wealth effects in emerging market economies

Wealth effects in emerging market economies

Tuomas A. Peltonen a, Ricardo M. Sousa b,c,⁎, Isabel S. Vansteenkiste d

a European Central Bank, Kaiserstraße 29, D-60311 Frankfurt am Main, Germanyb University of Minho, Economic Policies Research Unit (NIPE), Campus of Gualtar, 4710-057, Braga, Portugalc London School of Economics, Financial Markets Group (FMG), Houghton Street, London WC2 2AE, United Kingdomd European Central Bank, Kaiserstraße 29, D-60311 Frankfurt am Main, Germany

a r t i c l e i n f o a b s t r a c t

Article history:Received 24 June 2010Received in revised form 17 January 2012Accepted 19 January 2012Available online 27 January 2012

We build a panel of 14 emerging economies to estimate the magnitude of wealth effects onconsumption. Usingmodern econometric techniques and quarterly data,we show that: (i)wealtheffects are statistically significant and relatively large inmagnitude; (ii) stockmarket and housingwealth effects are, generally, smaller for Latin American emerging markets; and (iii) housingwealth effects have substantially increased for Asian emerging economies in recent years.Additionally, while housing wealth effects are more important in countries with low levelof financial development or low income level, financial wealth effects are stronger for coun-tries with high stock market capitalization.

© 2012 Elsevier Inc. All rights reserved.

JEL classification:E21E44D12

Keywords:Wealth effectsConsumptionEmerging markets

1. Introduction

Household consumption is affected not only by income (Mallick, 2008), but also by wealth such as real estate and stock own-ership (Sousa, 2010a). When real estate or stock prices rise, the wealth of homeowners or shareholders increases and householdconsumption can rise even when labor income remains constant. Such rise in consumption due to the increase in real estate pricesis called housing wealth effect, whereas the rise in consumption that is due to the increase in stock market prices is called stockmarket wealth effect.

There is a large body of literature that studies the effect of asset price fluctuations on private consumption and authors haveused different econometric techniques (such as panel versus single equation models) and databases (like micro panel data andaggregate time series) to address the issue. More recently, interest in the topic has regained ground against the background ofthe strong linkages between the macroeconomy (in particular, private consumption) and the wealth dynamics, which has ledto concerns by numerous academics, central banks and governments about the potential implications of downturns in housingand equity prices (Sousa, 2010b).

Despite the wide range of studies, most of the empirical evidence refers to advanced economies and mostly to the UnitedStates, where data is more readily available. Extending the existing literature to assess the macroeconomic impact of assetprice fluctuations in emerging markets may, therefore, be important as these economies are becoming a key engine of growth

International Review of Economics and Finance 24 (2012) 155–166

⁎ Corresponding author at: University of Minho, Department of Economics and Economic Policies Research Unit (NIPE), Campus of Gualtar, 4710-057 Braga,Portugal. Tel.: +351 253601936, +44 2079557542; fax: +351 253676375, +44 2079556592.

E-mail addresses: [email protected] (T.A. Peltonen), [email protected], [email protected] (R.M. Sousa),[email protected] (I.S. Vansteenkiste).

1059-0560/$ – see front matter © 2012 Elsevier Inc. All rights reserved.doi:10.1016/j.iref.2012.01.006

Contents lists available at SciVerse ScienceDirect

International Review of Economics and Finance

j ourna l homepage: www.e lsev ie r .com/ locate / i re f

http://dx.doi.org/10.1016/j.iref.2012.01.006

mailto:[email protected]




http://dx.doi.org/10.1016/j.iref.2012.01.006

http://www.sciencedirect.com/science/journal/10590560

in the world economy and may play an important role in the resolution of global imbalances. In addition, since an increasinglylarge number of emerging market economies is becoming financially developed, their access to financial assets and the possibilityto extract equity from them has also risen, hence, amplifying the potential macroeconomic impact of domestic asset price move-ments.1 This may, in turn, generate a de-synchronization of the business cycle (Rafiq and Mallick, 2008; Mallick and Mohsin,2010) or negatively impinge on the nexus between monetary stability and financial stability (Granville and Mallick, 2009;Sousa, 2010c; Castro, 2011).

The importance of financial assets in emerging economies is inter alia reflected both in the rise in stock market capitalizationwhich currently represents more than 20% of the world's stock market capitalization,2 and as a share of its domestic size which isin many cases higher than for developed economies. For real estate assets, emerging markets have been recording an importantrise in homeownership rates, which are now estimated to be around 62% for Latin America and 55% for Emerging Asian urbanareas.

In this paper, we intend to quantify the wealth effects in emerging markets. Using a panel of 14 emerging economies, weshow that wealth effects are statistically significant and relatively large: a 10% rise in housing prices leads to an increase in pri-vate consumption of between 0.28% and 0.5%; an increase of 10% in stock prices is associated with a 0.26% to 0.30% increase inconsumption; and when money wealth rises by 10%, consumption increases by 0.43% to 0.54%. Additionally, the empirical find-ings suggest that: (i) stock market wealth and housing wealth effects are, in general, smaller for Latin American emerging mar-kets; and (ii) housing wealth effects have substantially increased in recent years for emerging Asian economies. Among Asiancountries, stock market wealth effects tend to be larger in the most developed financial markets (for instance, Singapore)while housing wealth effects are only statistically significant in the cases of Taiwan and Thailand. Finally, the results suggestthat consumption growth exhibits a substantial persistence and responds sluggishly to shocks. This may be an important reasonfor concern – particularly, in the case of a negative downturn – taking into account that these economies have often witnessedepisodes of economic, financial and currency crises.

Additionally, we show that, while housing wealth effects are typically stronger for countries with low level of financial devel-opment, stockmarket andmoney wealth effects are larger in the case of countries with high stockmarket capitalization. Similarly,housing market wealth effects seem to be quantitatively more important for countries with low income level.

All in all, these features highlight the relative importance of housing assets for households in countries with low level of finan-cial development or low income level and the sensitivity of consumption to financial wealth for countries with high stock marketcapitalization.

The rest of the paper is organized as follows. Section 2 reviews the existing literature of wealth effects on consumption.Section 3 presents the estimation methodology. Section 4 describes the data. Section 5 discusses the results. Section 6 providesthe sensitivity analysis. Finally, Section 7 concludes with the main findings and policy implications.

2. A brief review of the existing literature

An extensive empirical literature has tried to estimate the magnitude of the wealth effects on consumption. For the US, com-monly cited estimates of the marginal propensity to consume out of wealth are typically in the range of 4 to 7 cents increasedconsumer spending from a dollar increase in aggregate wealth (Gale and Sabelhaus, 1999). Mankiw and Zeldes (1991) showthat the consumption of stockholders is more volatile and more strongly correlated with stock market returns than for non-stockholders. Ludvigson and Steindel (1999) also identify a wealth effect on consumption but show that the effect is unstableover time.

Other studies find modest wealth effects. Cochrane (1994), Mayer and Simons (1994) and Lettau and Ludvigson (2001) showthat the overall impact on consumption is small and mainly transitory. Otoo (1999) shows that the correlation between stockprices and the consumer confidence level (either for stockholders or non-stockholders) does not change with the property ofstocks, that is, consumers use stocks mainly as a leading indicator of real economic activity. Poterba (2000) points out that theconcentrated nature of wealth, the desire to leave bequests, and precautionary motives in the consumer's behavior are importantdeterminants of the modest wealth effects. Caporale and Williams (2001) suggest a small marginal propensity to consume out ofwealth, but emphasize that the processes of financial liberalization/deregulation have strengthened wealth effects. Starr-McCluer(2002) suggests that concerns relative to trend inversions in stock prices can lead stockholders not to spend realized gains.

At the international level, the evidence is also quite diversified. In Japan, Horioka (1996) and Ogawa et al. (1996) suggestestimates for the marginal propensity to consume out of wealth of around between 1% and 4%, varying, considerably, withthe definitions of wealth and income. In Canada, Boone et al. (2001) suggest the existence of a wealth effect of the order of3% to 8%. In Australia, Tan and Voss (2003) estimate the marginal propensity to consume out of wealth in the range of 2, 4and 5 cents, respectively. For the UK, Fernandez-Corugedo et al. (2007) quantify the marginal propensity to consume out ofwealth at 5%.

1 For a revision of China's financial research, see, for instance, Chan et al. (2007). Similarly, Ergungor (2008) investigates the relationship between economicgrowth and the development of the structure of the financial system. In addition, Marques and Mallick (2010) look at the role played by exchange rate fluctua-tions and, in particular, Pan et al. (2007) and Lin (2012) assess their linkages with stock markets.

2 The stock market capitalization of the 14 emerging market economies studied in this paper accounted for 19.2% of the world's total stock market capitaliza-tion in September 2008. Only three years ago, their share of the world stock market capitalization was 12.3%. Harvey (1995) analyses the risk exposure of equityinvestments in emerging markets.

156 T.A. Peltonen et al. / International Review of Economics and Finance 24 (2012) 155–166

The studies mentioned above are based on the life cycle model and the permanent income hypothesis, as they suggest thatconsumers distribute increases in anticipated wealth over time and that the marginal propensity to consume out of wealth shouldbe the same no matter what asset categories are considered.

A companion literature has, however, argued that shocks to different forms of wealth (such as equity versus housing wealth)can elicit varying consumption responses. There is, in fact, a number of reasons for why the responsiveness of consumers to finan-cial asset shocks and housing asset shocks can be different: liquidity reasons (Pissarides, 1978), utility derived from the propertyright of an asset as housing services or bequest motives (Poterba, 2000; Bajari et al., 2005), different distributions of assets acrossincome groups, expected permanency of changes of different categories of assets, mismeasurement of wealth and ‘psychologicalfactors’.

Each of these motives suggests a distinction between the impact of financial wealth and housing wealth on consumption. Atthis level, the empirical findings are not yet conclusive, namely, in what concerns the significance of housing wealth effect. Elliott(1980) and Mehra (2001) find essentially that the wealth effect is independent of the category of asset considered. Thaler (1990)and Sheiner (1995) investigate the correlation between individual savings rates and changes in house prices and find a weak re-lation. In contrast, Skinner (1999) and Case et al. (2005) find evidence of a considerable housing wealth effect on consumption.For the Euro Area as a whole, Sousa (2010b) finds that housing wealth effects are virtually nil, while financial wealth effects arerelatively large.

Despite the wide evidence on wealth effects for developed countries, little attention has been given to emerging markets. Tothe best of our knowledge, the closest to our paper is the work of Funke (2004), who uses an annual panel of 16 emerging econ-omies for the period 1985–2000. By means of feasible generalized least squares (FGLS), the author estimates a panel equationwhich includes log changes in real GDP per capita, log changes in real money wealth per capita and log real equity returns asexplanatory variables for changes in real consumption per capita. Based on this model, the author finds a small, but statisticallysignificant stock market wealth effect: a 10% decline (increase) in stock prices is associated with a 0.2–0.4% decrease (increase)in private consumption over a 3-year period.

We improve and extend the existing literature in several directions. First, we look not only at the effects of money and stockmarket wealth but also at housing wealth effects, therefore capturing the impact of an important component of household wealth.Ignoring this may lead to omitted variable bias and, consequently, to biased estimates of the wealth effects.3 Second, we use quar-terly data and for a longer time period and are, therefore, able to obtain more precise estimates of the magnitude of the wealtheffects on consumption. Third, we consider labor income (rather than disposable income or real GDP per capita) for our panelof countries. Finally, we build upon modern estimation techniques — namely, by using a system GMM estimator to control forendogeneity.

All in all, these features may impact on the assessment of the linkages between consumption and wealth (Sousa, 2010a), inparticular, for emerging markets, and represent important gaps that we try to address in the present work.

3. Estimation methodology

The empirical model for the estimation of wealth effects on consumption can be summarized as follows:

Ci;t ¼ β0Ci;t−1 þ β1Yi;t þ β2Wi;t þ X′i;tβ3 þ νi þ εi;t i ¼ 1;…;N t ¼ 1;…; Ti ð1Þ

where Ci,t stands for the consumption level of country i at time t, Yi,t represents labor income,Wi,t is the asset wealth, Xi,t is a vectorof strictly exogenous covariates, the βs are parameters to estimate, νi is country-specific effects, and, εi, t is the error term.

The inclusion of a lag of consumption in Eq. (1) is aimed at capturing the presence of habit formation and simultaneously teststhe permanent income hypotheses. This is in line with the findings of Campbell and Mankiw (1989) and Lettau and Ludvigson(2001), who show that consumption growth is somewhat predictable by its lag.

When model (1) is estimated using ordinary least squares (OLS), substantial complications arise. In fact, in both the fixed andrandom effect settings, the lagged dependent variable is correlated with the error term, even if we assume that the disturbancesare not themselves autocorrelated. Moreover, the estimation of the dynamic panel defined above suffers from the Nickell (1981)bias, which disappears only if T tends to infinity.

In order to solve the problems referred above, Blundell and Bond (1998) develop a generalized method of moments (GMM)estimator that allows one to get rid of country specific effects or any time invariant country specific variable. Additionally, thisalso tackles the endogeneity issue that may be due to the correlation of the country specific effects and the independent variables.

Consequently, first differencing (1) removes νi, and produces an equation estimable by instrumental variables:

ΔCi;t ¼ β0ΔCi;t−1 þ β1ΔYi;t þ β2ΔWi;t þ ΔX′i;tβ3 þ Δεi;t i ¼ 1;…;N t ¼ 1;…; Ti ð2Þ

where Δ is the first difference operator, while the variables and parameters are defined as in Eq. (1). Following Holtz-Eakin et al.(1988), Blundell and Bond (1998) instrument the differenced pre-determined and endogenous variables with their available lagsin levels: levels of the dependent and endogenous variables, lagged two or more periods; levels of the pre-determined variables,lagged two or more periods. The exogenous variables can be used as their own instruments.

3 Interestingly, Guo and Huang (2010) assess the impact of speculative capital inflows on China's real estate and stock price indexes.

157T.A. Peltonen et al. / International Review of Economics and Finance 24 (2012) 155–166

A problem of the difference-GMM estimator is that lagged levels are weak instruments for first-differences if the series arevery persistent. Efficiency can be increased by adding the original equation in levels (1) to the equation in first-differences (2)forming a system with 2 equations. If the first-differences of explanatory variables are uncorrelated with the individual effects,both lagged values of the first-differences of the explanatory variables and of the dependent variable can be used as instrumentsin the equation in levels. In this case, the estimation combines the set of moment conditions available for the first-differencedequations with the additional moment conditions implied for the levels equation.

4. Data and summary statistics

The dataset consists of an unbalanced panel of 14 main emerging economies, 8 from emerging Asia (China, Hong Kong, Indo-nesia, Korea, Malaysia, Singapore, Taiwan, and Thailand), 4 from Latin America (Argentina, Brazil, Chile and Mexico), and 2 others(Russia and South Africa). These countries account for 45.9% of world GDP, and 97.3% of the emerging economies' GDP, measuredin 2007 PPP terms.

The sample period is 1975:1–2008:2 for which quarterly data are available. Data on housing and equity wealth is not availableon a broad basis for emerging economies. We, therefore, use stock market and house price indices as proxy variables for thesewealth components. This is, however, in line with the studies that have investigated the (in)direct impact of stock market priceson aggregate consumption (Romer, 1990) or the role played by housing prices (Miles, 1992; Aoki et al., 2002).

Housing price (residential property) indices are provided by CEIC (for the emerging Asian countries), the IMF (for the LatinAmerican countries), and Haver Analytics (for the other two economies). Stock price indices (composite indices) are obtainedfrom the Global Financial Database. Money wealth is proxied by broad money, M2, available from Haver Analytics, which, there-fore, also captures indirectly the role of monetary policy in emerging market economies (Mallick and Mohsin, 2007).4

With regard to the other series, real private consumption is provided by Haver Analytics, with the exception of China, HongKong, Indonesia, and Singapore for which the data comes from CEIC. We consider a measure of aggregate consumption andhence we cannot distinguish between non-durable and durable consumption.

Conventional theories look at the flow of non-durable and services consumption, since durable consumption can be thoughtof as a replacement and addition to the capital stock. Nevertheless, as Mehra (2001) points out, total consumption is the variableof interest when investigating the consumption-wealth channel. In particular, stock market crashes are more likely to lead to apostponement of durable consumption decisions, while the reduction of non-durable consumption might be of minor impor-tance (Romer, 1990). Furthermore, durable consumption goods are among the major entities on which resources raised bymortgage refinancing are spent on. In addition, although total consumption includes expenditures on housing services,Ludvigson and Steindel (1999) also show that the response of total consumption to a wealth shock is similar to those fornondurables consumption, as durables consumption typically represents a small share of total consumption. Finally, the evidencesupports the hypothesis that total consumption, wealth, and labor income are cointegrated.

Summing up, although data on nondurables consumption is not available, the use of total consumption allows us to obtainvery similar evidence on the long-term relationship and the short-term dynamics among consumption, wealth (and its majorcomponents) and labor income.

Data on income (either salary or wage income) is obtained from CEIC (for emerging Asian countries), and from HaverAnalytics (Latin American economies, Russia, and South Africa). The CPI price index is mainly from Haver Analytics, with the ex-ception of Argentina, Brazil, and Chile, for which the data source is the IMF. Finally, population statistics are obtained from theUN World Population Statistics database. Table A.1 in Appendix A provides a detailed description of the variables and datasources used in the analysis.

For the regression analysis, data are transformed in several ways. First, the wealth variables are deflated using CPI price index(all items), while the real private consumption data is deflated by the national authorities using National Accounts data. Second,we divide real money by the population in order to express it in per capita terms. Third, income corresponds to real wage or salaryprovided by National Statistics authorities, except for Argentina, China, Indonesia, Malaysia, Russia, and Thailand, where nominalwages (or salaries) are deflated using the CPI price index. Fourth, data on population and real private consumption for China areannual, and, therefore, we interpolate it using a cubic conversion method. Fifth, the semi-annual nominal wage data for HongKong is interpolated using the same method for the period 1990:1–1998:4.

Given that emerging markets have frequently been the stage for episodes of economic, financial and/or currency crises, wecreate a dummy for these events and define it as follows: it takes the value of 1 if either the change (year-on-year) of realGDP, real property price, or real equity price index is more than two times the country-specific standard deviation of the variable;and 0, otherwise. In addition, the quarters before and after the peak of crisis are also marked with 1. All other periods (normalperiods) are marked with 0.5 Therefore, this variable is aimed at capturing specific events, such as the 1997 East Asian financialcrisis, the 1998 Russian financial crisis and the 1998–1999 financial crisis, given that these episodes were typically characterizedby severe economic recession and substantial instability in financial markets. A similar approach is used by Peltonen et al.(forthcoming) while looking at the impact of crises on the long-run investment for emerging market economies and byPeltonen et al. (2011), while uncovering the effects of such events on the short-term dynamics of private investment in emerging

4 For Thailand, we use M3 instead of M2.5 Agnello and Schuknecht (2011) analyze episodes of booms and busts in real estate price in industrialized countries, and show that recent housing booms have

been persistent.


markets. Tables A.2 and A.3 also present a range of descriptive statistics. In this context, it is noteworthy to mention Table A.3,which reports that the average correlation coefficient between consumption and equity prices is around 0.4 over the entire sam-ple; for property prices, this coefficient is even higher, namely around 0.5.

Finally, we use the panel unit root tests of Levin et al. (2002), Im et al. (2003), and Maddala and Wu (1999) to assess the pres-ence of unit roots in the data. The empirical findings show that the log differences of all key variables are stationary following therejection of the null hypothesis of a unit root (see Table A.4).

5. Estimation results

We start by considering the GMM estimation of the dynamic panel defined in Eq. (2). We include the lag of consumption, thelabor income, the housing wealth (proxied by the property index), the financial wealth (proxied by the equity price index), andthe money wealth (proxied by the per capita broad money) among the set of endogenous variables.6

Despite the unpredictability of consumption (as advocated by Hall (1978)), other authors argue that consumption growth issignificantly affected by past information (Flavin, 1981; Campbell and Mankiw, 1989; Lettau and Ludvigson, 2001). This justifiesthe inclusion of lagged consumption growth among the set of explanatory variables. In addition, given that changes in somewealth components are typically correlated with changes in expected permanent income, the issue of endogeneity is important,in particular, regarding wealth (Calomiris et al., 2009). This explains the treatment given to wealth and labor income in oureconometric framework.

In the set of strictly exogenous variables, we include a linear trend and a quadratic trend to capture potential nonlinearities inthe behavior of consumption, a dummy variable for economic/financial crises, and a constant.

The moment conditions use the orthogonality conditions between the differenced errors and lagged values of the dependentvariable, which are particularly important when the first-differences of the series are very persistent (Blundell and Bond, 1998).In addition, we assess the autocorrelation of the error terms, by using two diagnostic tests for first-order and second-order serialcorrelations in the disturbances. One should reject the null hypothesis of the absence of first-order serial correlation and not rejectthe null hypothesis of the absence of second order-serial correlation. The tests confirm these hypotheses. The estimation resultsare summarized in Table 1. Columns 1 to 4 display the point estimates using the entire sample: in Column 1, we include all com-ponents of wealth in the estimation; in Columns 2, 3 and 4, we consider only one component of wealth (respectively, housing,financial, and money wealth). Columns 5 and 6 provide a comparison between the wealth effects in Asian and Latin Americancountries in the period 2001:1–2006:4 (where all variables are available for both regions). Columns 7 and 8 allow us to analyzethe wealth effects for Asian countries in two sub-samples: 1975:1–1999:4 and 2000:1–2008:4. In Column 9, we assess the exis-tence of a “ratchet” effect and, therefore, separate between positive and negative changes in financial wealth and housing wealth.The major reason consumer's reaction to a fall in wealth may be greater than to an increase is the assumption of diminishingmarginal utility of wealth. Under this assumption, investor preferences can be described by convex utility functions. Such utilityfunctions reflect risk aversion implying that consumer values an increase in wealth less highly than an equivalent decrease.

In all specifications, the lag of consumption is statistically significant, therefore, reflecting the strong persistence of consump-tion growth and its sluggish response to shocks.7 Additionally, the different components of wealth are statistically significantwhen all observations are included in the estimation (Columns 1 to 4). In fact, the empirical evidence suggests that: when hous-ing wealth rises by 10%, private consumption increases by 0.28%–0.50%; a 10% increase in stock market wealth leads to an increasein consumption of 0.26%–0.30%; and a rise of 10% in money wealth is associated with a 0.43%–0.54% increase in consumption.Additionally, there is some evidence of non-linearity in the behavior of consumption.

When we focus on the period 2001:1–2006:4, the results suggest that whilst wealth effects are still statistically significant forAsian countries (Column 5), the same does not hold for Latin American countries (Column 6). This result can be related with thework of Hargis (2000), who shows that international cross-listings can strongly impact on stock market development in emergingmarket economies. However, it is important to note that this result may be associated with the smaller number of data pointsavailable for these emerging markets. Columns 7 and 8 show that housing wealth effects have been particularly strong forAsian countries in the second sub-sample (2000:1–2008:2). In contrast, financial wealth effects were larger in the first sampleperiod (1975:1–1999:4). The rise is housing wealth effects is in accordance with the increased financial development in the re-gion, which allows easier access to equity from housing than before. Finally, Column 9 suggests that consumption reacts asym-metrically to positive versus negative shocks in housing and/or financial wealth. However, neither of the variables included inthe regression to capture the “ratchet” effect are statistically significant.

Table 2 provides a summary of the country level evidence, that is, we estimate by GMM and for each country the dynamicmodel defined in Eq. (2). We focus on the analysis of 6 emerging Asian economies (China, Hong Kong, Korea, Singapore, Taiwanand Thailand) for which the number of observations is the largest. The availability of such data for these countries allows us: (i) toprovide country-based evidence on the wealth effects on consumption without costs in terms of the efficiency of the statisticalinference which would, otherwise, suffer from a small number of degrees of freedom; and (ii) to assess how those effects haveevolved over time and, thereby, to capture the key wealth components that drove consumption over different sub-sampleperiods.

6 All variables are expressed in log differences.7 The persistence of consumption growth may be due to household inattention, evaluation of household finances at periodic intervals, adjustment costs to

change consumption, and habit formation.


Consistently with the previous findings, the lag of consumption is always statistically significant and large in magnitude, re-vealing a strong persistence in consumption growth. In addition, stock market wealth effects are also statistically significant forthe majority of the countries, ranging between 0.18% (in the case of Thailand) and 0.65% (in the case of Singapore) of increasein consumption following a 10% rise in stock prices, that is, they tend to be larger in the most developed financial markets.

Table 1Dynamic GMM — panel regression results.

All All All All EM Asia Lat Am EM Asia EM Asia All

1975–2008 1975–2008 1975–2008 1975–2008 2001–2006 2001–2006 1975–99 2000–2008 1975–2008

ΔCons 0.710*** 0.731*** 0.7945*** 0.760*** 0.665*** 0.519*** 0.635*** 0.683*** 0.689***[0.018] [0.041] [0.038] [0.030] [0.039] [0.129] [0.100] [0.051] [0.026]

ΔWage 0.039 0.052 0.034** 0.025 0.001 0.003 0.141 0.040 0.057**[0.026] [0.045] [0.017] [0.018] [0.025] [0.034] [0.109] [0.053] [0.029]

ΔProp. price 0.028*** 0.050*** 0.025** 0.100 0.027** 0.056*** 0.026[0.009] [0.014] [0.012] [0.125] [0.011] [0.018] [0.017]

ΔEquity price 0.030*** 0.026*** 0.030*** 0.009 0.032*** 0.026*** 0.023***[0.003] [0.005] [0.008] [0.006] [0.009] [0.005] [0.007]

ΔMoney 0.043*** 0.054*** 0.093*** 0.034 0.037 0.051 0.052***[0.014] [0.012] [0.031] [0.024] [0.045] [0.054] [0.013]

ΔProp. price (−) 0.011[0.014]

ΔEquity price (−) 0.020[0.016]

Linear trend 0.022* 0.002 0.009 0.008 −0.207 1.837*** −0.028 −0.380* 0.023*[0.012] [0.018] [0.010] [0.008] [0.879] [0.427] [0.038] [0.228] [0.013]

Quadratic trend −0.000** 0.000 −0.000 0.000 0.001 −0.008*** 0.000 0.001 −0.000*[0.000] [0.000] [0.000] [0.000] [0.004] [0.002] [0.000] [0.001] [0.000]

Crisis 0.041 −0.095 −0.136 −1.011** 0.446 0.351 −0.813 0.394 0.125[0.323] [0.448] [0.415] [0.422] [0.305] [0.821] [0.837] [0.443] [0.395]

Constant 0.137 1.022 0.608 0.467* 14.513 −109.714*** 2.416 25.119* 0.412[0.373] [0.827] [0.406] [0.260] [51.844] [23.937] [1.662] [13.524] [0.361]

Observations 650 650 870 864 189 83 171 194 650Number of countries 14 14 14 14 8 4 6 6 14AR1 p-value 0.01 0.02 0.01 0.01 0.04 0.20 0.07 0.05 0.01AR2 p-value 0.42 0.35 0.24 0.18 0.98 0.41 0.21 0.82 0.44

Note: Estimation method Blundell and Bond (1998). Heteroscedasticity and serial correlation robust standard errors in brackets.*significant at 10%; **significant at 5%; ***significant at 1%.

Table 2Dynamic GMM — Asian country-by-country regression results.

China Hong Kong Korea Singapore Taiwan Thailand

ΔCons 0.601** 0.662*** 0.567*** 0.838*** 0.659*** 0.680***[0.248] [0.117] [0.158] [0.196] [0.122] [0.105]

ΔWage −0.014 −0.197 0.288* 0.050 −0.269 −0.017[0.057] [0.146] [0.157] [0.273] [0.243] [0.111]

ΔProp. price 0.059 0.004 −0.027 0.009 0.084* 0.091***[0.050] [0.022] [0.056] [0.053] [0.048] [0.029]

ΔEquity price 0.010 0.030*** 0.009 0.065* 0.026*** 0.018***[0.008] [0.011] [0.008] [0.036] [0.007] [0.006]

ΔMoney 0.069 0.131** −0.048 0.015 −0.201 −0.086[0.055] [0.054] [0.095] [0.073] [0.129] [0.129]

Linear trend −1.1969** 0.050 0.020 0.706 −0.368 0.489[0.533] [0.066] [0.089] [0.769] [0.228] [0.577]

Quadratic trend 0.005** −0.000 −0.000 −0.003 0.001 −0.002[0.002] [0.000] [0.001] [0.003] [0.001] [0.003]

Crisis 0.612*** −2.226** −2.270* 0.144 0.314 3.629***[0.227] [0.985] [1.216] [3.135] [0.668] [1.059]

Constant 70.171** −0.940 1.206 −36.803 27.074* −24.386[33.493] [3.030] [5.188] [43.045] [14.933] [33.020]

Observations 36 101 84 42 63 35R-squared 0.95 0.78 0.8 0.84 0.66 0.82J-stat 5.81 7.03 8.56 7.5 5.66 7.86p-value 0.67 0.53 0.38 0.48 0.69 0.45

Note: Heteroscedasticity and serial correlation robust standard errors in brackets.*significant at 10%; **significant at 5%; ***significant at 1%.


In contrast to the previous findings, there is only evidence of a statistically significant housing wealth effect in the case ofTaiwan and Thailand. This is somehow surprising as one would expect this effect to be also statistically significant and strongerfor Hong Kong and Singapore, that is, where the housingmarket boom has beenmore dramatic. In the case of money wealth, theempirical findings suggest that this category of wealth is statistically significant only for Hong Kong, where the associatedwealth effect is quite large (a 10% increase in money wealth leads to an increase in consumption of around 1.31%). Notably,for China, Hong Kong, Korea and Thailand, the dummy variable aimed at capturing episodes of economic/financial crises isstatistically significant.

6. Sensitivity analysis

In this section, we present some sensitivity analysis. More specifically, we estimate wealth effects in emerging market econ-omies using different econometric methodologies and assess the importance of (i) the financial development and (ii) the incomelevel in explaining the differences in wealth effects on consumption across different groups of countries.

6.1. Different econometric methodologies

We start by considering different econometric techniques, namely: (i) the pooled OLS (dynamic OLS) estimator; (ii) the pooledOLS estimator with time effects; (iii) the pooled OLS with both time and country effects; (iv) the fixed effects estimator; (v) the ran-dom effects estimator; (vi) the IV/GMM estimator; (vii) the bias corrected Least Squares with Dummy Variables (LSDV) dynamicpanel data estimator;8 and (viii) the Blundell and Bond (1998) estimator, for comparison (Table 3). In the cases of the IV/GMM,the bias corrected LSDV dynamic panel and the Blundell and Bond (1998) estimators, a linear trend, a quadratic trend, the crisisdummy and a constant are included in the set of exogenous variables, while the remaining variables are treated as endogenous.

The results corroborate the previous findings where we used the Blundell–Bond estimator. First, all econometric methodolo-gies show that the lag of consumption growth is statistically significant, thereby supporting the estimation of a dynamic model.Moreover, the associated coefficient is large in magnitude, which suggests that consumption exhibits strong inertia. It also impliesthat, while the short-run wealth effects are important, the long-run wealth effects are quantitatively relevant. Second, the incomeelasticity is statistically significant for a reasonable number of specifications, although its magnitude is relatively small. Third, allwealth components have a statistically significant impact on consumption in the eight different econometric techniques and theirestimated effects are quantitatively similar. In particular: (i) a 10% increase in housing wealth leads to an increase in consumptionof between 0.04% and 0.44%; (ii) when stock market wealth rises by 10%, consumption is boosted by between 0.24% and 0.30%;and (iii) a 10% increase in money wealth has a positive impact on consumption of between 0.16% and 0.49%. Finally, we find thatconsumption growth displays a relevant non-linear behavior.

6.2. Financial development

We now split the sample in two sub-samples based on the level of financial development, which is proxied by the ratio of stockmarket capitalization to GDP. This, in turn, is measured at the end of 2008 and countries are ranked according to the ratio.

We recognize that the ratio of stock market capitalization to GDP is just an approximation of the level of financial develop-ment. However, we note that our aim is to show that results are robust to the choice of the estimation sample.

With this caveat in mind, the countries with high stock market capitalization to GDP ratio are as follows: Hong Kong, Taiwan,Singapore, Korea, Malaysia, China and South Africa.

The results are summarized in Table 4. The left panel provides information for the sub-sample of countries with high level offinancial development. Columns 1 to 4 display the point estimates using the entire sample (1975–2008): in Column 1, we consid-er all components of wealth in the model specification; in Columns 2, 3 and 4, we consider only one component of wealth at time(respectively, housing, financial, and money wealth). In Column 5, we assess the existence of a “ratchet” effect, thereby, disentan-gling between positive and negative changes in financial wealth and housing wealth. Columns 6 to 10 display a summary of theevidence for the sub-sample of countries with low level of financial development.

The empirical findings suggest that the lag of consumption is statistically significant for both sub-samples, therefore, reflectingthe strong persistence of consumption growth (which may be due to the existence of habit-formation utility preferences or im-portant adjustment costs) and its sluggish response to shocks.

In addition, all components of wealth are statistically significant. However, there are relevant differences that are worthemphasizing. In particular, while housing wealth effects are larger for countries with low stock market capitalization, stock mar-ket and money wealth effects are stronger in countries with high financial development (Columns 1 and 6). In fact, the empiricalevidence suggests that: (i) when housing wealth rises by 10%, private consumption increases by 0.49% in countries with lowfinancial development (against 0.36% in countries with high financial development); and (ii) a 10% increase in stock market ormoney wealth leads to an increase in consumption of, respectively, 0.28% or 0.44%, for countries with high stock market capital-ization (which compares to, respectively, 0.25% or 0.21%, in the case of countries with low stock market capitalization). This pieceof evidence emphasizes the importance of housing as the major asset of households in countries with low level of financial

8 See, for instance, Kiviet (1995) and Bruno (2005).


development, as well as the sensitivity of consumption to changes in financial wealth for countries with high stock marketcapitalization.

Additionally, there is evidence of non-linearity in the behavior of consumption, especially, for countries with low financial de-velopment, where both the time and the quadratic trends are statistically significant. Finally, the crisis dummy seems to have astatistically significant and negative impact on consumption when the level of financial development is high. This implies that

Table 3Wealth effects in emerging market economies — different econometric methodologies.

OLS pooled OLS time effects OLS time and country effects Fixed effects Random effects IV GMM Bias corrected LSDV Blundell Bond

ΔCons 0.723*** 0.714*** 0.672*** 0.699*** 0.723*** 0.737*** 0.699*** 0.710***[0.038] [0.038] [0.040] [0.024] [0.023] [0.178] [0.018] [0.018]

ΔWage 0.036** 0.044** 0.062** 0.042** 0.036** 0.025 0.0416 0.039[0.018] [0.019] [0.024] [0.017] [0.015] [0.047] [0.026] [0.026]

ΔProp. price 0.031*** 0.033*** 0.044*** 0.041*** 0.031*** 0.004 0.041*** 0.028***[0.010] [0.009] [0.011] [0.010] [0.009] [0.027] [0.009] [0.009]

ΔEquity price 0.029*** 0.025*** 0.026*** 0.030*** 0.029*** 0.024*** 0.030*** 0.030***[0.003] [0.004] [0.005] [0.003] [0.003] [0.006] [0.003] [0.003]

ΔMoney 0.033** 0.029** 0.016 0.023* 0.033*** 0.049** 0.023*** 0.043***[0.014] [0.014] [0.019] [0.012] [0.010] [0.024] [0.014] [0.014]

Linear trend 0.021 −0.005 −0.015 0.018 0.021 0.017 0.0179* 0.022*[0.014] [0.020] [0.024] [0.014] [0.013] [0.014] [0.012] [0.012]

Quadratic trend −0.000* 0.000 0.000 −0.000* −0.000* −0.000 −0.0002** −0.000**[0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000]

Crisis 0.155 0.488 0.594 0.163 0.155 −0.476 0.1615 0.041[0.397] [0.365] [0.416] [0.324] [0.287] [0.569] [0.323] [0.323]

Constant 0.151 0.702 3.207*** 0.823 0.151 0.129 0.6989 0.137[0.539] [0.459] [1.090] [0.550] [0.490] [0.782] [0.373] [0.373]

Observations 650 650 650 650 650 631 650 650Number of countries 14 14 14 14 14 14 14 14R-squared 0.80 0.85 0.86 0.77 0.80 0.79Hansen J-stat 6.43AR1 p-value 0.01AR2 p-value 0.42

*significant at 10%; **significant at 5%; ***significant at 1%.

Table 4Dynamic GMM — high versus low financial development.

High financial development Low financial development

ΔCons 0.708*** 0.719*** 0.756*** 0.755*** 0.694*** 0.726*** 0.775*** 0.755*** 0.652*** 0.734***[0.021] [0.047] [0.054] [0.053] [0.031] [0.062] [0.071] [0.071] [0.115] [0.042]

ΔWage 0.069 0.087 0.088** 0.078* 0.085* 0.004 −0.009 0.022 −0.002 0.008[0.042] [0.078] [0.030] [0.046] [0.050] [0.020] [0.024] [0.015] [0.019] [0.016]

ΔProp. price 0.036*** 0.054*** 0.044*** 0.049*** 0.053*** 0.041[0.010] [0.011] [0.010] [0.016] [0.008] [0.030]

ΔEquity price 0.028*** 0.026*** 0.020*** 0.025** 0.028*** 0.009[0.003] [0.007] [0.008] [0.011] [0.005] [0.012]

ΔMoney 0.044*** 0.052*** 0.045*** 0.021* 0.086*** 0.035***[0.020] [0.020] [0.021] [0.012] [0.025] [0.013]

ΔProp. price (−) −0.018 0.029[0.016] [0.062]

ΔEquity price (−) 0.016 0.058***[0.016] [0.014]

Linear trend 0.016 0.003 0.018 0.014 0.014 0.355 0.774* 0.194** 0.221** −0.118[0.014] [0.023] [0.012] [0.015] [0.016] [0.387] [0.429] [0.097] [0.103] [0.441]

Quadratic trend −0.000 −0.000 −0.000 −0.000 −0.000 −0.002 −0.003* −0.001** −0.001* 0.000[0.000] [0.000] [0.000] [0.000] [0.000] [0.002] [0.002] [0.000] [0.000] [0.002]

Crisis −0.101 −0.432 −0.666* −1.520*** −0.207 0.648 10.377 0.058 −0.899 10.712[0.305] [0.527] [0.377] [0.544] [0.406] [0.845] [1.079] [1.004] [0.958] [0.844]

Constant 0.320 10..759 0.353 0.122 0.499 193.681 −44.972* −10.197* −12.033** 99.700[0.376] [0.954] [0.287] [0.520] [0.442] [23.225] [24.937] [5.881] [5.912] [26.770]

Observations 460 460 589 589 460 190 190 281 275 190Number of countries 7 7 7 7 7 7 7 7 7 7AR1 p-value 0.03 0.04 0.02 0.02 0.03 0.12 0.15 0.10 0.09 0.10AR2 p-value 0.56 0.46 0.39 0.30 0.57 0.63 0.77 0.74 0.55 0.28



these countries are more vulnerable economic and financial crises, as capital markets are an important funding source for con-sumption expenditure.

6.3. Income level

Another way of splitting the sample is based on the GDP per capita (i.e., the income level). Again, the main empirical findingsare corroborated and the results are available upon request. In this case, the countries with high GDP per capita include: HongKong, Korea, Singapore and Taiwan.

A summary of the findings can be found in Table 5. As in the case of Table 4, the left panel (Columns 1 to 5) displays the resultsfor the sub-sample of countries with high income level, while the right panel (Columns 6 to 10) refers to countries with low in-come level.

It can be seen that consumption exhibits strong persistence, but this is generally larger for countries with low income level,which suggests that adjustment costs are typically larger for this sub-sample. In line with this piece of evidence, we also findthat, while wage income typically has a positive and significant effect on consumption in the case of countries with high incomelevel, it rarely is significant for countries with low income level.

Additionally, we show that: (i) a 10% increase in housing wealth leads to an increase in consumption of 0.31% in the case ofcountries with high income level (which compares to 0.48%, for countries with low income level); (ii) stock market wealth effectsare quantitatively similar for both groups of countries (i.e., when stock market wealth rises by 10%, private consumption increasesby 0.27%); and (iii) money wealth effects do not seem to be statistically significant. These results highlight the role played byhousing in countries with low income level.

The empirical findings also corroborate the existence of a statistically significant and negative effect of crisis episodes onconsumption, in particular, for countries with high income level.

Finally, consumption seems to react asymmetrically to positive and negative changes in financial and housing wealth in thecase of poor countries. “Ratchet” effects are statistically significant in this sub-sample, which reflects that liquidity and credit con-straints may be binding when there are sharp falls in wealth.

7. Conclusion

In this paper, we analyze the relationship between consumption and several wealth components for a panel of 14 main emerg-ing economies. We estimate the magnitude of the effects of stock market wealth, housing wealth, and money wealth on privateconsumption using modern panel data econometric techniques.

Table 5Dynamic GMM — high versus low income level.

High income level Low income level

ΔCons 0.658*** 0.616*** 0.687*** 0.707*** 0.664*** 0.716*** 0.772*** 0.815*** 0.751*** 0.684***[0.035] [0.038] [0.058] [0.053] [0.048] [0.034] [0.054] [0.038] [0.059] [0.028]

ΔWage 0.156*** 0.253*** 0.109** 0.144*** 0.166*** 0.010 −0.011 0.028** 0.012 0.018[0.042] [0.031] [0.044] [0.038] [0.046] [0.013] [0.025] [0.011] [0.012] [0.013]

ΔProp. price 0.031*** 0.055*** 0.040*** 0.048*** 0.058*** 0.065***[0.009] [0.010] [0.006] [0.012] [0.009] [0.017]

ΔEquity price 0.027*** 0.027*** 0.015*** 0.027*** 0.026*** 0.020***[0.004] [0.005] [0.005] [0.006] [0.007] [0.008]

ΔMoney 0.044 0.105*** 0.052* 0.019 0.056*** 0.022[0.032] [0.028] [0.028] [0.012] [0.018] [0.017]

ΔProp. price (−) −0.024 0.043**[0.019] [0.018]

ΔEquity price (−) 0.025* 0.035**[0.015] [0.015]

Linear trend 0.008 −0.065 0.027 0.050*** 0.0191 0.008 −0.009 0.016 −0.004 0.007[0.041] [0.061] [0.028] [0.013] [0.047] [0.012] [0.010] [0.012] [0.008] [0.015]

Quadratic trend −0.000 0.000 −0.000 −0.000*** −0.000 −0.000 0.000 −0.000 0.000 −0.000[0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000]

Crisis −0.678*** −1.109* −0.752 −1.995*** −0.700** 0.489 0.709* 0.137 −0.635 1.144**[0.256] [0.572] [0.611] [0.678] [0.288] [0.426] [0.412] [0.614] [0.625] [0.542]

Constant 0.859 4.487** 0.576 −1.982*** 0.424 0.300 0.603* −0.133 0.588* 0.931***[1.117] [2.221] [0.757] [0.461] [1.163] [0.344] [0.336] [0.246] [0.325] [0.344]

Observations 269 269 358 358 269 381 381 512 506 381Number of countries 4 4 4 4 4 10 10 10 10 10AR1 p-value 0.06 0.07 0.07 0.07 0.06 0.08 0.10 0.04 0.03 0.07AR2 p-value 0.60 0.42 0.75 0.45 0.66 0.70 0.82 0.46 0.45 0.54



Drawing upon quarterly data, we show that wealth effects are statistically significant and relatively large: a 10% rise in housingprices leads to an increase in private consumption of between 0.28% and 0.5%; an increase of 10% in stock prices is associated witha 0.26% to 0.30% increase in consumption; and when money wealth rises by 10%, consumption increases by 0.43% to 0.54%. Ad-ditionally, the empirical findings suggest that: (i) stock market and housing wealth effects are, in general, smaller for Latin Amer-ican emerging markets; and (ii) housing wealth effects have substantially increased for Asian emerging economies in recentyears. These results are robust to the use of different econometric methodologies.

Among Asian countries, stock market wealth effects tend to be larger in the most developed financial markets (for instance,Singapore). Moreover, housing wealth effects are particularly important in Taiwan and Thailand.

Splitting the sample according to the level of financial development and the level of income, we also find that: (i) housingwealth effects are particularly relevant for countries with low stock market capitalization; (ii) financial and money wealth effectsare quantitatively larger in the case of countries with high level of financial development; and (iii) consumption is more sensitiveto changes in housing wealth when the income level is low.

Finally, our results suggest that consumption growth exhibits a substantial persistence and responds sluggishly to shocks. Thismay be an important reason for concern – particularly, in case of a negative downturn –, given that these economies have oftenwitnessed episodes of economic, financial and currency crises.

Acknowledgments

The authors would like to thank the Editor, Professor Hamid Beladi, two anonymous reviewers and seminar participants at theECB for useful comments and suggestions.

Table A.1Data sources.

Country Consumption Income Property index Equity index Money CPI index Population

Argentina Haver: privateconsumption(real GDP)

Haver: total salary index(deflated by CPI index)

IMF: houseprice index

GFD: BuenosAires SE GeneralIndex

Haver:M2

IMF: nationalCPI

UN World PopulationProspects (interpolated)

Brazil Haver: privateconsumption(real GDP)

Haver: real averageearnings of employedpersons


GFD: BrazilBolsa de Valoresde São Paulo

Haver:M2

IMF: nationalCPI


Chile Haver: privateconsumption(real GDP)

Haver: real hourly wageindex


GFD: SantiagoSE IndiceGeneral dePrecios deAcciones

Haver:M2

IMF: nationalCPI


China CEIC: privateconsumptiondeflated with theGDP deflator,interpolated

CEIC: average earningper employee(deflated by CPI index)

CEIC: residentialproperty price index

GFD: ShanghaiSE composite

Haver:M2

Haver:nationalaverage CPIindex


HK CEIC: privateconsumption(real GDP)

CEIC: real wage index(missing obs.interpolated)

CEIC: domesticpremise propertyprice index

GFD: Hang Sengcomposite index

Haver:M2

Haver: CPIcompositeindex


Indonesia CEIC: privateconsumption(real GDP)

Haver: average ofmining, hotel andmanufacturing wages(deflated by CPI index)

CEIC: residentialpropertyprice index(either 12, 13 or14 cities)

GFD: Jakart SEcomposite index

Haver:M2

Haver: CPItotal


Korea Haver: privateconsumption(real GDP)

CEIC: monthly earningsall industries(deflated by CPI index)

CEIC: total housingprice index

GFD: Koreastock priceindex (KOSPI)

Haver:M2

Haver: CPI allitems


Malaysia Haver: privateconsumption(real GDP)

CEIC: monthly earningsmanufacturing(deflated by CPI index)

CEIC: houseprice index

GFD: MalaysiaKLSE composite

Haver:M2

Haver: CPI allitems


Mexico Haver: privateconsumption(real GDP)

Haver: realremunerations inmanufacturing


GFD: Mexico SEIndice de Preciosy Cotizaciones

Haver:M2

Haver: CPI allitems


Appendix A. Data and summary statistics


Table A.2Sample period and number of observations per country.

Country Obs Sample

Argentina 17 4/2002–4/2006Brazil 18 3/2002–4/2006Chile 24 1/2001–4/2006China 36 1/1999–4/2007HK 101 1/1983–1/2008Indonesia 26 4/2001–1/2008Korea 84 1/1987–4/2007Malaysia 28 1/2001–4/2007Mexico 24 1/2001–4/2006Russia 21 1/2003–1/2008Singapore 42 1/1997–1/2008South Africa 127 2/1976–4/2007Taiwan 63 3/1992–1/2008Thailand 39 1/1998–3/2007

Table A.1 (continued)

Country Consumption Income Property index Equity index Money CPI index Population

Russia Haver: privateconsumption(real GDP)

Haver: nominal accruedmonthly wages(deflated by CPI index)

Haver: pricesfor existing homes

GFD: MoscowTimes RoubleIndex

Haver:M2

Haver: CPI allitems


Singapore CEIC: privateconsumption(real GDP)

CEIC: average realmonthly earnings, total

CEIC: privateresidential propertyprice index

GFD: FTSEStraits Timesindex

Haver:M2

Haver: CPI allitems


South Africa Haver: privateconsumption(real GDP)

Haver: realremuneration perworker

Haver: ABSA houseprice index

GFD: FTSE JSE allshare index

Haver:M2

Haver: CPI(metropolitanárea) all items


Taiwan Haver: privateconsumption(real GDP)

CEIC: average realmonthly earnings:industry and service

CEIC: Sinyiresidential propertyprice index

GFD: Taiwan SEcapitalizationweighted index

Haver:M2

Haver: CPI allitems

CEIC: population

Thailand Haver: privateconsumption(real GDP)

CEIC: averagemonthly wage(deflated by CPI index)

CEIC: average ofhousing price indicesof single detachedhouse and townhouse including land

GFD: ThailandSET generalindex

Haver:M3

Haver: CPI allcommodities


Table A.3Correlation coefficients.

Variable Consumption Wage Property Equity Money

Consumption 1.000Wage 0.400 1.0000Property 0.508 0.193 1.000Equity 0.386 0.056 0.238 1.000Money 0.408 0.385 0.243 0.207 1.000

Table A.4Panel unit root test results.

Test Consumption Income Property index Equity index Broad money

Levin, Lin and Chu t-stat −2.440 −2.795 −5.914 −2.923 −2.148p-value 0.007 0.007 0.000 0.002 0.016Im, Pesaran and Shin W-stat −5.083 −8.8207 −5.760 −6.670 −4.957p-value 0.000 0.000 0.000 0.000 0.000ADF — Fisher Chi-square 83.352 125.711 85.312 106.938 85.997p-value 0.000 0.000 0.000 0.000 0.000PP — Fisher Chi-square 73.118 93.240 89.924 85.108 67.362p-value 0.000 0.000 0.000 0.000 0.000

Note: All series are in log differences.


References

Agnello, L., & Schuknecht, L. (2011). Booms and busts in housing markets: determinants and implications. Journal of Housing Economics, 20(3), 171–190.Aoki, K., Proudman, J., & Vlieghe, G. (2002). Houses as collateral: has the link between house prices and consumption in the U.K. changed? Economic Policy Review,

8(1), 163–177.Bajari, P., Benkard, C., & Krainer, J. (2005). House prices and consumer welfare. Journal of Urban Economics, 58(3), 474–487.Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87, 115–143.Boone, L., Girouard, N., & Wanner, I. (2001). Financial liberalization, wealth and consumption. Organization for Economic Co-Operation and DevelopmentOECD

Working Paper No. 308.Bruno, G. S. F. (2005). Estimation, inference and Monte Carlo analysis in dynamic panel-data models with a small number of individuals. The Stata Journal, 5(4),

473–500.Calomiris, C., Longhofer, S. D., & Miles, W. (2009). The (mythical?) housing wealth effect. National Bureau of Economic ResearchNBER Working Paper No. 15075.Campbell, J., & Mankiw, G. (1989). Consumption, income and interest rates: re-interpreting the time series evidence. In S. Fischer, & O. Blanchard (Eds.), NBER

Macroeconomics Annual. MIT Press: Cambridge.Caporale, G., &Williams, G. (2001). Monetary policy and financial liberalization: the case of United Kingdom consumption. Journal of Macroeconomics, 23(2), 177–197.Case, K., Shiller, R., & Quigley, J. (2005). Comparing wealth effects: the stock market versus the housing market. Advances in Macroeconomics, 5(1), 1–32.Castro, V. (2011). Can central banks' monetary policy be described by a linear (augmented) or by a nonlinear rule? Journal of Financial Stability, 7(4), 228–246.Chan, K. C., Fung, H. G., & Thapa, S. (2007). China financial research: a review and synthesis. International Review of Economics and Finance, 16(3), 416–428.Cochrane, J. (1994). Permanent and transitory components of GDP and stock prices. Quarterly Journal of Economics, 109(1), 241–265.Elliott, J. (1980). Wealth and wealth proxies in a permanent income model. Quarterly Journal of Economics, 95(3), 509–535.Ergungor, O. E. (2008). Financial system structure and economic growth: structure matters. International Review of Economics and Finance, 17(2), 292–305.Fernandez-Corugedo, E., Price, S., & Blake, A. (2007). The dynamics of consumers' expenditure: the UK consumption ECM redux. Economic Modelling, 24, 453–469.Flavin, M. (1981). The adjustment of consumption to changing expectations about future income. Journal of Political Economy, 89(5), 974–1009.Funke, N. (2004). Is there a stock market wealth effect in emerging markets? Economics Letters, 83(3), 417–421.Gale, W., & Sabelhaus, J. (1999). Perspectives on the household savings rate. Brookings Papers on Economic Activity, 1, 1181–1224.Granville, B., & Mallick, S. K. (2009). Monetary and financial stability in the Euro Area: pro-cyclicality versus trade-off. Journal of International Financial Markets

Institutions & Money, 19(4), 662–674.Guo, F., & Huang, Y. S. (2010). Does “hot money” drive China's real estate and stock markets? International Review of Economics and Finance, 19(3), 452–466.Hall, R. (1978). Stochastic implications of the life cycle-permanent income hypothesis: theory and evidence. Journal of Political Economy, 86(6), 971–987.Hargis, K. (2000). International cross-listing and stock market development in emerging economies. International Review of Economics and Finance, 9, 101–122.Harvey, C. R. (1995). The risk exposure of emerging equity markets. World Bank Economic Review, 9(1), 19–50.Holtz-Eakin, D., Newey, W., & Rosen, H. S. (1988). Estimating vector autoregressions with panel data. Econometrica, 56, 1371–1395.Horioka, C. (1996). Capital gains in Japan: their magnitude and impact on consumption. The Economic Journal, 106(436), 560–577.Im, K. S., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics, 115, 53–74.Kiviet, J. F. (1995). On bias, inconsistency and efficiency of various estimators in dynamic panel data models. Journal of Econometrics, 68, 53–78.Lettau, M., & Ludvigson, S. (2001). Consumption, aggregate wealth and expected stock returns. Journal of Finance, 56(9), 815–849.Levin, A., Lin, C. F., & Chu, C. (2002). Unit root tests in panel data: asymptotic and finite-sample properties. Journal of Econometrics, 108, 1–24.Lin, C. -H. (2012). The comovement between exchange rates and stock prices in the Asian emergingmarkets. International Review of Economics and Finance, 22(1), 161–172.Ludvigson, S., & Steindel, C. (1999). How important is the stock market effect on consumption. FRBNY Economic Policy Review, 29–51.Maddala, G. S., & Wu, S. (1999). A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and Statistics, 61, 631–652.Mallick, S. K. (2008). Income distribution and consumption deprivation: an analytical link. Journal of Income Distribution, 17(2), 25–36.Mallick, S. K., & Mohsin, M. (2007). Monetary policy in high inflation open economies: evidence from Israel and Turkey. International Journal of Finance and Economics,

12(4), 405–415.Mallick, S. K., & Mohsin, M. (2010). On the real effects of inflation in open economies: theory and empirics. Empirical Economics, 39(3), 643–673.Mankiw, N., & Zeldes, S. (1991). The consumption of stockholders and nonstockholders. Journal of Financial Economics, 29(1), 97–112.Marques, H., & Mallick, S. K. (2010). Data frequency and exchange rate pass-through: evidence from India's exports. International Review of Economics and Finance,

19(1), 13–22.Mayer, C., & Simons, K. (1994). Reverse mortgages and the equity of housing wealth. Journal of the American Real Estate and Urban Economics, 22(2), 235–255.Mehra, Y. P. (2001). The wealth effect in empirical life-cycle aggregate consumption equations. Federal Reserve Bank of Richmond Economic Quarterly, 87(2), 45–68.Miles, D. K. (1992). Housing markets, consumption and financial liberalization in the major economies. European Economic Review, 36, 1093–1136.Nickell, S. (1981). Biases in dynamic models with fixed effects. Econometrica, 49, 1417–1426.Ogawa, K., Kitasaka, S., Yamaoka, H., & Iwata, Y. (1996). An empirical re-evaluation of wealth effect in Japanese household behavior. Japan and the World Economy,

8(4), 423–442.Otoo, M. (1999). Consumer sentiment and the stock market. Board of Governors of the Federal Reserve SystemFinance and Economics Discussion Paper No. 60.Pan, M. S., Folk, R. C. W., & Liu, Y. A. (2007). Dynamic linkages between exchange rates and stock prices: evidence from East Asian markets. International Review of

Economics and Finance, 16(4), 503–520.Peltonen, T., Sousa, R. M., & Vansteenkiste, I. (2011). Fundamentals, financial factors and the dynamics of investment in emerging markets. Emerging Markets Finance

and Trade, 47(03S2), 88–105.Peltonen, T, Sousa, R. M., & Vansteenkiste, I. (forthcoming). Investment in emerging market economies. Empirical Economics.Pissarides, C. (1978). Liquidity considerations in the theory of consumption. Quarterly Journal of Economics, 92(2), 279–296.Poterba, J. M. (2000). Stock market wealth and consumption. Journal of Economic Perspectives, 13, 91–118.Rafiq, M. S., & Mallick, S. K. (2008). The effect of monetary policy on output in EMU3: a sign restriction approach. Journal of Macroeconomics, 30, 1756–1791.Romer, C. (1990). The great crash and the onset of the great depression. Quarterly Journal of Economics, 105, 597–624.Sheiner, L. (1995). House prices and the savings of renters. Journal of Urban Economics, 38(1), 94–125.Skinner, J. (1999). Housing wealth and aggregate saving. Regional Science and Urban Economics, 19, 305–324.Sousa, R. M. (2010). Consumption, (dis)aggregate wealth and asset returns. Journal of Empirical Finance, 17(4), 606–622.Sousa, R. M. (2010). Wealth effects on consumption: evidence from the Euro Area. Banks and Bank Systems, 5(2), 78–87.Sousa, R. M. (2010). Housing wealth, financial wealth, money demand and policy rule: evidence from the Euro Area. The North American Journal of Economics and

Finance, 21(1), 88–105.Starr-McCluer, M. (2002). Stock market wealth and consumer spending. Economic Inquiry, 40(1), 69–79.Tan, A., & Voss, G. (2003). Consumption and wealth in Australia. The Economic Record, 79(244), 39–56.Thaler, R. (1990). Anomalies: saving, fungibility and mental accounts. Journal of Economic Perspectives, 4, 193–206.


wealth effects in emerging market economies

Documents