Open Access

Interdependence between the stock market and the bond market in one country: evidence from the subprime crisis and the European debt crisis

Financial Innovation20173:5

DOI: 10.1186/s40854-017-0055-z

Received: 27 October 2015

Accepted: 16 March 2017

Published: 9 April 2017

Abstract

Background

Once a global financial crisis breaks out, the interdependence between different financial markets suddenly increases and leads to a significant contagion.

Methods

With 39 countries used as samples, this paper analyzes the interdependence between the stock market and the government bond market during the crisis periods.

Results

It proves that the investor focuses more on the safety of their portfolio so there is neither a flight from quality nor a positive spillover during a crisis period. When one market is safer than the other market in the same country, a flight to quality occurs between the two markets; however, when the two markets in one country are both risky, negative spillover appears between these two markets.

Conclusions

This means a flight to quality from the stock market to the short-term government bond will occur more frequently than will occur from the stock market to the long-term government bond markets. In addition, a flight to quality always emerges in developed markets, while negative spillovers take place in emerging markets and in the PIIGS countries (Portugal, Italy, Ireland, Greece, and Spain, referred to hereon as “PIIGS”) in the European Debt Crisis.

Background

In the 21th century, there have been two destructive global financial crises, i.e., the Subprime Crisis in 2007–2009 and the 2009’s European Debt Crisis. Once a global financial crisis breaks out, the interdependence between different financial markets suddenly increases and leads to a significant contagion.

Although most of the co-movements between the stock market and the bond market demonstrate interdependence instead of a contagion, as Forbes and Rigobon (2002) mentioned, this interdependence could be deemed an overture of financial contagion. Because interdependence occurs more frequently than contagion, which is a special type of interdependence, it is more important to examine the mechanism of interdependence during a crisis period. The stock market is the largest and most liquid market in most regions. Meanwhile, the markets of bonds, commodities and derivatives also develop quickly in many countries. In addition to the stock market, the bond market is another financial market with large trade volumes. Therefore, any interdependence between the stock market and the bond market is more important and significant than that between some other markets.

Most studies have proven the negative correlation between the stock market and the bond market in non-crisis periods. Theoretically, there are four types of interdependence between a stock market and bond market that exist in the same country. During a period of financial crisis, investors tend to transfer their investments from the risky markets to safer markets in order to avoid significant losses. This flight to quality accelerates the fund run-off from markets in crisis and amplifies the contagion effect, while it might increase the investors’ utility. After the crisis, investors often rebalance their investments from safe assets (e.g. bonds) to riskier assets (for example, stocks). This rebalance is known as a flight from quality. Additionally, as another common interdependence in the financial crisis, negative spillover refers to the negative co-movement of asset prices between several financial markets in the crisis. This is because the markets might all exist in an unsafe region, so investors consequently rebalance their portfolios and investments to flow into another, safer region. This positive spillover may also occur in similar markets as an economy starts to recover and the confidence of the investors increases.

Different types of interdependence suggest different patterns of investor behavior during a period of financial crisis. As such, it is necessary to investigate different types of interdependence. Consequently, the stock market's returns generally decrease. If the correlation between the stock and the bond is negative before the crisis and becomes even more significantly negative during the crisis, this phenomenon is defined as a flight to quality, while bonds have lower risk than stocks in the same country. This suggests that investments flow from the stock market to the bond market because of higher risk aversion during a crisis. However, if the change mentioned above occurs while the return of the stock market increases, this special phenomenon is defined as a flight from quality. This phenomenon occurs during a crisis recovery period when investors seek greater risks. In addition, if the correlation between the stock and bond is negative before the crisis and becomes positive during the crisis, this phenomenon is defined as the negative spillover. This often occurs in markets that are shocked by a severe crisis and decrease all investments in different assets. Similarly, a positive spillover may happen during a crisis recovery period, which means the correlation between the stock market and the bond market becomes positive and the returns in both markets increase. The above four phenomena are defined with regard to the possibilities of relationships between the stock market and the bond market during a crisis. Most of the existing literature focuses on the flight to quality or the negative spillover, but few studies review all the four types of interdependence. However, this paper analyzes all four types of interdependence between the stock markets and the government bond markets, and their implications in two recent financial crises. In addition, the government bond markets are classified with different maturities to enrich the results. Due to the availability of the bond data, this paper focuses on the European Debt Crisis and the Subprime Crisis. For the purposes of this paper, the bankruptcy of American Home Mortgage Holdings, Inc., the 10th largest mortgage company on August 6, 2007, marks the start of the Subprime Crisis. Additionally, the European Debt Crisis began with the first degradation of Greece on October 22, 2008.

This paper also focuses on different types of interdependence between the stock markets and government bond markets in the same region. First, the interdependence between the stock markets and the bond markets in the same region is much more important than the interdependence across different regions (Baur and Lucey 2009). Second, the paper investigates the interdependence between the stock markets and the bond markets in 39 countries. The samples already cover approximately 80% of the global GDP, and therefore, it can be concluded that this paper’s analysis is reliable. The interdependence across different regions may result in an overly complicated analysis. In addition, over recent years the interdependence between the stock markets and the bond markets has become quite valuable because the debt crisis is one of the most important parts in the Subprime Crisis and the European Debt Crisis. The financial derivatives or the real estate markets are less liquid and important than government bond markets in most of the emerging markets. Thus, in this paper, our analysis emphasizes the stock markets and the government bond markets in 39 countries.

The primary result in this paper is that there are the significant instances of flight to quality and the negative spillover in the European Debt Crisis and the Subprime Crisis. Furthermore, the interdependence between the stock markets and the bond markets in these recent crises is classified and the characteristics are summarized according to region. Compared to existing studies, our analysis is based on a rich dataset that includes the stock markets and the government bond markets in 39 countries around the world. Lastly, a general implication is provided for the characteristics of interdependence in the recent two global financial crises.

The rest of this paper is organized as follows. Section II begins with a brief review of the theory on interdependence and contagion during times of financial crisis. In Section III, the analysis methodology is investigated; additionally, our hypotheses are listed in this section. In Section IV, a brief description of the data in the analysis is provided. Next, in Section V, we present discussions about the interdependence between the stock markets and the government bond markets in 39 countries, including three sub-topics. The analysis in this work proves that the interdependence between the stock markets and the government bond markets occurs in most of the 39 countries. Next, the characteristics of the interdependence in those different regions and crises are, respectively, analyzed. In Section VI, the robustness of the results is emphasized and the findings are reconciled using the same dataset. Finally, Section VII includes this work’s results and consequential policy implications.

Compared to the existing literature, the latest global financial crises are currently studied using more abundant samples, i.e., 39 different countries and regions, a method that is in favor of financial contagion studies. Additionally, in the final results, conducting empirical research is successful in classifying possible types of interdependence between the stock markets and the government bond markets. Furthermore, it has been found that risk aversion frequently causes investment flows across different regions in financial crises, which is simple but convincing logic regarding the special patterns of interdependence during the Subprime Crisis and the European Debt Crisis.

Literature review

Many studies have revealed the significant interdependence between the stock markets and the bond markets in certain countries during past financial crises.

Aslanidisa and Christiansen (2012) proved the interdependence between the U.S. stock markets and bond markets by adopting the Markov-switching method proposed by Hamilton (Hamilton 1989). Furthermore, Chan et al. (2011) analyzed the contagion between different financial markets, including stock markets, bond markets, commodity markets and real estate markets. Additionally, interdependence evidences different correlations during different stages. Based on these findings, Gulko (2002), Baur and Lucey (2009) defined the flight to quality between the stock markets and the bond markets. Cheng et al. (2011) took the Copula Contagion Index as a method of measuring the financial contagion of 50 countries. They investigated the contagion in 13 financial crises during recent 20 years, but their studies were limited to the stock market.

These previous studies have seldom involved different types of interdependence across different asset markets in the same region. Most of the papers mentioned above focus on the interdependence between different regional markets for one identified financial asset. For instance, Brièrea et al. (2012) investigated the interdependence between different regions. They proved that the flight to quality is more frequent than the contagion during a financial crisis. The existing studies on different types of interdependence feature two limitations. First, the majority of studies focus on the logic of the flight to quality or co-movements, because cash flow direction is not published and is in reality difficult to track. Bernanke et al. (1996) analyzed the mechanism of flight to quality in the credit market by employing the general equilibrium model. Vayanos (2004) compared the flight to quality and the flight to liquidity in the different asset markets using dynamic stochastic general equilibrium modeling. Caballero and Krishnamurthy (2008) also investigated the lender of the last resort during the flight to quality period under the guidance of general equilibrium theory. Second, some studies on interdependence between different assets focus on interdependence in one specific country but seldom conduct a comparison of interdependence between various asset markets in different regions, especially the comparisons between developed and emerging markets. Lang and Nakamura (1995) summarized the rules of flight to quality in the U.S. credit markets, while Baur and Lucey (2009) focused on flight to quality in the stock markets and corporate bond markets in the G8 countries; however, they failed to study a similar phenomenon in emerging markets. Beber et al. (2009) found that interdependence is aroused by flight to liquidity instead of quality when investigating interdependence in the European government bond markets.

Methods

In this paper, the interdependence between the stock markets and the government bond markets of 39 countries is studied in three steps.

Firstly, using the Markov-switching method, this work tests whether the interdependence between the stock markets and the government bond markets occurred during the Subprime Crisis and the European Debt Crisis. A sudden increase in correlations during the crises would prove significant interdependence between the stock markets and the government bond markets. Secondly, this work clarifies all kinds of interdependence by measuring the flight to quality, the flight from quality, the negative spillover and the positive spillover in conjunction with the different changing directions of interdependence. This work also summarizes the features of interdependence between the stock markets and government bond markets in various regions. Lastly, the characteristics of the interdependence in the Subprime Crisis and the European Debt Crisis are also mentioned, respectively.

The test for interdependence between the stock markets and the government bond markets in the same country

Based on the Markov-switching method, we tested the correlation between the stock markets and the short-term and long-term government bond markets in the same country to investigate whether the interdependence exists during the Subprime Crisis and the European Debt Crisis. Actually, the Markov-switching method is similar to the Markov-switching regression proposed by Hamilton (1989):
$$ {\mathrm{R}}_{\mathrm{i},\mathrm{t}}^S={c}_i+{\beta}_i^S{\mathrm{R}}_{\mathrm{i},\mathrm{t}-1}^{\mathrm{S}}+{\beta}_i^{sB}{\mathrm{R}}_{\mathrm{i},\mathrm{t}}^{\mathrm{sB}}+{\beta}_i^{lB}{\mathrm{R}}_{\mathrm{i},\mathrm{t}}^{\mathrm{lB}}+{\upepsilon}_{i, t},{\upepsilon}_{i, t}\sim N\left(0,{\sigma}_i^2\right), $$
(1)
$$ {\mathrm{p}}_{\mathrm{i},\mathrm{j}}= Prob\left({s}_t= j\left|{s}_{t-1}= i\right.\right),{p}_{i1}+{p}_{i2}=1, i=1,2 $$

In Formula (1), the dependent variable is \( {\mathrm{R}}_{i, t}^S \), i.e., the return of the stock market. To study the dependence and enrich the results, the returns of short-term and long-term government bonds (i.e., \( {\mathrm{R}}_{i, t}^{sB} \) and \( {\mathrm{R}}_{i, t}^{lB} \), respectively) are introduced into Formula (1) to test their correlations with the stock market in the same country. In order to take the momentum effect into account, the 1-period (i.e., 1-month) lag return \( {\mathrm{R}}_{i, t-1}^S \) of the stock market is considered an independent variable.

All the independent and dependent variables have two transition statuses in Formula (1). The transition status can be defined as the non-crisis status that refers to the status with a higher volatility of returns and the crisis status, which is related to the lower volatility of returns.

If the estimation of the bond coefficients is significant enough in Formula (1), it can be proved that some interdependence exists between the stock markets and the related bond markets. Therefore, the implication of the interdependence can be analyzed by verifying its direction in the following sections.

The investigation of different types of interdependence

The interdependence in this paper is defined as the significant increase in the correlation between the two markets. Based on the direction of the correlations as well as Baur and Lucey (2009)’s analysis, there are four types of interdependence, as follows (Table 1).
Table 1

Possible interdependences between the stock and the government bond markets

Returns

Change

Short-term (1-3 year) government bond

Long-term (7-10 year) government bond

Negative Correlation & Negative Changing Direction

Positive Correlation & Positive Changing Direction

Negative Correlation & Negative Changing Direction

Positive Correlation & Positive Changing Direction

Stock Market

Downside

Flight to Quality (FTQ)

Negative Spillover (NS)

Flight to Quality (FTQ)

Negative Spillover (NS)

Upside

Flight from Quality (FFQ)

Positive Spillover (PS)

Flight from Quality (FFQ)

Positive Spillover (PS)

Different types of the interdependence are measured through the regressions using crisis dummy variables. Similar to Formula (1), the return of the stock market is a dependent variable, while the returns of the short-term and the long-term government bond markets are independent variables in Formula (2). To decrease the autocorrelation, the lag item of the stock return is introduced into Formula (2).
$$ \begin{array}{l}{\mathrm{R}}_{\mathrm{t}}^{\mathrm{S}}={c}_t+{\beta}^S{\mathrm{R}}_{\mathrm{t}-1}^{\mathrm{S}}+{\beta}^{sB}{\mathrm{R}}_{\mathrm{t}}^{\mathrm{sB}}+{\beta}^{lB}{\mathrm{R}}_{\mathrm{t}}^{\mathrm{lB}}+{\gamma}^{sB}{R}_t^{sB}{D}_t^{Crisis}\\ {}+{\gamma}^{lB}{R}_t^{lB}{D}_t^{Crisis}+{\upepsilon}_t,{\upepsilon}_t\sim N\left(0,{\sigma}_i^2\right)\end{array} $$
(2)

In Formula (2), \( {\mathrm{R}}_t^S,{\mathrm{R}}_t^{lB}\ \mathrm{and}\ {\mathrm{R}}_t^{sB} \) refer to the returns of the stock market, the long-term government bond market and the short-term government bond market at Period t, respectively. \( {D}_t^{Crisis} \) refers to the crisis dummy variable. As the lag item of the stock return is incorporated, the error follows the normal distribution and passes the DW test. Then, types of interdependences can be tested through the coefficients β and γ.

Furthermore, in order to compare different types of interdependence during the two crises in recent years, different dummy variables are considered in the European Debt Crisis and the U.S. subprime mortgage crisis on the basis of Formula (2).
$$ \begin{array}{l}{\mathrm{R}}_{\mathrm{t}}^{\mathrm{S}}={c}_t+{\beta}^S{\mathrm{R}}_{\mathrm{t}-1}^{\mathrm{S}}+{\beta}^{sB}{\mathrm{R}}_{\mathrm{t}}^{\mathrm{sB}}+{\beta}^{lB}{\mathrm{R}}_{\mathrm{t}}^{\mathrm{lB}}+{\gamma}^{sB}{R}_t^{sB}{D}_t^{Crisis1}\\ {}+{\gamma}^{lB}{R}_t^{lB}{D}_t^{Crisis1}+{\upmu}^{\mathrm{sB}}{\mathrm{R}}_{\mathrm{t}}^{\mathrm{sB}}{D}_t^{Crisis2}+{\upmu}^{1\mathrm{B}}{\mathrm{R}}_{\mathrm{t}}^{1\mathrm{B}}{D}_t^{Crisis2}+{\upepsilon}_t,{\upepsilon}_t\sim N\left(0,{\sigma}_i^2\right)\end{array} $$
(3)

Hypotheses

Based on the above definitions, several hypotheses are proposed for the following analysis.

Hypothesis 1

The interdependence becomes more significant once the crisis breaks out.

Hypothesis 2

There is neither a flight from quality nor a positive spillover during the periods of the Subprime Crisis and the European Debt Crisis.

The two hypotheses mentioned above are somewhat trivial. To begin with, when the financial crisis occurs, most investors will rebalance their portfolio to reduce any risk to their investments. Considering that a great number of investors take different positions in different financial markets, this should increase the interdependence across financial markets. Furthermore, the definitions of the flight from quality and the positive spillover need the increasing return of the stock markets. Because the stock market is the most vulnerable financial market in most countries, any increase of the stock return is nearly impossible during a crisis period.

Hypothesis 3

Compared to the long-term government bond market, the flight to quality between the stock markets and the short-term government bond markets is more significant.

In comparison to the short-term government bonds, the performance of the long-term government bonds relies more on the long-term economic trend of the related countries rather than the investor’s emotion. However, the flight to quality is often a short-term phenomenon and depends more on the investors’ emotion during the crisis period. Considering that the sovereign debt of the related countries might increase markedly, the probability of default will also be much higher for the long-term government bonds. In this situation, the long-term government bonds cannot be deemed as safe financial assets. Therefore, it is mentioned in Hypothesis 3 that the fight-to-quality is more dramatic between the stock markets and the short-term government bond markets.

Hypothesis 4

In regards to interdependence between the stock markets and the government bond markets in the same country, the flight to quality often occurs in developed markets while the negative spillover often appears in the emerging markets.

It is obvious that the emerging markets are more vulnerable than the developed markets because emerging countries sometimes have an unstable political system, especially during the crisis. That is why government bonds in emerging countries are considered an unsafe asset in a financial crisis. In addition, those government bonds will not attract international investors due to greater political risk. In this case, most investors choose to turn away from the emerging markets instead of changing their stock position to the bond position. Thus, the negative spillover emerges more frequently than the flight to quality in the emerging markets.

Hypothesis 5

The countries where the flight to quality happens become less frequent in the European Debt Crisis than in the U.S. subprime mortgage crisis.

Similar to Hypothesis 4, the number of countries that suffered from the sovereign debt crisis increased during the European Debt Crisis. Furthermore, as the European Debt Crisis is regarded to be a sovereign debt crisis and also the sequel of the Subprime Crisis, there are fewer government bond markets of high quality. Therefore, government bonds cannot be treated as quality assets and the flight to quality also became less frequent during the European Debt Crisis.

Data

Government bonds

There are many kinds of government bonds in the financial markets. In addition, short-term and long-term government bonds often perform differently in different countries. Therefore, choosing the proper government bonds is a key question in the study.

All the government bonds in this paper were chosen as the bond indices from the Bloomberg database. Based on the maturity of the bonds, the 7–10 year government bond index was selected as the proxy for the long-term government bonds, while the 1–3 year government bond index was chosen as the proxy for the short-term government bonds (Table 2). Then, all the countries that have reliable government bond indices in Bloomberg were selected. In terms of geographic location, the samples involve 39 countries on six continents, i.e. Europe, Asia, North America, Latin America, and Oceania. Economically speaking, these 39 countries can be divided into developed markets, and those emerging markets take up more than 75% of the global GDP.
Table 2

List of the stock and government bond indices

Country

Type of Market

Government Bond Index

Stock Index

Argentina

Emerging America

Inflation-Adjusted

MERV

Australia

Developed Asia

EFFA

S&P 200

Austria

Developed Europe

EFFA

AUX

Belgium

Developed Europe

EFFA

BFX

Brazil

Emerging America

Inflation-Adjusted

IBOVESPA

Canada

Developed America

EFFA

Toronto 300

Chile

Emerging America

Inflation-Adjusted

IPSA

China

Emerging Asia

Zhongzhai All-value

Shanghai Composite

Czech

Emerging Europe

EFFA

Prague Composite

Denmark

Developed Europe

EFFA

OMX20

Finland

Developed Europe

EFFA

OMX

France

Developed Europe

EFFA

CAC40

Germany

Developed Europe

EFFA

DAC

Greece

Developed Europe

EFFA

ASE Composite

Hungary

Emerging Europe

EFFA

BUX

India

Emerging Asia

Value-weighted Yield

SENSEX30

Indonesia

Emerging Asia

Value-weighted Yield

Jakarta Composite

Ireland

Developed Europe

EFFA

Ireland Composite

Israel

Developed Asia

Inflation-Adjusted

TA100

Italy

Developed Europe

EFFA

FTSEMIB

Japan

Developed Asia

EFFA

Nikkei 225

Korea

Developed Asia

EFFA

Korea Composite

Mexico

Emerging America

Inflation-Adjusted

MXX

Netherlands

Developed Europe

EFFA

AEX

New Zealand

Developed Asia

EFFA

NZ50

Norway

Developed Europe

EFFA

OSEAX

Poland

Emerging Europe

EFFA

WIG20

Portugal

Developed Europe

EFFA

PSI20

Russia

Emerging Europe

Exposure-weighted Price

RTS

South Africa

Emerging Asia

EFFA

JALSH

Spain

Developed Europe

EFFA

IBEX35

Sweden

Developed Europe

EFFA

OMXSPI

Switzerland

Developed Europe

EFFA

SMI

Taiwan

Emerging Asia

Value-weighted Yield

Taiwan Value-added

Thailand

Emerging Asia

Value-weighted Yield

Thailand Composite

Turkey

Emerging Europe

Inflation-Adjusted

Istanbul 30

UK

Developed Europe

EFFA

FTSE 100

US

Developed America

EFFA

S&P 500

Vietnam

Emerging Asia

Value-weighted Yield

VNINDEX

All the government bond indices in these 39 countries can be classified into 4 types, according to their data source. Firstly, the government bond indices in most developed markets are the EFFA Government Bond Indices, which are calculated by value weights and the liquidity consideration of Bloomberg. Their value reflects the value-weighted clean prices of most liquid government bonds with the related maturity. The second type of index is the inflation-adjusted government bond index for Latin American countries. This index was chosen for Latin American countries because of the high inflation in those countries. Therefore, the inflation-adjusted government bonds are more popular and liquid in Latin America. There are no EFFA Government Bond Indices for certain Southeast Asian countries or regions in the Bloomberg database. In this case, we chose the value-weighted return of the government bonds and generated the clean price indices for these countries. Finally, there are no reliable government bond indices for Russia and Mainland China in the above three categories. Thus, we have taken into consideration the exposure-weighted government bond index for Russia and Zhongzhai All-value Government Bond Index for China, respectively.

The above government bond indices in the recent 10 years and 39 countries were collected. The time range is from October 2002 to June 2012. To eliminate the asynchronous effect and maintain the maximal information, the weekly returns were generated with the weekly bond indices using the logarithm function, i.e. the log (the close price in the end of this week/the close price in the end of last week).

Stocks

As for the stock market, the value-weighted indices produced by most liquid stocks were used as the proxy, since more than half of the sample countries are emerging markets with a number of illiquid stocks, which do not accurately reflect the prosperity of the stock market.

To compare the stock data with the bond data in the model, a similar method was adopted to generate the weekly log-return from October 2002 to June 2012. This period is the longest available period from the Wind database and the Bloomberg database.

Results and discussion

Investors always hold several kinds of financial positions as a portfolio to diversify their risk in the financial markets. Rebalancing their portfolio often leads to interdependence between different financial markets once a financial crisis breaks out. Thus, in this section, various types of interdependence between the stock markets and the government bond markets in the listed 39 countries were analyzed by adopting the methods mentioned in Section III.

The existence of interdependence

The analysis concerning interdependence between the stock markets and the government bond markets in 39 countries is made through the 39 independent Markov Switching regressions as Formula (1). All the estimations are summarized in Table 3. The estimations in the 39 regressions are statistically significant because most of their P-values are lower than 10% and their AIC statistics are significant enough. In addition, the coefficients of the lag item of the stock market return are much more positive during a non-crisis status than during a crisis status for the majority of regressions. It is shown that, for most stock markets, the momentum effects become less significant and the stock returns influence the bond markets more easily in crisis periods.
Table 3

Detailed results of markov switching regressions

Parameters

Samples

AIC

Standard Error

Transition Possibility

Status

  

Status 1

Status 2

p11

p12

p21

p22

Argentina

368

−1474

0.001a

0.003a

0.982a

0.040a

0.018a

0.960a

Austria

508

−2303

0.000a

0.002a

0.974a

0.050a

0.026a

0.950a

Australia

508

−2756

0.000a

0.001a

0.970a

0.033a

0.030a

0.967a

Belgium

508

−2519

0.000a

0.001a

0.978a

0.047a

0.022a

0.953a

Brazil

364

−1534

0.001a

0.002a

0.978a

0.115a

0.022a

0.885a

Canada

508

−2670

0.000a

0.001a

0.970a

0.081a

0.030a

0.919a

Switzerland

508

−2620

0.000a

0.001a

0.975a

0.064a

0.025a

0.936a

Chile

432

−2155

0.000a

0.001a

0.973a

0.059a

0.027a

0.941a

China

293

−1165

0.000a

0.002a

1.000a

0.006a

0.000

0.994a

Czech

508

−2396

0.000a

0.002a

0.972a

0.095a

0.028a

0.905a

Germany

508

−2363

0.000a

0.002a

0.987a

0.050a

0.013a

0.950a

Denmark

508

−2462

0.000a

0.001a

0.972a

0.076a

0.028a

0.924a

Spain

508

−2399

0.000a

0.001a

0.970a

0.053a

0.030a

0.947a

Finland

508

−2297

0.000a

0.001a

0.977a

0.039a

0.023a

0.961a

France

508

−2432

0.000a

0.001a

0.955a

0.061a

0.045a

0.939a

Greece

508

−2168

0.000

0.002a

0.991a

0.005a

0.009a

0.995a

Hungary

508

−2215

0.000a

0.002a

0.981a

0.068a

0.019a

0.932a

Indonesia

465

−2061

0.000a

0.003a

0.978a

0.145a

0.022a

0.855a

Ireland

503

−2363

0.000a

0.002a

0.970a

0.054a

0.030a

0.946a

Israel

503

−2394

0.000a

0.001a

0.991a

0.028a

0.009a

0.972a

India

507

−2181

0.000a

0.002a

0.990a

0.028a

0.010a

0.972a

Italy

507

−2391

0.000a

0.001a

0.982a

0.024a

0.018a

0.976a

Japan

508

−2330

0.000a

0.003a

0.991a

0.116a

0.009a

0.884a

Korea

508

−2307

0.000a

0.002a

0.979a

0.053a

0.021a

0.947a

Mexico

294

−1335

0.000a

0.002a

0.987a

0.045a

0.013a

0.955a

Netherlands

508

−2401

0.000a

0.001a

0.964a

0.064a

0.036a

0.936a

Norway

508

−2322

0.000a

0.002a

0.979a

0.070a

0.021a

0.930a

New Zealand

508

−2981

0.000a

0.000a

0.945a

0.157a

0.055a

0.843a

Poland

508

−2210

0.000a

0.002a

0.983a

0.051a

0.017a

0.949a

Portugal

508

−2580

0.000a

0.001a

0.969a

0.057a

0.031a

0.943a

Russia

170

−633

0.001a

0.004a

0.990a

0.066a

0.010a

0.934a

Sweden

508

−2429

0.000a

0.001a

0.981a

0.041a

0.019a

0.959a

Thailand

508

−2277

0.001a

0.004a

0.993a

0.202a

0.007a

0.798a

Turkey

172

−732

0.001a

0.000a

0.968a

0.861a

0.032a

0.139a

Taiwan

507

−2322

0.000a

0.001a

0.982a

0.035a

0.018a

0.965a

UK

508

−2640

0.000a

0.001a

0.976a

0.052a

0.024a

0.948a

US

508

−2686

0.000a

0.001a

0.987a

0.046a

0.013a

0.954a

Vietnam

309

−1107

0.000a

0.003a

0.896a

0.120a

0.104a

0.880a

South Africa

508

−2451

0.000a

0.001a

0.986a

0.041a

0.014a

0.959a

Parameters

Intercept

Lag of Stock

Short-term Bond

Long-term Bond

Status

Status 1

Status 2

Status 1

Status 2

Status 1

Status 2

Status 1

Status 2

Argentina

0.003a

−0.004a

0.165a

0.101a

−0.047a

0.128a

0.023a

−0.393a

Austria

0.005a

−0.006a

0.214a

0.103a

0.491a

−1.399a

−0.327a

−0.244a

Australia

0.003a

−0.001a

0.261a

0.006a

0.015a

−1.638a

0.028a

−0.183a

Belgium

0.003a

−0.003a

0.190a

0.166a

0.359a

−1.125a

−0.165a

−0.025a

Brazil

0.002a

−0.015a

0.196a

−0.138a

0.819a

1.187a

0.025a

0.631a

Canada

0.003a

−0.001a

0.109a

0.060a

−0.172a

−2.359a

0.012a

−0.205a

Switzerland

0.003a

−0.005a

0.114a

−0.080a

0.410a

−1.767a

−0.171a

0.044a

Chile

0.004a

−0.003a

0.192a

0.100a

0.204a

−0.577a

0.028a

−0.062a

China

−0.002a

0.002a

0.146a

0.199a

3.415a

−0.809a

−2.131a

−0.743a

Czech

0.004a

−0.006a

0.246a

0.206a

0.219a

−2.789a

−0.191a

0.120a

Germany

0.003a

0.001a

0.093a

−0.011a

0.339a

−2.758a

−0.308a

−0.149a

Denmark

0.004a

0.000a

0.111a

0.108a

0.410a

−3.466a

−0.206a

0.045a

Spain

0.002a

−0.004a

0.215a

0.055a

0.308a

0.171a

0.011a

0.117a

Finland

0.003a

−0.004a

0.216a

0.092a

0.633a

0.342a

−0.532a

−0.539a

France

0.003a

−0.003a

0.132a

0.013a

0.589a

0.081a

−0.251a

−0.362a

Greece

0.003a

−0.007

0.241

0.212a

0.899

−0.151a

−0.396a

0.197a

Hungary

0.003a

−0.009a

0.208a

0.062a

−0.117a

3.166a

0.232a

−0.497a

Indonesia

0.007a

−0.020a

0.153a

−0.258a

−0.002a

0.022a

−0.080a

−0.464a

Ireland

0.004a

−0.008a

0.165a

0.135a

0.034a

0.304a

0.039a

−0.436a

Israel

0.003a

−0.002a

0.151a

0.141a

−0.171a

0.713a

0.184a

−0.170a

India

0.004a

−0.002a

0.251a

0.109a

0.042a

0.072a

−0.056a

−0.246a

Italy

0.001a

−0.004a

0.218a

0.127a

1.021a

−0.961a

−0.009a

0.226a

Japan

0.001a

−0.011a

0.097a

0.070a

1.004a

−1.676a

−0.787a

1.407a

Korea

0.005a

−0.005a

0.143a

0.156a

−0.303a

−1.576a

−0.070a

1.348a

Mexico

0.002a

0.000a

0.167a

0.026a

0.474a

−0.635a

−0.018a

0.210a

Netherlands

0.002a

0.001a

0.214a

0.022a

0.631a

−4.025a

−0.268a

−0.116a

Norway

0.006a

−0.007a

0.227a

0.030a

−0.874a

−1.047a

−0.045a

−0.041a

New Zealand

0.002a

−0.003a

0.288a

0.206a

0.179a

0.910a

0.022a

−0.501a

Poland

0.002a

−0.001a

0.166a

0.084a

0.906a

−2.034a

−0.049a

0.907a

Portugal

0.003a

−0.007a

0.244a

0.108a

0.438a

0.074a

0.140a

−0.183a

Russia

0.000a

0.004a

0.186a

0.192a

1.753a

−0.353a

0.923a

0.518a

Sweden

0.004a

−0.002a

0.090a

0.009a

0.015a

0.091a

−0.239a

−0.432a

Thailand

0.003a

−0.041a

0.206a

−0.265a

−0.041a

0.594a

−0.028a

−0.150a

Turkey

0.005a

−0.055a

0.056a

2.663a

0.772a

−1.889a

0.261a

3.885a

Taiwan

0.003a

−0.003a

0.170a

0.088a

−0.002a

−0.031a

0.029a

0.163a

UK

0.002a

−0.002a

0.179a

0.055a

0.405a

−0.418a

−0.086a

−0.128a

US

0.003a

0.002a

0.150a

0.022a

−0.439a

−6.529a

−0.048a

−0.094a

Vietnam

−0.003a

0.005a

0.276a

0.333a

−0.182a

0.247a

−0.077a

−0.221a

South Africa

0.004a

−0.007a

0.155a

0.015a

−0.120a

1.894a

0.207a

−0.182a

ameans significance with 10% confidential interval

As Krolzig (2004) proved that the market returns are more volatile during the crisis period, the crisis status is defined as switching status with one more volatile (i.e., Status 1 in Table 4) in this paper. The intercepts refer to the mean value of the stock market return, excluding the lag effect and the bond market effect in different switching statuses. In a crisis status defined in terms of volatility, the mean value of the stock market returns is lower for most of the regressions except for those in Mainland China, Russia and Vietnam. This also proves that in this paper, the definition of the crisis status is appropriate.
Table 4

Bond coefficients in markov switching regressions

Short-term Bonds

Non-Crisis

Crisis

Regions

Mean

Median

Standard Error

Negative Coefficient

Mean

Median

Standard Error

Negative Coefficient

Developed America

−0.31

−0.31

0.19

100%

−4.44

−4.44

2.95

100%

Developed Europe

0.38

0.41

0.42

6%

−1.00

−0.69

1.38

63%

Developed Asia

0.14

0.01

0.51

40%

−0.65

−1.58

1.34

60%

Emerging America

0.36

0.34

0.37

25%

0.03

−0.22

0.85

50%

Emerging Europe

0.71

0.77

0.72

20%

−0.78

−1.89

2.38

80%

Emerging Asia

0.44

−0.00

1.31

71%

0.28

0.07

0.83

29%

Developed Market

0.27

0.36

0.46

22%

−1.23

−1.05

1.75

65%

Emerging Markets

0.51

0.12

0.93

44%

−0.11

−0.00

1.47

50%

All

0.37

0.31

0.70

31%

−0.77

−0.42

1.71

59%

Long-term Bonds

Non-Crisis

Crisis

Regions

Mean

Median

Standard Error

Negative Coefficient

Mean

Median

Standard Error

Negative Coefficient

Developed America

−0.02

−0.02

0.04

50%

−0.15

−0.15

0.08

100%

Developed Europe

−0.18

−0.19

0.18

81%

−0.13

−0.12

0.23

69%

Developed Asia

−0.12

0.02

0.38

40%

0.38

−0.17

0.92

60%

Emerging America

0.01

0.02

0.02

25%

0.10

0.07

0.43

50%

Emerging Europe

0.24

0.23

0.43

40%

0.99

0.52

1.70

20%

Emerging Asia

−0.31

−0.06

0.81

71%

−0.26

−0.22

0.28

86%

Developed Market

−0.15

−0.09

0.22

70%

−0.02

−0.13

0.49

70%

Emerging Markets

−0.06

0.00

0.61

50%

0.22

−0.11

1.07

56%

All

−0.11

−0.05

0.42

62%

0.08

−0.13

0.78

64%

Moreover, this formula is analyzed in Hypothesis 1 in Section III. In most of the regressions, the coefficients of the long-term bonds have smaller absolute values than the coefficients of short-term bonds. When compared with the short-term government bonds, the long-term bonds have less of an effect on the stock market in the same country Table 5.
Table 5

Detailed results of formula 2

Country

Adjusted R-Square

F Stat

DW Stat

VIF

Intercept

Lag of Stock

Short-term Bond

Long-Term Bond

Short-term Bonda Crisis Dummy

Long-Term Bonda Crisis Dummy

Argentina

0.10

9.16b

1.92

9.73

0.00

0.13a

−0.12a

0.07

0.21b

−0.34b

Austria

0.05

6.37b

2.00

6.34

0.00

0.16b

0.41

−0.13

−1.72

−0.19

Australia

0.09

10.59b

1.98

6.75

0.00a

0.07

0.06

−0.11

−2.37b

0.13

Belgium

0.04

5.61b

2.00

7.60

0.00

0.19b

−0.13

−0.21

−0.74

0.18

Brazil

0.05

4.56b

1.97

6.18

0.00

0.09a

3.01b

0.04

−2.64a

0.20

Canada

0.04

5.43b

1.98

4.90

0.00a

0.10a

0.05

−0.02

−2.15b

−0.04

Switzerland

0.01

1.74

1.98

4.85

0.00

0.02

−0.40

−0.10

−1.39

0.10

Chile

0.02

2.67a

1.98

5.95

0.00a

0.16b

0.48

0.13

−0.48

−0.24

China

0.07

5.58b

2.03

14.89

0.00

0.20b

−6.89

−0.43

8.31

−0.82

Czech

0.07

8.79b

2.04

5.87

0.00

0.26b

0.54

−0.14

−3.84b

0.26

Germany

0.07

8.77b

2.05

5.59

0.00b

0.03

0.05

−0.34a

−3.74b

0.26

Denmark

0.05

6.82b

2.03

7.85

0.00a

0.13b

0.01

−0.18

−2.72b

0.16

Spain

0.02

2.68a

2.01

8.97

0.00

0.11a

0.93

−0.20

−1.11

0.37

Finland

0.06

7.47b

2.01

6.55

0.00

0.12b

1.27a

−1.04b

−3.33b

0.78a

France

0.05

5.93b

2.01

5.53

0.00

0.03

0.82

−0.26

−3.46b

0.10

Greece

0.09

10.51b

1.99

34.89

0.00

0.25b

1.27

−0.33

−1.40

0.52a

Hungary

0.08

9.77b

1.96

10.13

0.00

0.17b

0.56

0.01

1.09

−0.22

Indonesia

0.05

5.76b

1.98

5.96

0.00b

0.10a

−0.02

0.04

0.03

−0.26a

Ireland

0.03

4.32b

1.97

17.95

0.00

0.18b

−0.23

0.00

0.41

−0.25

Israel

0.02

2.98a

1.99

3.13

0.00

0.15b

−0.04

0.17

−0.03

−0.22

India

0.03

4.21b

2.00

6.65

0.00a

0.17b

0.17

−0.22

−0.15

0.05

Italy

0.02

3.33b

1.98

6.97

0.00

0.14b

−0.64

0.03

−0.12

0.23

Japan

0.04

4.71b

1.99

2.97

0.00

0.10a

0.63

−0.49a

−2.62

−0.08

Korea

0.03

4.04b

1.99

11.42

0.00

0.18b

−0.26

−0.28

−1.10

1.37

Mexico

0.02

2.16a

1.99

368.46

0.00

0.09

0.23

0.30

−0.23

−0.17

Netherlands

0.08

10.35b

2.04

5.96

0.00a

0.07

−1.08

−0.23

−3.04b

0.14

Norway

0.04

4.70b

2.03

2.78

0.00a

0.12b

−0.18

0.06

−1.89a

−0.33

New Zealand

0.08

9.22b

1.98

10.65

0.00

0.28b

0.41

0.04

−0.47

−0.03

Poland

0.03

3.93b

1.99

4.66

0.00

0.13b

0.35

0.22

−0.97

0.18

Portugal

0.03

3.66b

1.99

16.87

0.00

0.17b

0.87

−0.18

−0.85

0.13

Russia

0.06

32.80b

2.23

900.75

0.00

−0.01

−0.01

0.09

0.11

−0.11

Sweden

0.04

4.95b

2.02

6.68

0.00a

0.04

0.28

−0.37

−3.01a

0.26

Thailand

0.03

4.00b

2.03

4.46

0.00a

0.17b

−0.07

0.02

0.07

−0.09

Turkey

0.01

4.24b

1.95

832.23

0.00

0.12

0.98

0.00

−0.98

0.00

Taiwan

0.05

6.87b

1.96

11.65

0.00

0.12b

−0.07a

0.02

0.06a

0.21b

UK

0.04

5.07b

1.98

6.46

0.00

0.09a

2.27b

−0.43a

−3.82b

0.45a

US

0.07

8.45b

1.99

6.48

0.00a

0.08a

−0.67

−0.08

−2.40a

0.00

Vietnam

0.12

9.29b

1.94

31.78

0.00

0.34b

−0.94

0.92

0.87

−1.00

South Africa

0.01

2.38a

1.97

6.42

0.00

0.09a

0.83

0.10

−0.35

−0.01

ameans significance with 10% confidential interval. bmeans significance with 1% confidential interval

To begin with, the coefficients of short-term government bonds are positive during non-crisis periods, while become significantly more negative during the crisis period for most regressions. This change is especially significant in regressions for developed countries. Even for those regressions, the coefficients of short-term government bonds remain positive, while the coefficients of short-term government bonds are also somewhat smaller than the related coefficients during the non-crisis period, which means the interdependence between the stock markets and the short-term government bond markets become negative or less positive during the crisis period, though they are positive during the non-crisis period. This suggests that more investments flow from the stock markets to the short-term government bond markets during a financial crisis due to the higher quality of short-term government bonds when compared to those of stocks in the same country.

Furthermore, no matter whether a crisis period or non-crisis period is in progress, the coefficients of long-term government bonds remain negative in most regressions, which shows that investment in long-term government bonds is strongly related to the prosperity of both the local economy and financial markets. During financial crisis periods, the interdependence between the stock markets and long-term government bond markets is still similar to the interdependence in the non-crisis periods. The interdependence between the stock markets and the long-term government bond markets is often negative due to different risk patterns.

Therefore, it is possible for the interdependence between the stock market, and the short-term and long-term government bond markets to be much higher during crisis periods than during non-crisis periods, which Hypothesis 1 already proves. In addition, it is necessary to look into different types of interdependence and determine their implications.

Different types of interdependence in the developed markets and the emerging markets

In this section, the author models different types of interdependence between the stock markets and the government bond markets during the crisis period using Formula (2). The F-statistics of the 39 regressions are significant in the 10% confidential interval. Therefore, there is no significant auto-correlation in the error series of these regressions. In addition, it should be noted that VIF factors are smaller than 10 in most of the 39 regressions, except the models for Mexico, Russia and Turkey, because the financial markets are less efficient in these three emerging countries. For most countries, however, the estimations in the regressions are remarkable and reliable.

From most of the estimated results from Formula (2), the short-term government bond markets are positively correlated with stock markets in the same country during non-crisis periods. Furthermore, their interdependence becomes negative for most of the 39 countries during crisis periods, which is consistent with the definition of flight to quality. Unlike the short-term government bond markets, the long-term government bond markets are negatively correlated with the stock markets. This pattern is more significant for developed markets; however, the interdependence between the stock markets and the long-term government bond markets becomes positive for most countries during crisis periods. This is noteworthy, since the Subprime Crisis and the European Debt Crisis are not only financial crises, but also economic crises. Once a depression appears, the quality of the long-term government bonds will be more affected than the short-term government bonds because short-term government bonds are more liquid. The performance of the developed American market is different from most of the other regions. The short-term and the long-term government bond markets are both negatively dependent on the stock markets, since America dominates the market and its government bonds are regarded as the country’s least risky financial asset, even though the U.S. was the source of the U.S. subprime mortgage crisis. Therefore, some investments flow into the U.S. government bond market from the stock market as a result of risk aversion.

Based on the definitions in the List of Tables:

Table 1: the characteristics of different types of interdependence are summarized in Table 5 and Fig. 1. There is neither a flight from quality (FFQ) nor positive spillover (PS) shown in Table 6, which proves that Hypothesis 2 is true.
Fig. 1

Significant interdependences between the stock and the government bond markets in the crisis

Table 6

Statistics to the results of Formula 2

Regions

Short-term Bond

Short-term Bond

Short-term Bond

Short-term Bond

Short-term Bond* Crisis Dummy

Short-term Bond* Crisis Dummy

Short-term Bond* Crisis Dummy

Short-term Bond* Crisis Dummy

Mean

Median

Standard Error

Negative Coefficient

Mean

Median

Standard Error

Negative Coefficient

Developed America

−0.31

−0.31

0.51

50%

−2.27

−2.27

0.18

100%

Developed Europe

0.35

0.16

0.85

38%

−2.00

−1.81

1.33

94%

Developed Asia

0.16

0.06

0.36

40%

−1.32

−1.10

1.14

100%

Emerging America

0.90

0.35

1.43

25%

−0.78

−0.35

1.27

75%

Emerging Europe

0.49

0.54

0.36

20%

−0.92

−0.97

1.85

60%

Emerging Asia

−1.00

−0.07

2.65

71%

1.26

0.06

3.13

29%

Developed Market

0.25

0.05

0.75

39%

−1.87

−1.89

1.24

96%

Emerging Markets

−0.06

0.20

2.00

44%

0.07

−0.06

2.52

50%

All

0.12

0.06

1.39

41%

−1.08

−0.98

2.08

77%

Regions

Long-Term Bond

Long-Term Bond

Long-Term Bond

Long-Term Bond

Long-Term Bond* Crisis Dummy

Long-Term Bond* Crisis Dummy

Long-Term Bond* Crisis Dummy

Long-Term Bond* Crisis Dummy

Mean

Median

Standard Error

Negative Coefficient

Mean

Median

Standard Error

Negative Coefficient

Developed America

−0.05

−0.05

0.04

100%

−0.02

−0.02

0.03

100%

Developed Europe

−0.24

−0.20

0.25

81%

0.18

0.17

0.28

19%

Developed Asia

−0.14

−0.11

0.26

60%

0.23

−0.03

0.65

60%

Emerging America

0.13

0.10

0.12

0%

−0.14

−0.20

0.23

75%

Emerging Europe

0.04

0.01

0.13

40%

0.02

0.00

0.20

40%

Emerging Asia

0.06

0.02

0.42

29%

−0.27

−0.09

0.46

71%

Developed Market

−0.20

−0.18

0.25

78%

0.18

0.13

0.37

35%

Emerging Markets

0.07

0.04

0.28

25%

−0.15

−0.10

0.35

63%

All

−0.09

−0.08

0.29

56%

0.04

0.05

0.39

46%

* refers to the multiple symbol between two explanatory variables, i.e. the item with * inside is an interaction term

To make different region’s markets more comparable, the number of the countries is standardized by dividing the total sample number in different regions. The flight to quality occurs from the stock markets to the long-term and short-term government bond markets in a majority of selected developed countries. However, regarding the emerging markets, in Asian countries the flight to quality only occurs between the stock markets and the long-term markets. Furthermore, in emerging European counties and a few developed European countries (i.e., Italy and Greece), regardless of whether the market is developed or emerging, the negative spillover only happens between the stock markets and the government bond markets. The negative spillover in Europe suggests that the investors lost their confidence in the recovery and solvency of the European countries during the Subprime Crisis and the European Debt Crisis periods. In addition, there is no flight from quality or positive spillover in these two crises. To summarize, all the negative spillovers occurred between the stock markets and the long-term government bond markets while the flight to quality emerged between the stock markets and the short-term government bond markets. For example, the remarkable flight to quality between the stock market and the long-term government bond market appears in the American market because the U.S. government bond is considered to be the safest asset in times of crisis. The long-term government bonds are less liquid than the short-term government bonds, such that the flight to quality between the stock market and the short-term government bond markets is of great significance, which proves that Hypothesis 3 is true.

Furthermore, the implications of the above results were analyzed by comparing the CDX indices to the returns of the government bonds. The CDXis an index on Credit Default Swaps of the government bonds, which is publicized by the Markit Company. In addition, the CDS is a popular financial derivative that uses the potential as the bond. In the CDS transaction, the buyer pays payments periodically to the seller and gains the right to sell the underlying bond to the seller at par when default occurs. Thus, a higher CDS spread causes a higher CDX index, which indicates a higher possibility of default for the related bonds. The correlation between the changing rate of the CDXs and the returns of the government bonds is positive in most developed countries, but negative in emerging markets. This is especially true in countries with high sovereign risks, such as the emerging European markets and Latin America. The significantly negative correlation between the changing rate of the CDXs and the returns of the government bonds exists, which means that the government bond is not a choice for risk aversion in countries with high sovereign risks (Table 7). That is why, in times of financial crisis, these countries suffer from negative spillover instead of flight to quality between the local stock markets and the government bond markets.
Table 7

Increasing Ratios of sovereign debts in the two crises

Regions

Mean of increasing in 2008-2009

Median of increasing in 2008-2009

Faster increasing in 2008–2009 than in 2006-2007

Mean of increasing in 2010-2011

Median of increasing in 2010-2011

Faster increasing in 2010–2011 than in 2006-2007

Developed America

6.7%

6.7%

100%

10.2%

10.2%

100%

Developed Europe

8.7%

8.3%

88%

8.5%

7.6%

94%

Developed Asia

12.8%

8.7%

60%

19.4%

8.4%

80%

Emerging America

4.6%

11.3%

33%

18.1%

11.4%

67%

Emerging Europe

9.7%

10.8%

100%

31.6%

28.6%

100%

Emerging Asia

13.4%

10.5%

100%

21.0%

23.2%

100%

Developed Market

9.4%

8.7%

83%

11.0%

8.4%

91%

Emerging Markets

9.0%

10.8%

83%

21.6%

11.4%

83%

All

9.0%

9.4%

81%

14.4%

10.6%

89%

Due to the implication from Beber et al. (Beber et al. 2009), the sharp changes in sovereign yield spreads is explained using differences in credit quality, though liquidity plays a non-trivial role, especially for low credit risk countries and during times of heightened market uncertainty. This result is similar to the above proof for Hypothesis 3. In addition, Beber et al. (Beber et al. 2009) suggests that the destination of large capital flows into the bond market is determined almost exclusively by liquidity. The above results supplement our studies (Table 8).
Table 8

Statistics to the different interdependences

Countries

Two Crises

US Subprime Crisis

European Debt Crisis

Robustness Test

Short-term Bond

Long-Term Bond

Short-term Bond

Long-Term Bond

Short-term Bond

Long-Term Bond

Short-term Bond

Long-Term Bond

Argentina

        

Austria

 

FTQ

 

FTQ

NS

FTQ

  

Australia

        

Belgium

FTQ

 

FTQa

   

FTQ

 

Brazil

 

NS

 

NSa

   

NS

Canada

 

FTQ

 

FTQ

 

FTQ

FTQa

FTQ

Switzerland

FTQa

 

FTQa

 

FTQa

FTQ

FTQ

 

Chile

        

China

 

FTQ

 

FTQ

 

FTQ

 

FTQa

Czech

        

Germany

      

FTQa

 

Denmark

      

FTQa

 

Spain

        

Finland

        

France

     

FTQ

  

Greece

      

FTQ

NSa

Hungary

NSa

 

NSa

  

NS

NSa

 

Indonesia

       

FTQa

Ireland

      

NS

FTQa

Israel

FTQ

 

FTQ

 

FTQ

   

India

      

NS

FTQa

Italy

FTQ

NS

FTQa

NS

 

NSa

FTQa

NSa

Japan

 

FTQa

  

NS

FTQa

NS

 

Korea

FTQ

 

FTQ

  

FTQ

FTQ

NS

Netherlands

FTQa

 

FTQa

 

FTQa

FTQ

FTQa

 

Norway

FTQa

 

FTQ

 

FTQa

NS

FTQa

 

New Zealand

  

NS

   

NS

 

Poland

 

NS

 

NS

   

NSa

Portugal

   

FTQ

   

FTQa

Sweden

      

FTQa

 

Thailand

       

FTQ

Taiwan

 

NSa

 

NSa

 

NSa

FTQa

NSa

UK

        

US

FTQa

FTQ

FTQa

 

FTQ

FTQ

FTQa

 

Vietnam

        

South Africa

   

NS

  

NS

NS

‘FTQ’ stands for Flight to Quality while ‘NQ’ stands for Negative Spillover

astands for significance in 10% confidential interval

Different types of interdependence in the two latest crises

This section examines different kinds of interdependence in the U.S. subprime mortgage crisis and the European Debt Crisis on the basis of Formula (3) (Table 9). This is similar to the previous section, which states the F-statistics are significant in the 10% confidential interval level for all the 39 regressions and there is no significant autocorrelation for the errors of 39 regressions. In addition, the estimation of the 39 regressions is significant and reliable.
Table 9

Correlations between CDX and returns of government bonds

Regions

Mean

Median

Standard Error

Negative Value

Mean

Median

Standard Error

Negative Value

Developed America

0.02

0.02

0.29

50%

0.01

0.01

0.16

50%

Developed Europe

0.06

0.27

0.42

38%

0.03

0.12

0.36

44%

Developed Asia

0.20

0.08

0.22

0%

0.18

0.13

0.23

0%

Emerging America

−0.23

−0.17

0.15

100%

−0.30

−0.21

0.17

100%

Emerging Europe

−0.21

−0.26

0.23

75%

−0.19

−0.27

0.37

75%

Emerging Asia

−0.05

−0.02

0.12

67%

0.01

0.06

0.23

33%

Developed Market

0.09

0.22

0.37

30%

0.06

0.13

0.32

35%

Emerging Markets

−0.13

−0.17

0.28

75%

−0.14

−0.23

0.35

67%

All

−0.02

0.04

0.37

49%

−0.03

0.01

0.35

49%

The interdependence between the stock markets and the government bond markets shows some similar patterns in the Subprime Crisis and the European Debt Crisis. During the Subprime Crisis the short-term government bond markets are negatively correlated with the stock markets in the same countries during the, which is consistent with the definition of flight to quality. However, the short-term government bond markets become positively correlated with the stock markets in certain developed European countries (i.e., Italy and Greece) and all emerging markets (Table 10), which shows that the flight to quality only occurs in developed countries, while negative spillover appears in the emerging markets and Europe’s PIIGs countries, which is consistent with Hypothesis 4.
Table 10

Statistics to the results of Formula 3

Regions

Short-term Bond

Short-term Bond

Short-term Bond*Subprime Crisis

Short-term Bond* Subprime Crisis

Short-term Bond*European Crisis

Short-term Bond* European Crisis

Mean

Negative Coefficient

Mean

Negative Coefficient

Mean

Negative Coefficient

Developed America

−0.32

50%

−2.88

100%

−0.93

100%

Developed Europe

0.30

38%

−3.66

94%

−1.32

75%

Developed Asia

0.15

40%

−1.70

80%

0.12

60%

Emerging America

0.86

25%

−1.30

75%

−0.74

75%

Emerging Europe

0.26

20%

−0.70

40%

−0.73

60%

Emerging Asia

−1.01

71%

1.04

29%

1.52

29%

Developed Market

0.22

39%

−3.16

91%

−0.97

74%

Emerging Markets

−0.14

44%

−0.09

44%

0.25

50%

All

0.07

41%

−1.90

72%

−0.47

64%

Regions

Long-Term Bond

Long-Term Bond

Long-Term Bond* Subprime Crisis

Long-Term Bond*Subprime Crisis

Long-Term Bond* European Crisis

Long-Term Bond* European Crisis

Mean

Negative Coefficient

Mean

Negative Coefficient

Mean

Negative Coefficient

Developed America

−0.05

100%

0.50

50%

1.00

100%

Developed Europe

−0.24

81%

0.25

25%

0.31

31%

Developed Asia

−0.13

60%

0.40

40%

0.80

80%

Emerging America

0.13

0%

0.75

75%

1.00

100%

Emerging Europe

0.01

40%

0.20

20%

0.40

40%

Emerging Asia

0.07

29%

0.57

57%

0.71

71%

Developed Market

−0.20

78%

0.30

30%

0.48

48%

Emerging Markets

0.06

25%

0.50

50%

0.69

69%

All

−0.09

56%

0.38

38%

0.56

56%

* refers to the multiple symbol between two explanatory variables, i.e. the item with * inside is an interaction term

Furthermore, this section focuses on different types of interdependence in the crisis periods. Figures 2 and 3 are both standardized as Fig. 1. In these two crises, the flight to quality between the stock markets and the short-term bond markets occurs more frequently in developed markets. However, in emerging Asian markets the flight to quality only occurs between the stock markets and the short-term government bond markets. However, the flight to quality occurred less frequently in the European Debt Crisis due to the worse quality of the European sovereign debts, which proves Hypothesis 5 is true. Additionally, the negative spillover also occurred more frequently during the Subprime Crisis than in the European Debt Crisis. However, during the European Debt Crisis, the negative spillover was more frequent between the stock markets and the long-term government bond markets in the European and Asian developed markets. The developed European countries that suffered from negative spillover during the two crises are the PIIGs countries, Greece and Italy in particular. To summarize, the above analysis proved that both the flight to quality and the negative spillover occur more frequently in the very beginning of the crisis. Additionally, the countries that suffered more from the crisis (such as the PIIGS countries) will also experience more frequent negative spillovers in the government bond markets and the stock markets. Similar to the conclusion drawn in the previous section, the developed American markets always have significant flight to quality since the U.S. government bond is regarded as the safest asset during times of crisis.
Fig. 2

Significant interdependences between the stock and the government bond markets in US Subprime crisis

Fig. 3

Significant interdependences between the stock and the government bond markets in European Debt Crisis

To show more evidence in the above conclusion, the above results of interdependence are compared with the sovereign debt in the related countries. For the purposes of this paper, we consider 2008 to 2009 as the period of the Subprime Crisis while 2009 to 2011 is regarded as the European Debt Crisis period. For purposes of comparison, the years between 2006–2007 have been chosen as the non-crisis period. The sovereign debts increased in all the emerging market countries during the two crises, but especially during the European Debt Crisis. Therefore, in emerging markets government bonds are more illiquid and risky than bonds in developed markets. Therefore, negative spillover occurs more frequently than flight to quality in these emerging markets. Once the European Debt Crisis worsened and had a greater effect on Europe’s developed countries, the flight to quality also appeared less frequently in those developed countries. Different kinds of interdependence implicate the different quality of sovereign debts in both the developed and emerging markets.

Robustness tests

To conduct the robustness test, Formula (2) is reestimated by defining the crisis period with the crisis status in the Markov-switching Regression [Formula (1)].

In the robustness test, the crisis period is defined using data and statistics instead of the pre-decided definition, which makes the estimation of Formula (2) more significant. In addition, the directions of the coefficients are consistent with the previous results. To be specific, in most of the developed markets, the flight to quality occurred between the stock markets and the short-term government bond markets. However, in the emerging markets, there was little flight to quality evident between the stock markets and the short-term government bond markets. In certain emerging Asian markets, the flight to quality was seldom evident between the stock markets and the long-term government bond markets. Second, with the exception of the developed American market, there is significant negative spillover between the stock markets and the long-term government bond markets. This is in line with the previous conclusions.

Conclusion

This work investigated different types of interdependence between the stock and the government bond markets in the same country, and special patterns of interdependence during the Subprime Crisis period and the European Debt Crisis period were summarized. In this paper, we suggest that risk aversion becomes the main cause of investment flowing across different regions during times of financial crisis. Once the financial crisis occurs, most investors prefer bonds to stocks because, in the same region, stock markets are riskier than the government bonds. This is why neither the flight from quality nor the positive negative occurs during times of crisis. Furthermore, investors tend to choose safer markets based on their judgment of the safe haven. Generally speaking, the cash flows are more likely to occur in the government bond markets of developed countries. In particular, government bonds in the emerging markets and the long-term government bonds in the sovereign debt crisis are not deemed to be safe assets, so they do not attract investors during crises.

To be specific, we have concluded the following items for interdependence:
  1. 1.

    During non-crisis periods, the short-term government bond markets and the stock markets are positively interdependent in most countries. However, during crisis periods, their interdependence becomes negative and leads to flight to quality, since the government bond markets involve less risk and more funds may flow from the stock markets to the government bond markets for risk aversion.

     
  2. 2.

    Compared with the short-term government bond market, a long-term government bond market is rarely dependent on the stock market in the same country because it is significantly correlated with the long-term trend of the local economy. During the crisis period, the interdependence between these two markets remains negative, especially for the countries with a higher sovereign debt risk.

     
  3. 3.

    During times of crisis, the flight to quality is more popular than the negative spillover between the stock markets and the government bond markets in developed countries. The negative spillover often occurs in the emerging markets as well as the PIIGS countries, which are developed European countries but ones that suffer from the high sovereign risk during crisis periods.

     
  4. 4.

    The flight to quality or negative spillover occurs more frequently at the beginning of a crisis. The developed American markets always witness significant flight to quality since the U.S. government bond is considered to be the safest asset during times of crisis.

     

Furthermore, this paper presents two implications for risk management. To be specific, it is first important to measure the quality gap between different assets in the same country. Timely monitoring with concern to the quality gap between different assets would be beneficial to risk management regarding uncertain cash flow across different financial markets. Secondly, the government should keep the policy stable to provide international investors more confidence in local financial markets, since having more faith in governmental policy would reduce the magnitude and destructiveness of the international cash flow.

Declarations

Acknowledgement

This paper is funded by China Postdoctoral Science Foundation (2014M550985) and National Science Foundation of China (71532013, 71528001).

Authors’ contributions

KC, carried out the data arrangement, conceived of the study, performed the statistical analysis and drafted the manuscript. XY, participated in the design of the study and helped to draft the manuscript. Both authors read and approved the final manuscript.

Competing interests

None of the authors have any competing interests in the manuscript.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors’ Affiliations

(1)
China University of Petroleum
(2)
Academy of Mathematics and Systems Sciences, China’s Academy of Sciences

References

  1. Aslanidisa N, Christiansen C (2012) Smooth transition patterns in the realized stock–bond correlation [J]. J Empir Financ 19(4):454–464View ArticleGoogle Scholar
  2. Baur D, Lucey B (2009) Flights and contagion - an empirical analysis of stock–bond correlations [J]. J Financ Stab 5(4):339–352View ArticleGoogle Scholar
  3. Beber A, Brandt MW, Kavajecz KA (2009) Flight-to-quality or flight-to-liquidity? evidence from the euro-area bond market [J]. Rev Financ Stud 22(3):925–957View ArticleGoogle Scholar
  4. Brièrea M, Chapelle A, Szafarz A (2012) No contagion, only globalization and flight to quality [J]. J Int Money Financ 31(6):1729–1744View ArticleGoogle Scholar
  5. Bernanke B, Gertler M, Gilchrist S (1996) The financial accelerator and the flight to quality [J]. Rev Econ Stat 78(1):1–15View ArticleGoogle Scholar
  6. Caballero RJ, Krishnamurthy A (2008) Collective risk management in a flight to quality episode [J]. J Financ 63(5):2195–2230View ArticleGoogle Scholar
  7. Chan K, Treepongkarunab S, Brooksc R, Gray S (2011) Asset market linkages: Evidence From financial, commodity and real estate assets [J]. J Bank Financ 35(6):1415–1426View ArticleGoogle Scholar
  8. Cheng K, Lu F, Yang X (2011) Copula contagion index and its efficiency. Appl Financ Econ 22(12):989–1002View ArticleGoogle Scholar
  9. Forbes KJ, Rigobon R (2002) No contagion, only interdependence - measuring stock market comovements [J]. J Financ 57(5):2223–2261View ArticleGoogle Scholar
  10. Gulko L (2002) De-coupling [J]. J Portfolio Manage 28(3):59–66View ArticleGoogle Scholar
  11. Hamilton J (1989) A new approach to the economic analysis of nonstationary time series and the business cycles [J]. Econometrica 57:357–384View ArticleGoogle Scholar
  12. Krolzig H (2004) Constructing turning point chronologies with Markov-switching vector autoregressive models: the euro-zone business cycle[M]. Eurostat Colloq on Mod Tools Bus Cycle Analy. 1:147–190.
  13. Lang WW, Nakamura LI (1995) Flight to quality in banking and economic activity [J]. J Monet Econ 36(1):145–164View ArticleGoogle Scholar
  14. Vayanos D (2004) Flight to quality, flight to liquidity, and the pricing of risk [R]. National Bureau of Economic Research (No. w10327). http://www.nber.org/papers/w10327

Copyright

© The Author(s). 2017

Advertisement