economics Analysis.

L. Cheng, Y. Jin / Economics Letters 119 (2013) 24–27 25

Fig. 1. Impulse responses.

important results, despite its apparent simplicity. The ordering ofvariables we adopt captures the idea that asset prices are the mostresponsive, while output and inflation are the most sluggish, re-flecting the common perception that changing the quantities andprices of most goods and services is subject to various sorts of ad-justment costs (Sims and Zha, 2006). Our sample period covers theVolcker–Greenspan period, i.e., 1979Q3–2006Q1. This reflects ourdesire to look at a historical period where monetary policy wasconducted in a relatively consistent manner, i.e., without majorstructural breaks. The same consideration leads us to defer to fu-ture research the investigation of the recent Global Financial Crisisand the ensuing Great Recession, when monetary policy appearedto be ‘‘unconventional’’. The analysis in this paper is thus moreabout the normal time of business of monetary policy making. Asfor the number of lags, the Akaike information criterion (AIC) sug-gests three, while the Schwarz information criterion suggests one.As a compromise, we take two for the benchmark.4

2. Results

As the dynamic interactions between monetary policy, output,and inflation have been well known from CEE and the relatedliterature, our presentation below focuses on the results relatedto asset prices. The impulse responses are presented in Fig. 1(the dashed lines represent 90% confidence intervals). We also useforecast error variance decomposition to gauge the contributionsof shocks to the volatilities of the endogenous variables.

4 We have also conducted robustness checks including (1) changing the orderingof variables, (1) changing the number of lags from two to three (as theAIC indicates),(2) using the consumer price index instead of GDP deflator to deflate stock andhouse prices, and (3) using the American Stock Exchange’s AMEX index to replacethe S&P500 index as themeasure of stock price. Broadly speaking, the results remainrobust against these changes.

2.1. Systematic monetary policy versus monetary policy shock

The monetary policy instrument – the federal funds rate –contains both a systematic component (reaction to the state ofthe economy) and a shock component (called ‘‘monetary policyshock’’). Our estimation results indicate that the funds rate de-clines following a positive term spread shock. This means that thecentral bank lowers the policy rate when it anticipates economictightening in the future, or when it believes that private agentsanticipate future tightening. On the other hand, a positive fundsrate shock reduces the term spread, implying that future short-term rates are expected to rise by less than the current short-termrates. We also find that the funds rate does not react directly tostock and house price movements, i.e., the estimated coefficientson stock and house prices in the funds rate equation are statisti-cally insignificant. However, this does not preclude the funds ratefrom responding indirectly to these prices via output and inflationmovements. Indeed, the impulse responses show that a positivestock price shock has positive, lagged effects on the funds rate.

For short forecast horizons, monetary policy shock is the domi-nant source of funds rate variability. Output, inflation, and term-spread shocks are also important, while stock and house pricemovements have negligible influences. Over time, inflation shockgains importance. And stock and house price shocks have some im-pacts on the funds rate volatility over long forecast horizons. Inthe unconditional variance decomposition, monetary policy shockaccounts for 31% of the funds rate volatility, inflation shock ac-counts for 22%, term spread shock 18%, output shock 14%, stockprice shock 10%, and house price shock 6%. Hence asset pricemove-ments are not to be neglected when accounting for the volatility ofthe monetary policy instrument.

The effects of a positive funds rate shock on stock and houseprices are both negative, with noticeable differences. First, thestock price declines by a much greater extent. In response to a

26 L. Cheng, Y. Jin / Economics Letters 119 (2013) 24–27

one standard deviation shock to the funds rate, the peak decline ofhouse price is about 0.3%, while the peak decline of stock price is aslarge as 1.5%. Second, the decline of house price is more persistentthan that of stock price. Although monetary policy shock has alarger impact on stock price in absolute terms, it is relatively moreimportant in generating house price fluctuations (this is possibleas the stock price is much more volatile than the house price).According to the variance decomposition, monetary policy shockcontributes around 20% to house price volatility, and only around10% to stock price volatility. Hence monetary policy shock is animportant source of house price volatility.

2.2. Asset prices and aggregate fluctuations

There is an important asymmetry in the interactions betweenasset prices and output fluctuations. The contributions of out-put shock to the volatilities of asset prices are very limited,around 4.4%, 6.6%, and 8.0% for stock price, house price, and theterm spread, respectively. In the reverse direction, however, as-set price shocks, especially stock price shock, appear to be impor-tant sources of output fluctuations. In the unconditional variancedecomposition, stock price, house price, and term spread shockscontribute around 14.2%, 7.3%, and 10.9%, respectively, to outputvolatility. From Fig. 1, a positive stock price shock and a positivelong-term bond price shock (negative term spread shock) bothlead to significant output expansions. One possible explanationfor the asymmetric output–asset price interactions is that non-fundamental factors might have disproportionate importance ingenerating asset price movements, so that asset prices are drivenmore by expectations, beliefs, and contagions than by output fluc-tuations. On the other hand, to the extent that net worth, collat-eral values, and costs of finance are important determinants ofinvestment, asset price movements can be important for outputfluctuations.

Asset price shocks are also important determinants of infla-tion volatility over long forecast horizons. Indeed, the term spreadshock is the most important source of inflation fluctuations otherthan the shock to inflation itself. In the unconditional variance de-composition, the contribution of the term spread shock to infla-tion volatility is 14.79%, larger than the 12.33% contribution fromthe funds rate shock. The contributions from house price and stockprice shocks are 9.87% and 7.56%, respectively, both greater thanthe 5.06% contribution of output shock. Positive stock price andhouse price shocks and negative term spread shocks (positive long-term bond price shock) all stimulate inflation. And the effects arequite persistent.

2.3. Asset price comovement

Another important message of our analysis is that asset pricesexhibit positive comovement, and that term spread movementsare important forces behind this. A negative term spread shock (ora positive shock to long-term bond price) drives up both the stockprice and the house price. In the unconditional variance decompo-sition, the term spread shock contributes 21% to stock price volatil-ity, amagnitude doubling the contribution of each of the funds rateand inflation shocks. The term spread shock is not only the largestsource of stock price volatility other than the shock to the stockprice itself, but also one of the most important sources of houseprice volatility other than the shock to the house price itself. Thecontribution of the term spread shock to house price volatility is19%, only slightly less than the 20.5% contribution from the fundsrate shock and somewhat larger than the 16% contribution from in-flation shock. On the other hand, stock price andhouse price shockscontribute little to the term spread volatility.

Although a positive stock price shock also drives up the houseprice, and vice versa, the responses are not significant. This does

not preclude the stock price and the house price from comovingpositively. In fact, a rise in the stock price (respectively, houseprice), as triggered by a decline in the term spread, can cause thehouse price (respectively, stock price) to increase. When we re-move the term spread from the SVAR, we find that the responseof house price to a stock price shock is significantly positive, andvice versa, and that the contribution to each other’s volatility is im-portant. This means that the stock price (respectively, house price)serves as an intermediate channel through which the term spreadaffects the house price (respectively, stock price).

There are various reasons for the positive comovement betweenasset prices. First, an increase in the price of one asset raises thewealth of households and firms, thereby pushing up the demandand hence the price for the other assets through a direct wealtheffect. Second, the increase in the price of any asset raises the networth of households and firms andhence strengthens their balancesheets. This allows them to borrow more at lower cost, whichstimulates investment and bids up asset prices. Third, an increasein one asset price might trigger the expectation that other assetpriceswill also increase, leading to expectation-drivenmovements(and comovement) in asset prices.

3. Conclusions

In this paper, we have presented a set of important results com-ing out of a simple recursive SVAR. The results should provide someguidance for future development of quantitative dynamic stochas-tic general equilibrium models on the interactions among assetprices, monetary policy, and aggregate fluctuations. In future re-search, we plan to explore other identification strategies, such aslong-run restrictions and sign restrictions. On another front, thefocus on a historical period where monetary policy was conductedin a relatively consistent manner has led us to work with the Vol-cker–Greenspan period. It is an important task in future research tostudy the interactions between asset prices, macroeconomic per-formance, and the ‘‘unconventional’’ monetary policy during theGlobal Financial Crisis and the ensuing Great Recession.

References

Bernanke, Ben S., Kuttner, Kenneth N., 2005. What explains the stock market’sreaction to federal reserve policy. Journal of Finance 60 (3), 1221–1257.

Bordo, Michael D., Dueker, Michael J., Wheelock, David C., 2007. Monetary policyand stock market booms and busts in the 20th century, Federal Reserve Bankof St. Louis Working Paper.

Carrington, Sarah J., Madsen, Jakob B., 2011. House prices, credit and willingness tolend. Economic Record 87 (279), 537–557.

Chang, Kuang-Liang, Chen, Nan-Kuang, Leung, Charles Ka Yui, 2011. Monetarypolicy, term structure and asset return: comparing REIT, housing and stock.Journal of Real Estate Finance and Economics 43, 221–257.

Chang, Kuang-Liang, Chen, Nan-Kuang, Leung, Charles Ka Yui, 2012. Would somemodel please give me some hints? In search of some stylized facts of themonetary policy and asset return dynamics, mimeo.

Chen, Nan-Kuang, Cheng, Han-Liang, Mao, Ching-Sheng, 2012. House price, mort-gage premium, and business fluctuations. Economic Modelling 29, 1388–1398.

Chen, Nan-Kuang, Leung, Charles Ka Yui, 2007. Asset price spillover, collateral andcrises: with an application to property market policy. Journal of Real EstateFinance and Economics 37, 351–385.

Christiano, Lawrence J., Eichenbaum, Martin, Evans, Charles, 1999. Monetarypolicy shocks: what have we learned and to what end? In: Taylor, John,Woodford, Michael (Eds.), Handbook of Macroeconomics, Vol. 1. Elsevier.

Del Negro, Marco, Otrok, Christopher, 2007. 99 Luftballons: monetary policy andthe house price boom across U.S. States. Journal of Monetary Economics 54,1962–1985.

DeStefano, Michael, 2004. Stock returns and the business cycle. The FinancialReview 39 (4), 527–547.

Fama, Eugene F., French, Kenneth R., 1989. Business conditions and expectedreturns on stocks and bonds. Journal of Financial Economics 25 (1), 23–49.

Iacoviello, Matteo, 2005. House prices, borrowing constraints, and monetary policyin the business cycle. American Economic Review 95 (3), 739–764.

Iacoviello, Matteo, Minetti, Raoul, 2003. Financial liberalisation and the sensitivityof house prices to monetary policy: theory and evidence. The ManchesterSchool 71 (1), 20–34.

L. Cheng, Y. Jin / Economics Letters 119 (2013) 24–27 27

Iacoviello, Matteo, Minetti, Raoul, 2008. The credit channel of monetary pol-icy: evidence from the housing market. Journal of Macroeconomics 30 (1),69–96.

Jaccard, Ivan, 2011. Asset pricing and housing supply in a production economy. TheB.E. Journal of Macroeconomics 11 (1), (Contributions), Article 33.

Jin, Yi, Zeng, Zhixiong, 2004. Residential investment and house prices in a multi-sector monetary business cycle model. Journal of Housing Economics 13,268–286.

Leung, Charles Ka Yui, 2004. Macroeconomics and housing: a review of theliterature. Journal of Housing Economics 13, 249–267.

Patelis, A., 1997. Stock return predictability and the role of monetary policy. Journalof Finance 52 (4), 1951–1972.

Punzi, Maria Teresa, Housing market and current account imbalances in theinternational economy. Review of International Economics (forthcoming).

Rigobon, Roberto, Sack, Brian, 2003. Measuring the reaction of monetary policy tothe stock market. Quarterly Journal of Economics 118 (2), 639–669.

Sims, Christopher A., Zha, Tao, 2006. Does monetary policy generate recessions?Macroeconomic Dynamics 10 (2), 231–272.

Thorbecke, Willem, 1997. On stock market returns and monetary policy. Journal ofFinance 52 (2), 635–654.