The Interaction between Market Sentiments in the U.S. Financial Market and Global Equity Market

This paper adopts the volatility index and Baker-Wurgler index as the U.S. financial market sentiment measures. Using monthly data from June 1965 to December 2010, we identify the causal relationships between sentiment and the performance of global equity markets. We include 23 G20 market indices, 28 European indices, 25 Asia-Pacific indices, and 10 Americas indices, and employ Granger causality procedure to explore the linkages. We find that the international equity markets are not greatly affected by the U.S. financial market sentiment. The type of extreme sentiment, whether it is optimistic or pessimistic, is irrelevant to its influential power. The equity markets that are affected by the volatility index do not cluster in any region. In contrast, the majority of global equity markets can Granger cause the U.S. investor sentiments, with optimistic market atmosphere being more affected. The equity markets in the Americas and Europe are highly influential to the U.S. investors, compared to the Asian markets.


Introduction
This paper employs the market sentiments time series data and global equity market prices and returns to exam the interaction between the U.S. investor sentiments and the international stock performances.We include 23 G20 market indices, 28 European indices, 25 Asia-Pacific indices, and 10 Americas indices, and employ Granger causality tests to explore the linkages.By converting the daily VIX indices into monthly variables, we identify three sentiment types: market panic, extreme market optimism, and market consensus.The pairwise causality tests then identify the mutual impacts of different market sentiments and cross-country equity returns.
While previous studies shed some light on the financial market contagion across countries, these researches mainly focus on the contagion among the homogeneous variables: equity prices, returns, or volatility.For example, Eryigit and Eryigit (2009) study the equity returns contagion and conclude that the market spillovers are geographically based clustering behaviors.Dungey, Fry, and Martin (2003) study the equity prices interactions and suggest that comovements in Asian and Australian stock markets are because of the common systemic interdependent factors.Karunanayake, Valadkhani, and O'Brien (2010) focus on equity volatility contagion and the spillover is unidirectional from the bigger markets to smaller ones.This paper takes the rare perspective that studies the interaction of heterogeneous financial variables: the spillover between the sentiments and returns.We attempt to address two issues: whether the market sentiment in the United States can spread beyond boundary and affect oversea markets, and whether the global equity market movements can explain the fluctuation of the sentiment in the American market.Exploring the conclusions for these two topics improve the conventional framework of return contagion, which identifies the homogeneous return contagion without revealing the endogenous factor of such contagion.Our study explores the driving force of the integration of oversea equity markets from the sentiment perspective, as sentiment is a measure that is more liquid than the fundamental factors of public listed companies around the world.
Our study does not assume the type of rationality of investors.The debate in terms of the validity of the rational agent assumption leads to the significant distinguishes in the classical theory of finance and the theory of behavior finance.The former is established on the setting of investor maximization of profit and the latter assumes limited rationality.However, this paper attempts to empirically identify the role of investor sentiment, and the conclusion does not necessarily support either side of the debate.In other words, if investors are prone to the impact of sentiment, it is not equivalent to the assertion that they are irrational, or emotional.Investors might interpret the sentiment delivered from the market rationally and update their investment decisions based on the premise that the sentiment contains information that they are not aware of.Actions on the observation of sentiment are might be rational moves, or irrational herding behaviors.
A few previous literatures emphasize the interaction of sentiment and the security returns (Brown & Cliff, 2004;Joseph, Wintoki, & Zhang, 2011), or the sentiment contagion among geographically different markets (Baker, Wurgler, & Yuan, 2012).The conclusions are inconsistent: Brown and Cliff (2004) do not support that sentiment primarily affects individual investors and small stocks, and sentiment has limited role for near-term future stock returns; on the other side, Baker and Wurgler (2006) suggest the strong role of sentiment in stock returns.Previous findings are dependent to the type of sentiment proxy selected.(Corredor, Ferrer, & Santamaria, 2013).We attempt to categorize and compare the measures of sentiment in the past studies as follows.
Many previous studies use the BW market sentiment index (Baker & Wurgler, 2006, 2007).Examples are Stambaugh, Yu, and Yuan (2012), and Laborda and Olmo (2013).According to Baker and Wurgler's method, the BW index is based on the first principal component of six orthogonal sentiment proxies: value-weighted dividend premium, IPO volume, first day returns on IPO, closed-end fund discount, equity share in new issues, and NYSE turnover.
The stock option based VIX index is also widely adopted as market sentiment variable, for example, Ben-Rephael, Kandel, & Wohl (2012) use the VIX as market sentiment proxy.The Chicago Board Options Exchange (CBOE) VIX indices, which is often referred to as the "investor fear gauge", are the benchmark for stock market volatility.It is based on market portfolio index option prices and incorporates information from the volatility skewness by setting a wide range of exercise prices.
The Index of Consumer Sentiment produced by the University of Michigan Survey Research Center is also utilized as a market sentiment indicator by some articles, for instance, Akhtar, Faff, Oliver, & Subrahmanyam (2012).We however do not involve this series in the current paper as it contains only one series of monthly data and does not separately provide the financial asset investors' extreme or modest emotions.
Other studies employ different methods by assuming the mood of investors as market sentiments from exogenous events.Al-Hajieh, Redhead, & Rodgers (2011) examine whether the mood brought by the holy month of Ramadan affects the Islamic Middle Eastern stock markets.Palomino, Renneboog, & Zhang (2009) uses the outcomes of soccer club games as investor moods to test its relation with the stock returns.
This study uses the BW index and the VIX indices as the proxy of market.The monthly BW data has one series, whereas the daily VIX indices include four series: option volatilities based on Stand and Poor's 100 and 500 indices, the Dow Jones Industrial Average, and the NASDAQ.For each of the daily index, we calculate its monthly maximum, minimum, and median levels as the investor extreme fear, optimistic, and consensus sentiments.
We find that international equity markets are not greatly affected by the U.S. financial market sentiment.Equity indices in different countries and regions are determined by their own fundamental factors of the firms publicly listed, rather than the market sentiment of the United States market.The significant impacts of sentiment do not show any unanimous pattern in term of the sentiment type.The ratio of significant maximum, minimum, and median VIX, which stand for the market panic, optimism, and consensus, are close to even.Hence the directions of market atmosphere are irrelevant to whether it can be contagious among the markets.
We also conclude that the equity markets that are affected by the volatility sentiment index do not fall into any category in terms of region or scale.Markets that can be infected by the U.S. VIX sentiment are from the following countries: Brazil, Canada, Mexico, the United States, Indonesia, Jordan, South Korea, and Austria.These countries vary from small developing economies to large developed countries.On the other hand, countries of which the equity markets are not affected also have various types.
As we reverse the causality direction and examine the impacts of the global equity market to the volatility sentiment, the results suggest that: Firstly, the majority of global equity markets can Granger cause the U.S. investor sentiments.Most of the countries that cannot lead to market atmosphere turbulence are small and have limited influence on the regional or global economy.Secondly, the type of sentiment being affected is not even.We detect 22 or 29% of causalities lead to influence of market panic, 32 or 42% lead to influence of market optimism, and the rest 22 or 29% result in the change of consensus market sentiment.Optimistic market atmosphere is more affected by the global equity market, rather than the pessimistic or modest ones.Thirdly, the driving forces of U.S. financial market sentiment are not evenly located in the world.The equity markets in the Americas are highly influential to the U.S. investors.In addition, the American financial market also observes the performances of European markets closely.However, the Asian markets are much less influential.The global equity markets and the U.S. financial markets are mutually less influential, when the U.S. sentiment is measured by the BW index.In addition, the markets that bear the relationship have changed.

Data and Methodology
This paper employs data from two categories: the market sentiments time series data and global equity market prices and returns.The Baker andWurgler (2006, 2007) index (BW) and the VIX indices serve as market sentiments in the United States financial market.The BW index is based on first principal component of six (standardized) sentiment proxies: value-weighted dividend premium, IPO volume, first day returns on IPO, closed-end fund discount, equity share in new issues, and NYSE turnover.Baker and Wurgler first use each of the six proxies as the explained variable and use a set of macroeconomic conditions as the explanatory variables to perform a linear regression.The residuals of the six regressions are orthogonal with respect to a set of macroeconomic conditions.These linear independent regression errors are linearly combined to be the BW index.The specific relationship for the orthogonalized variables is: where SENTIMENT is the BW market sentiment index, CEFD is the closed-end fund discount; TURN is the natural log of the raw turnover ratio, detrended by the 5-year moving average; NIPO is number of IPOs; RIPO is the average first-day returns; S is the share of equity issues in total equity and debt issues; P D-ND is the dividend premium by calculating the log difference of the average market-to-book ratios of payers and nonpayers.
The VIX indices are another measure of market sentiment from the investor side.The Chicago Board Options Exchange (CBOE) publishes the CBOE Volatility Index (VIX) as the scale of stock market volatility.CBOE first creates the weighted average value of options with a constant maturity of 30 days to expiration.The options are based on market portfolio index option prices and incorporate information from the volatility skewness by setting a wide range of exercise prices.Four market portfolio indices are included: the Standard and Poor's 100 and 500 index, the Dow Jones Industrial Index, and the NASDAQ returns.VIX is often cited as an indicator of investor panic, as volatility signifies financial turbulence.During financial stress and periods of significant security price drops, VIX increases, and vice versa.We adopt the S&P 500-based VIX index in this paper.
We convert the daily VIX indices into monthly variables.The daily index is the average of daily high and low.For each of the daily index based on different market portfolios, we calculate its monthly maximum, minimum, and median levels.Monthly VIX maximum is the highest point of market panic; in contrast, the monthly VIX minimum is the extreme market optimism.The monthly median VIX, as the market consensus, is the general investor attitude without extreme gauges.
The series of major equity market indices are organized by the Yahoo!Finance database.We categorize the indices, with some repeating ones, into four categories: the G20 group, markets, the European markets, the Asia-Pacific markets, and the Americas markets.The G20 group includes 23 indices; the European markets include 28 indices; the Asia-Pacific markets include 25 indices, and the Americas markets include 10 indices.Table 1 provides a detailed list of index names and codes.All the series involved includes monthly data from June 1965 to December 2010, with some missing data.The series use the close level of the index and are not adjusted by split and dividend payout.The reason that we do not employed the adjusted close level is the nature of sentiment contagion is not directly related to the fundamentals of the firm in the short term.Before we proceed to the Granger causality tests, we first run the standard ADF unit root tests on all the time series variables.The variables that do not reject the hull hypothesis of non-stationary variable cannot be included in the Granger causality regressions because of the autocorrelation violation.These variables are then converted into the first order difference form, which are tested to be covariance stationary.We involve two series for each equity market index: the plain price levels, and the return levels.The results are reported in Table 2.

Results and Discussion
We present the Granger causality results in this section.Table 3 and Table 4 employ the VIX index as the indicator of market sentiments, whereas Table 5 and 3.4 replace the VIX index with the BW index.All the tests are performed in four groups of markets: the Americas, the Asia-Pacific area, Europe, and G20.The prefixes RET of the equity indices refer to the return level of these indices, and the prefixes DIF of the indices denote the first order difference of the price levels of these indices.For the VIX index, we calculate the monthly maximum (MAX), minimum (MIN), and median (MEDIAN) levels to represent the market panic, optimism, and consensus sentiments, respectively.For both the tables 3 to 4, statistically significant results imply that the variable in the left column can Granger cause the variable in the right column.
Table 3 describes the causality from the market sentiment in the United States financial market to the global equity market.The results reveal three major facts: firstly, international equity markets are not greatly affected by the U.S. financial market sentiment.Equity indices in different countries and regions are more independent from the impact of the U.S., and are determined by their own fundamental factors of the firms publicly listed.
Secondly, the significant impacts of sentiment do not show any unanimous pattern in term of the sentiment type.
The ratio of significant maximum, minimum, and median VIX, which stand for the market panic, optimism, and consensus, are close to even.We thus draw the conclusion that the directions of market atmosphere are irrelevant to whether it can be contagious among the markets.Global investors do not weigh extreme or modest emotions more than one the other.
Thirdly, the equity markets that are affected by the volatility sentiment index do not fall into any category in terms of region or scale.Markets that can be infected by the U.S. VIX sentiment are from the following countries: Brazil, Canada, Mexico, the United States, Indonesia, Jordan, South Korea, and Austria.countries vary from small developing economies like Indonesia to large developed economies, such as Canada.
On the other hand, countries of which the equity markets are not affected also have various types.For instance, Ecuador, Sri Lanka, Japan, and the U.K., are independent from such sentiment.Note.Series with † rejects the null hypothesis that the series on the left fails to Granger cause the series on the right at 5% level of significance.
In the next step, we reverse the causality direction and examine the impacts of the global equity market to the volatility sentiment.Such tests are intuitively correct, because investors do not generate their judgments and attitudes based on pure emotion.The VIX index is an indicator of investors' belief based on their observations in the international markets.Table 4 reports the regression results.The causal relationships from the equity markets to the VIX sentiment index are much more significant, compared to Table 3.We summarize the conclusions below: Firstly, the majority of global equity markets can Granger cause the U.S. investor sentiments.Only a few countries cannot lead to market atmosphere turbulence, which are: Ecuador, Indonesia, Jordan, South Korea, Sri Lanka, Malaysia, Estonia, Italy, and Iceland.Most of these economies are small and have limited influence in the regional or global economy, other than South Korea and Italy.Somewhat surprising is the fact that South Korea, Italy, and Indonesia are the G20 countries.Therefore it seems that the U.S. investors form their sentiments not from a fully reasonable source, but are slightly biased.Such biasness is not significant to affect their rationality, especially given the fact that the entire G8 group is considering expel Italy.
Secondly, the type of sentiment being affected is not even.We detect 76 existing significant causal relationships in the non-overlapping markets from the following four groups.We identify an existing causal relationship as long as either of the price level or the return level of the index can Granger cause the investor attitude.In the 76 significant causalities, 22 or 29% of them lead to influence of market panic, 32 or 42% of them lead to influence of market optimism, and the rest 22 or 29% result in the change of consensus market sentiment.Optimistic market atmosphere is more affected by the global equity market, rather than the pessimistic or modest ones.
Thirdly, the driving forces of U.S. financial market sentiment are not evenly located in the world.The equity markets in the Americas are highly influential to the U.S. investors, with Ecuador being the only exception.In addition, the American financial market also observes the performances of European markets closely, except Estonia and Iceland.However, the Asian markets are much less influential.The number of causality and the robustness of such relationship are lower.Some countries merely present weak impact on the U.S. market attitude by exhibiting only one causal link.For example, only the price level of the Shanghai Composite Index in China affects the optimistic end of the U.S. sentiment, while its return level has no role in the U.S. market in terms of all types of sentiment.
In order to compare the effectiveness of the sentiment measures, we proceed to perform the pairwise Granger causality tests between the BW sentiment index and the international equity market.Table 5 reports the causality from the BW sentiment index to the global equity market, whereas Table 6 reports the causality in the opposite direction.This further step generates non-collinear conclusions, because of the fundamentally different natures of the VIX index and the BW index.The former is based on the volatility of option prices of the S&P 500 index, yet the latter is a linear combination of fundamental market variables.
The global equity markets and the U.S. financial markets are mutually less influential, when the U.S. sentiment is measured by the BW index.In addition, the markets that bear the relationship have changed.The previously closely-linked Americas markets show independence.A plausible explanation is that the components of the BW index focus on the U.S. firms.As part of the BW index, the number of IPOs, the average IPO first-day returns, the share of equity issues in total equity and debt issues, and the dividend premium are strongly affected by the profitability of the individual firms publicly listed in the U.S.These factors, however, do not respond to the global equity market turbulences in the short run.We also conclude that the VIX index is a better measure of sentiment contagion in the global context.

Table 1 .
The equity markets, stock indices and codes in groups This table presents the four groups of equity markets in this paper: the G20 group markets, the European markets, The Asia-Pacific markets, and the Americas markets.The codes of the indices involved in the regressions are established by the International Organization for Standardization (ISO) in 2000.

Table 2 .
Unit root tests of the prices and returns of global equity series Note.The null hypothesis is unit root does not exist.Significant p values therefore imply the series is non-stationary.The variables with only codes presented in Table1are the price levels of the series, and the variables with RET added as the prefix are the return levels of the series.

Table 2
suggests the existence of unit root in all the price level series of the equity market indices.Hence we proceed to replace the original variables with first order difference series and use the stationary variables in the following Granger causality tests.For a bivariate linear autoregressive model with pairwise variables  1 and  2 , the test regression is: P in the regression equations is the maximum number of lags included, and the matrix A is the plain vanilla VAR coefficients. • () is the regression residual.If the variance of  • () is improved by adding  1 or  2 , it implies that  1 or  2 Granger causes  2 or  1 .The way to detect such improvement is by testing whether, for example, the coefficients carried by  12 are jointly different from zero.If the null hypothesis of  12 = 0 is rejected significantly by the F test,  2 Granger causes  1 .We use the Bayesian Information Criterion (BIC) to determine the number of lags.

Table 3 .
Significant granger causalities from the VIX-based sentiment to the global equity markets Series with † rejects the null hypothesis that the series on the left fails to Granger cause the series on the right at 5% level of significance.

Table 6 .
Granger causality from the global equity markets to the BW-based sentiment Series with † rejects the null hypothesis that the series on the left fails to Granger cause the series on the right at 5% level of significance.