_________________________________
* I thank Enrico Perotti, Esther Eiling, Frank de Jong and Radomir Todorov for their valuable comments. Any errors or omissions remain the sole responsibility of the author.
Global Financial Transmission of Economic Policy Uncertainty:
Evidence from the U.S. and Europe*
Author: Ran Zhu
Supervisor: Prof. Dr. E.C. Perotti
University of Amsterdam
Summer 2014
Abstract
Using quantification of economic policy uncertainty, we show economic policy uncertainty (EPU) in a home country has a negative effect on domestic equity returns in both the U.S. and Europe. The financial market incorporates uncertainty effects into prices in a short period. In addition, policy uncertainty spillover effects have large positive long run effects on equity returns via bond and equity flows between the U.S. and Europe. The U.S. EPU spillover effect is larger than the European EPU effect. Our results suggest that financial markets are more sensitive to U.S. policy uncertainty than European EPU.
JEL Classification: F21, F32, F42, G12, G18
2
1. Introduction
Since the global financial crisis, economic policy uncertainty has become prominent in developed economies. Popular newspapers are awash with reports about stock market reactions to possible changing economic policies in both their countries as well as foreign countries. Unfortunately, popular interest has not been matched by academic research. This paper fills that gap by conducting an empirical analysis of the global financial transmission of economic policy uncertainty between the U.S. and Europe.
Theoretically, increased economic policy uncertainty in a home economy would decrease domestic stock market return and increase volatility. However, the financial market will require a higher risk premium due to a persistently high level of economic policy uncertainty. In addition, changing the economic policy uncertainty for a sizeable country would not only influence domestic capital markets, but also have an international spillover effect to other capital markets. One potential way in which this occurs is that a less predictable economic policy environment impedes a home country’s growth expectation, which decreases the attractiveness of domestic investment. That is, policy uncertainty in domestic economy could result in investors shifting their investments to other capital markets. Another argument is that there could be lower demand from main import countries due to higher policy uncertainty; therefore lower economic growth expectation for main export countries. As a result, this lower growth would drive down stock performance in those countries. Apparently, there is a strong relationship between economic policy uncertainty and financial market performance.
Using data from the U.S. and Europe, we find that increases in economic policy uncertainty in home countries tend to reduce domestic equity returns. Conversely, increases in U.S. economic policy uncertainty tend to increase equity returns in Europe. Furthermore, U.S. economic policy uncertainty transfer through portfolio inflows into Europe is associated with sizeable positive stock returns in Europe, and a similar pattern holds for the effect of European economic policy on the U.S. equity market. The effect becomes more economically significant after taking into account the safe haven effect. These empirical findings are consistent with traditional economic theory. Both bond and equity inflows caused by policy uncertainty should generate positive effects on stock markets by direct (equity inflows) or indirect effects (bond inflows). The direct effect refers to increasing demand for equity that will drive up equity price as well as returns. The indirect effect refers to increasing demand on bonds that will reduce the cost of debt for
3
enterprises, and thus increase equity values and returns. According to the author’s knowledge, the present paper is the first to formally test the international transmission of economic policy uncertainty to equity returns among developed economies. The main contribution is to examine the global financial transmission of economic policy uncertainty to equity markets through different channels within advanced economies.
There is a modest amount of research relating the domestic effect of economic policy uncertainty to financial markets. For a theoretical framework, Pastor and Veronesi (2012) provide a general equilibrium model that predicts that the policy uncertainty commands a risk premium, and it is larger in a weaker economy. This provides theory-based arguments that financial markets require a risk premium to compensate for higher economic policy uncertainty.
For an empirical framework, one of the crucial issues is how to measure economic policy uncertainty. Traditionally, there are two different approaches to measure economic policy uncertainty. One method is to use event study as a proxy of economic policy uncertainty. Belo, Gala, and Li’s (2011) research uses elections as a proxy of economic policy uncertainty to examine the relationship between economic policy uncertainty and cross-section stock market returns. They find that there is a positive relationship between a company’s exposure to government spending and stock returns across various industries during Democratic presidencies, while there is a negative relationship during Republican presidencies. Julio and Yook (2013) study the link from policy uncertainty to cross board capital flows by using election cycles. They show that policy uncertainty decreases foreign direct investment from U.S firms to other countries. Gao and Qi (2013) find that higher policy uncertainty during election periods sharply increases yields of municipal bonds, especially in times of economic stress. Santa-Clara and Valkanov (2003) link stock market returns to political cycles. Kelly, Pástor and Veronesi (2014) examine the political uncertainty effect on option pricing by using national elections and global summits as measurements of uncertainty. The main advantage of using event study is that it allows the researchers to precisely draw up a timeline and incorporate effects from the event. However, since each event is different and those studies only capture short intervals, the results have limited generalizability.
The second method is the quantification of the economic policy uncertainty. Typically, researchers use search-based measurements to quantify economic policy uncertainty. Baker, Bloom and Davis (2013) create an economic policy index by using a search-based model. The
4
index reflects the policy uncertainty from three perspectives: the percentage of mentions in newspapers of policy uncertainty; tax code provision in future years; and disagreement among economic forecasters over government institutions. Using similar methodology, Brogaard and Detzel (2012) examine the relationships between economic policy uncertainty and stock return, volatility, risk premium, and cash flow. They find that economic policy uncertainty is negatively correlated with both stock returns and aggregate cash flow. Brogaard and Detzel’s (2013) research using the economic policy uncertainty index (Baker, Bloom and Davis 2013) find that increased economic policy uncertainty is associated with lower stock returns in the United States. Wright (2011) highlights a positive relationship between inflation uncertainty and bond risk premiums by using forecasting disagreement to measure uncertainty.
The advantage of the quantification method is that it provides continuous measurement over time, which is better in terms of generalizability. However, quantification itself tends to be problematic and complicated. Both selection of search criterions and keywords vary results significantly.
This present research is also related to literature on the policy spillover effect from domestic economies to foreign economies. Forbes et al. (2012) show that policy uncertainty in countries that are perceived to conduct capital controls, such as Brazil, decreases capital inflow. They argue that the effect of capital controls on capital inflow is mainly due to investors’ changing expectations about future policies. Gauvin, McLoughlin and Reinhardt’s (2013) research shows that negative policy uncertainty shocks in the U.S and EU led to portfolio capital inflows into emerging markets, especially after the financial crisis. Fratzscher et al. (2012) found that global spillovers of U.S quantitative easing caused reallocation of global portfolios and that such a policy effect was unique.
The paper proceeds as follows. Section 2 introduces the data and the notation used in the paper. Section 3 discusses the empirical results of the domestic effect of economic policy uncertainty on equity returns. Section 4 presents the international spillover effect of economic policy uncertainty on stock returns between the U.S. and Europe. Section 5 shows the policy spillover effect on equity returns between the U.S. and Europe via portfolio shift channel. Section 6 gives the conclusion and some suggestions for future research.
5
2. Data
In this section, we describe the definitions and sources of data used in the research. For clarity, the data is categorized into economic policy uncertainty variables, equity market variables, international portfolio flows and control variables. All variables are at monthly frequency. For the economic policy uncertainty effect on stock performance within the U.S., the sample includes 229 monthly observations from December 1994 (1994:12) to December 2013 (2013:12). For the economic policy uncertainty effect on equity markets between the U.S. and Europe, the sample covers 156 months from January 2001 (2001:01) to December 2013 (2013:12). Table 1 provides summary statistics and data sources which are given in Appendix A. Table 1 Panel A shows variables that are used in domestic effect analysis, and Panel B presents variables that are applied in international spillover analysis. The portfolio flows are given in Panel C.
2.1. Economic policy uncertainty
The economic policy uncertainty measurement is taken from Baker, Bloom and Davis (2013). They create a new index to measure economic policy uncertainty by taking into account factors such as newspaper coverage of economic policy uncertainty, tax code expiration in the future, and forecasting disagreements between institutions. The newspaper coverage uses search-based measurements to count the frequency of newspaper exposure related to economic policy uncertainty. The tax code expiration measures the number of tax code provisions set to expire in the future. This component is only applicable to the U.S. economic policy uncertainty index. The forecasting disagreement component reflects different forecasts about future fiscal and monetary policy among federal, state and local governments.
2.2. Equity market variables
For the U.S. economic policy uncertainty effect on U.S. equity market, we use S&P 500 index return to test EPU implication for stock returns, as well as using 10 size decile portfolios to examine the effect due to the size of market capitalization. For economic policy uncertainty spillover effect between the U.S. and Europe, the MSCI U.S. and Europe equity index returns are used as proxies for equity market performance.
6 Table 1
Descriptive Statistics
Panel A presents the variables that are used for domestic effects between EPU and equity returns within the U.S. and Europe. The control factors include dividend yields, the term spread measured as a yield difference between 10-year government bonds and 3-month government bonds, the default spread calculated as the yield of BAA rated bonds minus the yield of AAA rated bonds, the relative interest rate which is the deviation of 3-month government bonds to 1-year averages, and the business cycle dummy for the U.S. and Europe (1 stands for recession). Panel B summarizes the additional global factors that are used for EPU global spillover effect between the U.S. and Europe. The global control variables include the world equity performance measured by MSCI World equity index returns, gold returns, yields of U.S. Treasury bills, percentage change exchange rate and risk appetite measured by the VIX index. All returns are expressed in monthly returns. Panel C shows the net portfolio flows between the U.S. and Europe. All amounts are shown in billions of dollars.
Panel A: Domestic Effect
Variables Mean Std. dev. Min Max Obs.
U.S.
U.S. economic policy uncertainty 106.62 37.69 57.20 245.13 229 S&P 500 index return 0.0072 0.0444 -0.1694 0.1077 229 MSCI U.S. index return 0.0053 0.0463 -0.1725 0.1083 203
Control factors
S&P 500 dividend yield 1.8497 0.4485 1.1100 3.6000 229
U.S. term spread 1.6996 1.1301 -0.6252 3.7753 229
U.S. Default spread 0.9928 0.4524 0.5500 3.3800 229
U.S. Relative interest rate 0.3086 0.2582 -0.4100 1.67 229
U.S. business cycle 0.1135 0.3179 0 1 229
Europe
European economic policy uncertainty 109.58 35.46 52.03 217.31 204 MSCI Europe index return 0.0050 0.0552 -0.2132 0.1316 156
Control factors
EU term spread 1.5929 1.0389 -0.6973 3.5202 156
EU default spread 0.7314 0.0379 0.4701 2.7530 156
EU relative interest rate 0.2364 0.2403 -0.2460 0.7644 156
EU business cycle 0.1469 0.2570 0 1 156
Panel B: International Spillover Effect Global factors
MSCI World index return 0.0045 0.0468 -0.1905 0.1091 203
Gold return 0.0069 0.0401 -0.1173 0.1738 203
U.S. 3 month T-bill 2.4514 2.0972 0.0100 6.1900 204
Change % USD to GBP 0.0004 0.0242 -0.0935 0.0882 203
Change % USD to EUR 0.0011 0.0297 -0.0972 0.0955 156
Change % USD to CHF 0.0025 0.0258 -0.1044 0.0779 203
VIX 22.21 8.33 10.82 62.64 148
Panel C: Portfolio flows
Europe bond net inflow -1125 495 -2508 -294 156
Europe equity net inflow -147 97 -552 -40 156
Change % Europe bond net inflow -0.0299 0.2180 -0.8118 0.4356 155 Change % Europe equity net inflow -0.0520 0.3236 -1.004 0.5285 155
7 2.3. International portfolio flow
The data of international portfolio inflows for bonds and equities between Europe and the U.S. is taken from the Treasury International Capital (TIC) system. It covers portfolio flows between the U.S. and Europe from 2001:01 to 2013:12. The portfolio data of Europe only includes countries from the European Monetary Union and the UK since European economic policy uncertainty indexes is based on large economies (Germany, the UK, France, Italy and Spain). The bond flows include both government bonds and corporate bonds.
2.4. Control variables
The conditioning variables are the dividend yield for stock index, the term spread measured as the difference in the yield of 10-year government bonds and the 3-month government bonds, the default spread calculated as the yield of BAA-rated corporate bonds minus the yield of AAA-rated corporate bonds, the relative interest rate which is the deviation of 3-month government bonds to 1-year average, and the business cycle indicators for the United States and Europe. The present research also controls for global factors such as the percentage change in exchange rate between EUR, GBP and USD, global risk appetite measured by the VIX index, world equity market performance based on MSCI World equity index returns. In order to separate the portfolio transmission effect and safe haven effect, we also control for safe haven assets performance. Safe assets we used in this research are Gold, U.S. 3 month Treasury bills, and Swiss Francs. All of these control variables have been commonly used by previous studies related to equity market performance and international spillover effect (Santa-Clara and Valkanov 2003; ***).
3. Domestic effect of economic policy uncertainty on equity returns
In this section, we study the domestic effect by examining the relationship between economic policy uncertainty and stock returns within the domestic economy. After controlling for the business cycle and equity return related factors, increasing economic policy uncertainty is associated with lower domestic stock returns. The empirical findings are consistent for both U.S. and European financial markets.8 Figure 1
Note: Figure 1 displays the time series data of U.S. economic policy uncertainty and S&P 500 index return from December 1994 to December 2013.
3.1. U.S. EPU effect on U.S. equity returns
In order to measure the relationship of U.S. economic policy uncertainty to stock returns, we employ regressions as following:
( ) ( ) ( )
where is the one-month holding period returns of S&P500 index; is the standardized economic policy uncertainty of the U.S. in month t; is the difference between and ; is the control variable matrix of U.S. domestic factors. Those factors comprise divide yield, term spread, default spread, relative interest rate, and business cycle indicators. The permanent effect is measured by long-run propensity ( ) while the short term effect is measured by the impact propensity ( ). If the policy uncertainty has no effect for stock returns, the coefficient of and should equal zero.
Table 2 shows the results of the overall effect of economic policy uncertainty on stock returns. The results are made robust by regressing changing economic policy uncertainty on equity returns. Column 1 shows that changing economic policy uncertainty is negatively correlated with stock returns in the short run. The results are both economically and statistically significant. Column 2
-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0 50 100 150 200 250 300 E co no m ic P o licy Uncer ta inty I nd ex
Correlation between U.S. EPU
and S&P 500 Index Return
U.S. EPU
9
lists the results after considering the control variables, and the results remain robust. Higher economic policy uncertainty will decrease stock returns in the short run while increase returns in the long run. Specifically, increasing standard deviation of economic policy uncertainty by one point is associated with a 4.05% reduction in stock returns. These results coincide with the theoretical argument that the adverse shock of economic uncertainty tends to increase volatility while decreasing stock returns. The positive long term effect of economic policy uncertainty is consistent with Pastor and Veronesi (2013), who also show that investors require a higher equity risk premium due to a high level of uncertainty, which leads to higher returns.
Table 2
Economic Policy Uncertainty and Equity Returns within U.S. market
The table reports estimates of the following regression:
( ) ( )
The dependent variable is the monthly returns from S&P 500 index. The sd_USEPU is the standardized U.S. economic policy uncertainty index (Baker, Bloom and Davis 2013). The domestic control variables are dividend yield, term spread measured as yield difference between 10-year Treasury notes and the 3-month Treasury bills, the default spread calculated as the yield of BAA-rated bonds minus the yield of AAA-rated bonds, the relative interest rate is the deviation of 3-month Treasury bills to 1-year average, and the NBER business cycle dummy of the U.S. (business cycle, 1 stands for recession); Standard deviations are reported in parentheses; */**/*** stand for statistical significance at the 10%, 5% and 1% levels. (1) (2) -0.0010 0.0080 (-0.32) (1.57) -0.0232*** -0.0274*** (-3.81) (-4.31) -0.0188*** -0.0211*** (-2.63) (-3.36) Dividend yield 0.0185** (2.57) Term spread -0.0023 (-0.68) Default spread -0.0376*** (-3.44)
Relative interest rate 0.0061
(0.39) Business cycle -0.0089 (-0.75) Constant 0.0070** 0.0134 (2.51) (0.78) N 227 227 Adj- 0.1140 0.2141
10
3.2. European EPU effect on European equity market returns
Figure 2 shows the correlation between the European economic policy uncertainty index and the MSCI Europe index monthly return time series. The overall correlation between the European economic policy uncertainty index and the MSCI Europe index return is -0.1177. European policy uncertainty is negative correlated with domestic equity returns, which is consistent with the relationship between EPU and equity returns in the U.S. financial market.
Figure 2
Note: Figure 2 displays the time series data of European economic policy uncertainty and MSCI Europe index return from January 2001 to December 2013.
To measure the relationship between economic policy uncertainty and equity returns within European financial markets, we run the following regression:
( ) ( ) ( )
where is the one-month holding period returns of the MSCI Europe equity index; is the standardized European economic policy uncertainty in month t; is the difference between and ; is the control variable matrix of European domestic factors. The permanent effect is measured by long-run propensity ( ) while the short term effect is measured by impact propensity ( ). If European policy uncertainty
-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0 50 100 150 200 250 E co no m ic po licy un ce rt a inty ind ex
Correlation between European EPU and
MSCI Europe Index Return
European EPU
11
has no effect for domestic stock returns, the coefficient of and should equal zero.
Table 3
Economic Policy Uncertainty and Equity Returns within European markets
The table reports estimates of the following regression:
( ) ( )
The dependent variable is the monthly returns from the MSCI Europe equity index. The sd_EUEPU is standardized European economic policy uncertainty index (Baker, Bloom and Davis 2013). The domestic control variables are term spread (EU term spread) measured as yield difference between 10-year ECB bonds and the 3-month ECB bonds, the default spread (EU default spread) calculated as the yield of BAA-rated bonds minus the yield of AAA-rated bonds, the relative interest rate (EU relative interest rate) which is the deviation of 3-month ECB bonds to 1-year average, and the business cycle dummy of Europe (EU business cycle, 1 stands for recession); Standard deviations are reported in parentheses; */**/*** stand for statistical significance at the 10%, 5% and 1% levels.
(1) (2) -0.0009 -0.0086 (-0.20) (-1.54) -0.0407*** -0.0328*** (-4.58) (-4.13) -0.0121 -0.0062 (-1.24) (-0.69) EU term spread 0.0158*** (3.07) EU default spread -0.0211*** (-2.97)
EU relative interest rate -0.0283*
(1.73) EU business cycle 0.0037 (0.90) Constant 0.0052 0.0132 (1.42) (1.51) N 155 155 Adj- 0.1140 0.1976
Table 3 shows the European economic policy uncertainty effect on domestic equity returns. In general, we find that increasing policy uncertainty in Europe decreases domestic equity returns. In particular, after controlling for other relevant factors, a one-point increase in the standard deviation of European EPU results in a 4.76% decrease in European stock returns. Hence, economic policy uncertainty has a larger impact on domestic equity returns in Europe than in the United States. However, unlike the U.S., one-month-lagged changing European policy
12
uncertainty has minor economic effects and is not statistically significant. In addition, the major impact of European policy uncertainty on domestic equity returns occurs within one month.
4. EPU spillover effect between the U.S. and Europe
Beyond domestic effect analysis, the present research seeks to understand how economic policy uncertainty from a major economy will impact other countries. In particular, we investigate the average effect of Europe stock market reactions to economic policy uncertainty changes in the U.S., as well as European economic policy uncertainty to U.S. stock returns.
In principle, economic policy uncertainty from a large economy could affect stock returns in another economy through several mechanisms. First, the policy uncertainty spillover effect can go through the corporate finance channel. For example, increasing economic policy uncertainty in the U.S. would be associated with a higher interest rate, which would raise the cost of debt for European enterprises financing themselves on the U.S. bond market. As a result, higher financing cost reduces firms’ value as well as equity returns. Similarly, Pástor and Veronesi (2013) argue that increasing economic policy uncertainty in the U.S. will lead to higher risk premiums in the U.S. equity market, which raises the cost of capital for European firms who finance themselves on the U.S. equity market. Hence, a higher cost of capital causes lower equity returns.
Second, domestic economic policy uncertainty can affect foreign stock markets by changing demand expectation. A less predictable economic policy environment in the U.S. would impede growth prospects as well as domestic demand, which reduces the profitability of European enterprises that run businesses in the United States. Those European firms’ stock will adjust to such a lower demand expectation and profitability by lowering stock price.
Third, the global financial transmission of economic policy uncertainty can follow the portfolio shift channel. That is, a high level of economic policy uncertainty in the U.S. would hinder domestic economic growth expectations; therefore, investors might shift their investments from U.S to European equity markets. A higher demand for European stocks will drive up the stock price as well as returns in that month. Alternatively, for the same reason, investors could also change their investment portfolios by purchasing European bonds instead of U.S. assets.
Figure 3 presents the overall correlations between standardized U.S. and European EPU indexes. The correlation between the two EPU indexes is 0.8589, which indicates that there is a significant global integration of policy uncertainty in today’s world economy.
13 Figure 3
Note: Figure 3 displays the time series data of U.S. economic policy uncertainty and European economic policy uncertainty index from January 1997 to December 2013.
To examine the global spillover effect from U.S. economic policy uncertainty to European stock market performance and vice versa, we apply the following regression models:
( ) ( )
( ) ( ) ( )
( ) ( )
( ) ( ) ( )
where and are the one-month holding period returns of the MSCI U.S. and Europe indexes; is the standardized economic policy uncertainty; is the difference of and . is the control variable matrix. The long term spillover effect is measured by long-run propensity ( ) while the short term spillover effect is measured by the impact propensity ( ). If the U.S policy uncertainty has no effect for European stock returns, the coefficient of and in equation (3) should equal zero.
-2 -1 0 1 2 3 4 E co no m ic P o licy Uncer ta inty I nd ex
Correlation between U.S. EPU
and European EPU
Standardized_U.S. EPU Standardized_European EPU
14 Table 4
U.S. Policy Uncertainty Spillover Effect on European Stock Market
The table reports estimates of the following regression:
( ) ( ) ( )
( )
The dependent variable is the monthly returns of the MSCI Europe index. The sd_USEPU and sd_EUEPU are the standardized economic policy uncertainty indexes of U.S. and Europe (Baker, Bloom and Davis 2013). The control variables of domestic factors are term spread (EU term spread) measured as yield difference between 10-year ECB bonds and the month ECB bonds, the relative interest rate (EU relative interest rate) which is the deviation of 3-month ECB bonds to 1-year average, and business cycle indicators (EU business cycle) for Europe. The control variables of global factors are monthly percentage changing exchange rate, global risk aversion measured by the VIX index, and global equity returns measured by MSCI World index returns. Standard deviations are reported in parentheses; */**/*** stand for statistical significance at the 10%, 5% and 1% levels.
(1) (2) -0.0075 -0.0065 (-0.78) (-1.54) -0.0131 -0.0030 (-1.03) (-0.64) -0.0117 0.0077* (-1.04) (1.95) 0.0051 0.0026 (0.53) (0.53) -0.0327*** -0.0096* (-2.69) (-1.78) -0.0034 0.0004 (-0.31) (0.10)
Control Var. No Yes
N 156 148
Adj- 0.1601 0.8367
Table 4 indicates the U.S. EPU spillover effect on European stock markets. First, consistent with Table 3, European economic policy uncertainty has a negative effect on domestic equity returns. More importantly, U.S. EPU has a positive spillover effect on European equity market performance. The results are robust even after controlling for equity co-movement and other relevant factors. Specifically, one-point deviation from one-month-lagged U.S. policy uncertainty will increase European stock returns by 77 basis points. The European EPU effect on U.S. equity returns is given by Table 5. Although European EPU also has a positive spillover effect on the U.S. stock market, the impact is lower in comparison with Table 4. Specifically, increasing the standard deviation of European EPU by one-point is associated with U.S. stock returns up 26
15
basis points in the current month. Overall, the evidence in this section seems to support that economic policy uncertainty has a positive international spillover effect on equity market returns. Furthermore, the U.S. EPU spillover effect on European stock markets is more significant than the European EPU spillover effect on the U.S. equity market. That is, European stock markets have a higher level of exposure to U.S economic uncertainty than vice versa. This evidence is consistent with findings from the IMF (2013) with respect to uncertainty spillover effects on economic growth.
Table 5
European Policy Uncertainty Spillover Effect on U.S. Stock Market
The table reports estimates of the following regression:
( ) ( ) ( )
( )
The dependent variable is the monthly returns of the MSCI U.S. index. The sd_USEPU and sd_EUEPU are standardized economic policy uncertainty indexes of the U.S. and Europe (Baker, Bloom and Davis 2013). The domestic control variables are term spread (U.S. term spread) measured as yield difference between 10-year Treasury notes and 3-month Treasury bills, the relative interest rate (U.S. relative interest rate) which is the deviation of 3-month Treasury bills to 1-year average, and the NBER business cycle dummy of U.S. (U.S business cycle). The control variables of global factors are monthly percentage changing exchange rate, global risk aversion measured by the VIX index, and global equity returns measured by the MSCI World index returns. Standard deviations are reported in parentheses; */**/*** stand for statistical significance at the 10%, 5% and 1% levels.
(1) (2) -0.0041 0.0046 (-0.56) (0.97) -0.0142 -0.0033 (-1.56) (-0.80) -0.0183** -0.0124*** (-2.06) (-3.51) 0.0052 0.0002 (0.72) (0.05) -0.0209** 0.0026* (-2.26) (1.85) 0.0004 0.0009 (0.04) (0.24)
Control Var. No Yes
N 156 148
16
5. Spillover effect transmission: portfolio shift channel
The previous analysis shows a statistically reliable link between U.S. economic policy uncertainty and European stock returns, and vice versa. In this section, we investigate the potential transmission channel of domestic policy uncertainty to foreign equity returns. As mentioned above, economic policy uncertainty from a large economy can influence stock returns in another economy through several mechanisms. Considering the scope of the present research, we only focus our analysis on the portfolio shift channel. Specifically, we examine the following plausible hypotheses.
U.S. economic policy uncertainty transmission to European equity market:
Hypothesis 1 (equity inflow effect). Higher economic policy uncertainty in the U.S. can lead to lower domestic economic growth expectation; hence, investors would reallocate their investment portfolios from U.S. to European equity markets. As a result, higher capital inflows drive up the equity prices as well as returns.
Hypothesis 2 (bond inflow effect). Increasing economic policy uncertainty in the U.S. can reduce expectations of economic growth; therefore, investors might shift investments from U.S. to European bond markets. A higher demand for European bonds will reduce the cost of debt for European firms and increase their equity values as well as returns.
European economic policy uncertainty transmission to the U.S. equity market:
Hypothesis 3 (equity inflow effect). Increasing European economic policy uncertainty would be associated with lower growth expectations, which would reduce the attractiveness of domestic financial market; hence, investors would make a significant shift in portfolio allocation away from European financial markets to the U.S. equity market. As a result, increasing equity inflow should drive up U.S. equity prices as well as returns.
Hypothesis 4 (bond inflow effect). Higher economic policy uncertainty in the U.S. can reduce expectations of domestic economic growth; hence, investors might shift investments from European financial markets to the U.S. bond market. A higher demand for U.S. bonds will reduce the cost of debt in the U.S. financial market; therefore, increasing equity values as well as returns.
According to hypotheses 1 and 2, if the financial transmission of U.S. economic policy uncertainty to European equity markets works through net portfolio inflow, then this uncertainty effect should be reflected in positive returns in European equity markets. Likewise, according to
17
hypotheses 3 and 4, higher economic policy uncertainty in Europe plus portfolio inflow to the U.S. should also be associated with positive equity returns in the United States.
One important issue is that a high level of economic policy uncertainty in a large economy may impact investors’ willingness to take risks and lead to safe haven flows. That is, investors might shift their investment away from equity markets to safe haven assets (gold, U.S. T-bills, Swiss Francs) that are perceived as performing reasonably well in difficult market situations. This phenomenon will drive up the price of safe haven assets and drive down the price of stocks. In order to separate the portfolio transmission effect and safe haven effect, we also control for safe haven assets performance in the model.
Before testing the portfolio shift transmission channels, we provide a simple correlation analysis between economic policy uncertainty and net portfolio inflows. Table 6 confirms that U.S. economic policy uncertainty is strongly positively correlated with portfolio inflows to Europe. Likewise, European economic policy uncertainty is also positively but weakly correlated with portfolio inflows to the U.S. These patterns are consistent with portfolio shift channels in which higher policy uncertainty in a large economy would make investors reallocate their investment portfolio to other developed countries. Furthermore, on average, bond inflows are more sensitive to policy uncertainty spillovers than equity inflows.
Table 6
Correlation between EPU and net portfolio inflows Net U.S. bond
inflows Net U.S. equity inflows Net EU bond inflows Net EU equity inflows 0.0709 -0.0310 0.1750 0.0813 0.0242 0.0190 -0.0028 0.0156 0.0175 -0.0165 0.0587 0.0513 0.0503 0.0379 0.0323 -0.0958 0.0004 0.0256 -0.0676 -0.0200 -0.0012 0.0201 -0.0251 0.0023
18
5.1. U.S. economic policy uncertainty transmission to European equity markets To examine Hypotheses 1 and 2, we estimate the regression as following:
∑ ∑
∑
( )
where is the one-month holding period return of the MSCI Europe index; is the standardized economic policy uncertainty; and is the difference of and . The safe haven assets are monthly return for gold, U.S. 3 month Treasury bills and Swiss Francs. The control variables cover relevant factors to both domestic and global asset pricing. is the monthly net portfolio inflows into Europe, and interaction terms ( ∑ ) are the net portfolio inflows times EPU index. The magnitude of the effect of the interaction terms can be perceived in the following way. Consider a situation where a U.S. economic policy uncertainty shock causes large amounts of safe haven inflows. Investors shift their investment away from equity markets, driving down stock prices in both U.S. and European equity markets. However, some less-risk-averse investors could shift their investments from the U.S. to Europe and increase net portfolio inflows into Europe. Such portfolio inflows alleviate the downward pressure on European stock prices that are caused by the safe haven effect. These mitigation effects of uncertainty transmission through portfolio inflows are captured by interaction terms.
Table 7 reports the results for two different portfolio specifications. In the first specification (Column 1), we test the U.S. economic policy uncertainty effect on European equity markets through bond inflows. In the second specification (Column 2), we examine U.S. policy uncertainty spillover to European stock markets through equity inflows. The results show that a positive relation exists between U.S. economic policy uncertainty and European equity returns if there are positive bond or equity inflows into Europe. Thus, this is consistent with Hypotheses 1 and 2. The estimated effects are both statistically and economically significant. In the long run, a one-point standard deviation increase in U.S. economic policy uncertainty and a one percent rise in net bond inflows increases European equity market returns by 2.89% in comparison with -0.78% on European stock returns without bond inflows. In contrast, U.S. policy uncertainty through equity inflows has a lower effect on European equity markets.
19 Table 7
U.S. EPU transmission to European stock market
The table reports estimates of the following regression:
∑ ∑
∑
The dependent variables are the monthly returns of the MSCI Europe index. The sd_USEPU and sd_EUEPU are standardized economic policy uncertainty index of U.S and Europe (Baker, Bloom and Davis 2013). is the monthly
net portfolio inflows to Europe, and interaction terms ( ) are the net portfolio inflows times U.S. EPU
index. The safe haven assets are the monthly returns of gold, U.S. 3 month Treasury bills, and Swiss Francs. The control variables include monthly percentage changing exchange rate, global risk aversion measured by the VIX index, global equity returns measured by MSCI World index returns, term spread measured as yield difference between 10-year government bonds and the 3-month government bonds, the relative interest rate which is the deviation of 3-month bond to 1-year average, and business cycle indicators for the U.S. and Europe. Standard deviations are reported in parentheses; */**/*** stand for statistical significance at the 10%, 5% and 1% levels.
U.S. EPU effect on European equity market (1) Bond inflows (2) Equity inflows Portfolio inflows 0.0193** 0.0016 (2.58) (0.37) -0.0078** -0.0069* (-2.04) (-1.75) 0.0016 0.0022 (0.44) (0.64) 0.0084*** 0.0096*** (2.62) (2.89) 0.0174*** 0.0158*** (2.74) (3.77) 0.0107 0.0104 (0.99) (1.39) -0.0082* -0.0079 (-1.71) (-0.78) 0.0050 0.0043 (1.25) (1.17) -0.0100** -0.0110*** (-2.33) (-2.90) -0.0018 -0.0013 (-0.52) (-0.39) Gold Return 0.0150* 0.0082* (1.65) (1.74)
U.S. T-bill Return 0.0107* 0.0140*
(1.72) (1.83)
CHF Returns 0.0091 -0.0496
(0.12) (-0.60)
Control Var. Yes Yes
N 148 148
20
A one-point standard deviation increase in U.S. EPU with a one percent increase in net equity inflows is associated with 1.05% stock returns in the long term. This might be partially because of the lower correlation between equity inflows and economic policy uncertainty.
Furthermore, in the short run, U.S. policy uncertainty with portfolio inflows on average is weakly positively correlated with European equity returns. The estimated effects are not economically significant. This could be because some investors are extremely risk-averse, so they just stop investing and leave the financial market during the higher uncertainty period. However, further research is required to establish the strength of this theory.
5.2. European economic policy uncertainty transmission to U.S. equity market To examine Hypotheses 3 and 4, we use a similar model to section 5.1:
∑ ∑
∑
( )
where is the one-month holding period return of the MSCI U.S. index; is the standardized economic policy uncertainty; and is the difference of and . The safe haven assets are monthly return for gold, U.S. 3 month Treasury bills and Swiss Francs. The control variables cover relevant factors for both domestic and global asset pricing. is monthly net portfolio inflows into the U.S., and interaction terms ( ∑ ) are the net portfolio inflows times EPU index.
Table 8 shows the results for two different portfolio specifications. In the first specification (Column 1), we test the European economic policy uncertainty effect on the U.S. equity market through bond inflows. In the second specification (Column 2), we examine European policy uncertainty spillover effect on the U.S. stock market through equity inflows. In general, there is a significant positive relation between European economic policy uncertainty and U.S. equity returns if there are positive equity inflows into the United States in the long term, consistent with Hypothesis 3. However, this relation is negative and insignificant if we consider bond inflows instead of equity inflows. This analysis suggests that European economic policy uncertainty transfer through equity inflows into the U.S. is associated with positive stock returns in the United States.
21 Table 8
European EPU transmission to the U.S. stock market
The table reports estimates of the following regression:
∑ ∑
∑
The dependent variables are the monthly returns of the MSCI U.S. index. The sd_USEPU and sd_EUEPU are standardized economic policy uncertainty index of U.S. and Europe (Baker, Bloom and Davis 2013). is the monthly
net portfolio inflows into U.S. and interaction terms ( ) are the net portfolio inflows times European
EPU index. The safe haven assets are monthly returns of gold, U.S. 3 month Treasury bills, and Swiss Francs. The control variables are monthly percentage changing exchange rate, global risk aversion measured by the VIX index, global equity returns measured by MSCI World index returns, term spread measured as yield difference between 10-year government bonds and the 3-month government bonds, the relative interest rate is the deviation of 3-month bond to 1-year average, and business cycle indicators for U.S and Europe. Standard deviations are reported in parentheses; */**/*** stand for statistical significance at the 10%, 5% and 1% levels.
European EPU effect on U.S. equity market (1) Bond inflows (2) Equity inflows Portfolio inflows 0.0156** 0.0005 (2.52) (0.11) 0.0073** 0.0072* (1.93) (1.76) -0.0044 -0.0053 (-1.35) (-1.54) -0.0111*** -0.0121*** (-3.68) (-3.66) -0.0023 -0.0031 (-0.66) (-0.77) 0.0054 0.0050 (1.06) (1.19) 0.0022 0.0013 (0.55) (0.32) -0.0103 0.0096** (1.62) (2.38) -0.0024 -0.0017 (-0.22) (-0.22) -0.0059 0.0031 (-0.62) (0.27) Gold Return -0.0565* -0.0489 (-1.70) (-1.48)
U.S. T-bill Return -0.0095* -0.0116*
(-1.68) (-1.84)
CHF Return -0.0310 0.0328
(-0.47) (0.45)
Control Var. Yes Yes
N 148 148
22
However, the evidence for European policy uncertainty transmission through bond inflows is rather weak. We do not find strong evidence to support European policy uncertainty with bond inflows into the U.S. would have a positive effect on the U.S. equity market.
5.3. Discussion of the empirical findings
The empirical evidence on the policy uncertainty spillover effect on equity returns in Sections 5.1 and 5.2 are overall more supportive of Hypotheses 1 and 2 than Hypotheses 3 & 4. On average, economic policy uncertainty transmission through portfolio inflows are associated with higher equity returns. This is consistent with evidence in Section 4, which shows that under portfolio shift channels, the U.S. has a larger economic policy uncertainty spillover effect on European equity markets than the European EPU effect on the U.S. stock market. This evidence then brings up the question of why financial markets are more sensitive to U.S. policy uncertainty than European EPU. Although we cannot provide a conclusive answer, we can conjecture several plausible explanations. First of all, as Rapach et al. (2013) identify, the U.S. stock market significantly predicts equity returns for other non-U.S. advanced economies. Despite the increasing economic status of the European Union, there is no substitute for the U.S. on the world economic stage. The U.S. still holds the leading position on the world financial market. In addition, both liquidity and market size of European financial markets are smaller than the United States, which means comparatively lower financial transactions in European markets, hence reducing the influence on the international financial market.
The alternative interpretation for these empirical findings is that the results are just the outcome of data mining. To address this concern, we also use the U.S. presidential election period as an instrumental variable to study the U.S. policy uncertainty spillover effect on European equity markets. In the United States, election timing is always fixed and cannot be influenced by individuals or firms. Meanwhile, the election generates policy uncertainty. Thus, it can be seen as a natural experiment to test the relationship between policy uncertainty spillover effects on European equity returns. In results not reported here, we find similar evidence that U.S. economic policy uncertainty with portfolio inflows is associated with positive equity returns in European financial markets in the long run.
23
6. Conclusion
In this paper, we study the economic policy uncertainty spillover effect on equity returns via portfolio shift channels between the U.S. and Europe. Our main empirical findings can be summarized as follows.
First, economic policy uncertainty in a home economy has a negative impact on domestic equity returns in both U.S. and European financial markets. Economic policy uncertainty has a larger impact on domestic equity returns in Europe than in the United States.
Second, without considering any transmission channels, U.S. economic policy uncertainty has a positive effect on European equity returns. A one-point deviation from one-month-lagged U.S. policy uncertainty will increase European stock returns by 77 basis points. A similar pattern also holds for European economic policy uncertainty effects on U.S. equity returns.
Third, the policy uncertainty spillover effect has a large positive long run effect on equity returns via bond and equity flows between the U.S. and Europe. A one-point standard deviation increase in U.S. economic policy uncertainty and one percent rise in net bond inflows increase European equity market returns by 2.89%.
Forth, the U.S. EPU spillover effect on European financial markets is larger than the European EPU effect on the U.S. market. These findings suggest that financial markets are more sensitive to U.S. policy uncertainty among advanced economies.
A fundamental question that we leave open is: What are the impacts of economic policy uncertainty spillovers on stock markets via other transmission channels? In this paper, we only test the portfolio shift channel. Other transmission channels might have different effects and would certainly improve understanding of policy uncertainty spillover effect on asset pricing. Thus, such research would help policymakers to improve their communication decisions and mitigate negative spillover effect to the rest of the world.
24
Reference
Baker, S.R., Boom, N., Davis S.J., 2013, Measuring economic policy uncertainty, Working paper, University of Chicago, Booth School of Business.
Banz, R.W., 1981, The relationship between return and market value of common stocks, Journal of Financial Economics 9, 3-18.
Beber, A., Brandt, M.W., Kavajecz, K.A., 2009, Flight-to-quality or flight-to-liquidity? Evidence from the Euro-Area bond market, Review of Financial Studies 22, 925-957.
Belo, F., Gala, V.D., Li, J., 2013, Governments pending, political cycles, and the cross-section of stock returns, Journal of Financial Economics 107, 305-324.
Berk, J.B., 1997, Does size really matter?, Financial Analysts Journal, Vol. 53, No. 5, 12-18.
Brogaard, J., Detzel, A., 2013, The asset pricing implications of government economic policy uncertainty, Working paper, University of Washington.
Carriere-Swallow, Y., Cespedes, LF., 2013, The impact of uncertainty shocks in emerging economies, Journal of International Economics 90, 316-325.
Chan, K.C., Chen, N,F., 1991, Structural and return characteristics of small and large firms, Journal of Finance, Vol. 46, No. 4, 1467-1484.
Forbes, K., 2012, The ‘Big C’: Identifying Contagion, NBER Working paper 18465, National Bureau of Economic Research, Inc.
Forbes, K., Fratzscher, M., Kostka, T., Straub, R., 2012, Bubble thy neighbor portfolio effects and externalities from capital controls, Working paper series 1456, European Central Bank.
Fratzscher, M., Lo Duca, M., Straub, R., 2012, A global monetary tsunami? On the spillovers of US Quantitative Easing, CEPR Discussion Papers 9195.
Gao, P., Qi, Y., 2013, Political uncertainty and public financing cost: Evidence from U.S municipal bond markets, Working paper, City University of Hong Kong.
Gauvin, L., McLoughlin, C., Reinhardt, D., 2013, Policy uncertainty spillovers to emerging markets – evidence from capital flows, Working paper, Banque de France.
25
Gonzalez, A., Terasvirta, T., van Dijk, D., 2005, Panel Smooth Transition Regression Models, Working paper series in Econometrics and Finance 604, Stockholm School of Economics.
IMF, 2013, Hopes, realities, and risks, World Economic Outlook, IMF.
Julio, B., Yook, Y., 2013, Policy uncertainty, irreversibility, and cross-border flows of capital, Working paper, Finance and Economics Discussion Series, Divisions of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, D.C.
Keim, D.B., 1983, Size-related anomalies and stock return seasonality: further empirical evidence, Journal of Financial Economics 12, 13-32.
Kelly, B., Pástor, L., Veronesi, P., 2014, The price of political uncertainty: theory and evidence from the option market, Working paper, University of Chicago.
Lo Duca, M., 2012, Modelling the time varying determinants of portfolio flows to emerging market, Working paper series 1468, European Central Bank.
Pástor, L., Veronesi, P., 2012, Uncertainty about government policy and stock prices, Journal of Finance 67, 1219-1264
Pástor, L., Veronesi, P., 2013, Political uncertainty and risk premia, Journal of Financial Economics 110, 520-545
Rapach, D.E., Strauss, J.K., Zhou, G., 2013, International stock return predictability: what is the role of the United States?, Journal of Finance, Vol. 8, No.4.
Roe, M, J., Siegel, J, I., 2011, Political instability: Effects on financial development, roots in the severity of economic inequality, Journal of Comparative Economics 39, 279-309.
Wright, J.H., 2011, Term premia and inflation uncertainty: Empirical evidence from an international panel dataset, American Economic Review, 101(4), 1514-34.
