Thesis
The effect of the latest terror attacks on the consumer
confidence of European countries.
Name: Mélanie Bosboom Student number: 10797661
Specialization: Economics and Finance Field: Economics
Number of credits thesis: 12 Name of supervisor: Péter Foldvari
Statement of Originality
This document is written by Melanie Bosboom who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Content
1. Introduction 4
2. Literature review
2.1 Terrorist attacks 6
2.2 Consumer confidence index 8
2.3 Conclusion and hypothesis 9
3. Data collection 10
4. Method and research design
4.1 Method 13
4.2 Fixed effects model 13
4.3 Random effects model 14
4.4 Hausman test 14
4.5 The model 14
4.6 Long term effects 15
5. Results
5.1 Random effects model with lagged dummy variables 16 5.2 Random effects model without lagged dummy variables 16
5.3 Long term effects results 17
6. Conclusion 18
7. Discussion 18
8. References 20
1. Introduction
On the 7th of January 2015 in Paris, multiple gunmen started shooting at journalists
from the magazine Charlie Hebdo. In total, 12 people were killed and 12 people were injured. This left France as well as all the Western European countries speechless. Everyone felt that this was a direct attack on the Western world. Unfortunately, the terrorist attacks did not stop after that. Several attacks have been executed since then and they only became deadlier. The most lethal attack occurred on the 13th of November 2015, where more than 135 people were
killed and more than 320 people were injured. Since that day, France declared a state of emergency, which has remained for over two years now. However, the attacks did not only occur in France. Belgium, Germany, United Kingdom and Spain were also victims of terrorist attacks. In the past two years, these countries together have lost more than 300 people and more than 1585 were injured. All these attacks create an atmosphere of anxiety and a fear of terrorism. Due to this, civilians and tourists decide to stay at home which results in less spending.
However, these attacks are not the first attacks to happen. Over the years, many research has been done to investigate the economic effects of terror attacks. Although all the research is very different, they all come to the conclusion that terroristic attacks negatively influence the economy of a country. Besides these direct economic effects, one of the indirect effects is that the terrorist attacks influence individuals’ fear and insecurity. These feelings affect the household’s trust in the future economy which affects their economic decisions (Christelis and Georgarakos, 2009). The consumer confidence is an indicator that reflects these feelings. To be precise, it reflects the consumers’ feelings about current and future economic conditions, which is used as an indicator of the overall state of the economy (Oxford dictionary). Many research finds that indeed, the consumer confidence index is a key determinant of economic growth. Leeper (1992) was one of the first researchers that looked at the role that consumer confidence plays in forecasting economic activity. He tested if the data from the consumer confidence index is informative and if the data is independently of other information about future economic conditions. Ludvigson (2004) and Acemoglu and Scott (1994) both measured if the consumer confidence index predicts consumer consumption, and if the index correlates with the state of the economy. All these researchers came to the conclusion that consumer confidence is a valuable indicator for economic future. Even though the consumer confidence index has proven to be an indicator of the current and future state of the economy, only one researcher has observed the effect of terrorist attacks on the consumer confidence of a country, and so on observed the effect on the economy of a country. Unfortunately, the effects of terrorist attacks on the consumer confidence index are relatively unexplored. This is why it is interesting to observe if terrorism has a significant effect on it.
This research is based on the investigation of Garner (2002). His research examined the effect of the 9/11 attacks on the consumer confidence of the United States. He used a regression equation and included lagged values of the consumer price, inflation, civilian
unemployment rate, change in stock prices and the confidence index itself. Dummy variables were included for unique events also. This study goes beyond Garner’s (2002) research through three big changes. Firstly, by investigating more countries to be able to give a more generalized answer to the question if terrorist attacks influence the consumer confidence index. Secondly, by using more explanatory variables to get a more sophisticated model and in effect, improve the predictive power of the regression equation. Thirdly, by also observing the effect of a terrorist attack in one country, on the consumer confidence of another country. For this, an extra dummy variable will be added.
To examine this, the consumer confidence index of five European countries that suffered the last couple of years from large terror attacks will be analyzed. These countries will be Belgium, France, Germany, United Kingdom and Spain. Using a panel data regression, we can explore the relationship between the consumer confidence, the terror attacks and other explanatory variables. All the data will be quarterly calculated in a time frame from January 2000 until September 2017.
The outline of this study will be as follows. Firstly, definitions and relevant literature about this topic will be explained and evaluated through a theoretical frame work. Secondly, the data used for this research will be explained and analyzed. Thirdly, the research design and chosen method will be explained. Fourthly, answers to the research question will be given, based on the results of the research.
2. Literature review
In this section multiple relevant literature and definitions will be discussed and evaluated. First of all, the definition of terrorism will be explained, as well as its effect on the economy. Moreover, research on the consumer confidence index will be discussed. The section closes with a conclusion and the research hypothesis.
2.1 Terroristic attacks
Many researchers give different definitions about terrorist attacks. In this thesis we will focus on the definition of the Global Terrorism Database. According to the Global Terrorism Database, each incident had to be an intentional act of violence or threat of violence by a non-state actor to be considered as a terrorist act. In addition, two of the following three criteria also have to be met: the violent act was aimed at attaining a political, economic, religious, or social goal; the violent act included evidence of an intention to coerce, intimidate, or convey some other message to a larger audience other than the immediate victims; and the violent act was outside the precepts of International Humanitarian Law. The Global Terrorism Database categorize each terrorist attack in the following eight categories. First of all, they categorize the attacks by dates of when the attack happened. Secondly, they categorize the attacks by location in terms of the attacked countries. Thirdly, they categorized the attacks by city where the attack occurred. Fourthly, they categorize the attacks by which perpetrator or group claimed the attack. Fifthly, they categorize the attacks by the number of fatalities. They define fatalities by the number of total confirmed deaths for the incident, this includes all victims and attackers. Sixthly, they categorize the attacks by how many people were injured. They define those injured by the number of confirmed non-fatal injuries to both perpetrators and victims. Seventhly, they categorize the attacks by what kind of attack it was. Lastly, they categorize them by which target the attack focused on.
Multiple studies have been done to investigate what the economic consequences are of terrorist attacks on a country. This literature review will discuss the two most relevant research studies on the economic effects. Firstly, Garner (2002) observed the effect of the terrorist attack of 9/11 on the consumer confidence of the United States. He used two measures on consumer confidence indexes, the Conference Board’s index and the University of Michigan’s index. His study shows that the consumer confidence indexes fell below their historical average in the fall of 2001, but recovered very fast after. To examine the effect of economic indicators and unique events on the consumer confidence, regression equations were estimated. He used a simple empirical model including four lagged values of the unemployment rate, the CPI inflation, the percentage change in the S&P500 and the confidence measure itself. A dummy variable was included to represent the 9/11 attack. His results showed a surprisingly resilient consumer confidence index after the attack. For both indexes, the dummy variable in the fourth quarter of 2001 was not statistically significant. All the other dependent variables were statistically significant which shows us that there is a direct correlation between consumer confidence and other economic indicators. The decline
in the fourth quarter could be explained by the worsening economic conditions in the third quarter earlier. Another explanation for this insignificance could be that the sample of Garner (2002) was too small. The attack ensued in November 2001 and he analyzed the effect in 2002. There could be a possibility that the effect of the attack had not hit the consumer confidence yet.
Arin, Ciferri and Spagnolo (2008) investigated the effects of terrorism on the financial markets. They tested whether terrorist attacks have a statistically significant causality effect on the stock market return and the stock market volatility. They tested this on six different countries (Indonesia, Israel, Spain, Thailand, Turkey and UK) by using the bivariate VAR-GARCH(1,1)-in-mean model. To control exogenous shocks, they included in the mean equation domestic interest rates (90-days treasury bill rate) and a proxy for the global stock market index (the US stock market returns). They procured data over the period of 1/1/2002 until 31/12/2006, with a total of 1368 observations. Their results suggested that there is evidence of statistically significant causality effects, both in mean and in variance and in all six countries. However, the responses to terror shocks vary across countries. The stock markets of Spain and UK are generally less effected by terror shocks. This shows that the financial market in these European countries are more resilient. An explanation for this difference could be that investors in Spain and UK are more aware that their authorities might be able to restore these terrorist shocks.
2.2 Consumer confidence
In the second part of this literature review, the consumer confidence index and related researchers will be discussed.
According to the OECD, the consumer confidence index is based on households’ plans for major purchases and their economic situation, both current and estimated for the immediate future. Opinions compared to a normal state are collected and the difference between positive and negative answers provides a qualitative index on economic conditions.
A couple of studies have been done to investigate if consumer confidence is correlated with macro-economic indicators and if the consumer confidence is valuable to predict economic future. Acemoglu and Scott (1994) tested two things. First they tested to see if consumer confidence is an indicator of consumption growth by using a regression analysis. Besides statistical evidence of a simple regression between consumer confidence and consumption growth, they argued that confidence indicators may be a useful sign of consumption if they predict variables relevant to the consumers’ spending. So they regressed the current change in the logarithm of labor income on lagged changes of consumer confidence, unemployment, inflation, real interest rates and changes in financial and housing wealth. Their evidence showed that consumer confidence is the only indicator that predicts future consumption growth. Secondly, they wanted to test if the consumer confidence indicator predicts unemployment, inflation, real interest rates, financial wealth and housing wealth by using a regression analysis. They constructed data from the UK in a period from 1955-1991 and with data adjusted quarterly. They found statistical evidence for all the
explanatory variables except financial wealth. They concluded that the confidence indicator is highly correlated with the current state of the economy and, is a predictor of future economic strength.
Ludvigson (2004) measured, just like Acemoglu and Scott (1994), if consumer confidence surveys predict consumer consumption. He used a regression including consumer confidence and five categories of household expenditures - total expenditure, motor vehicle expenditure, on all goods (except motor vehicles), services expenditure, and durable goods (except motor vehicles). He also included a dummy variable that will be set equal to one in the quarters corresponding to the 1990-1991 recession. He retrieved the data in a period from 1968 until 2002. The evidence from the regression suggested that measures of consumer confidence has important predictive power over consumer expenditure growth. His next question is whether the confidence measured contains predictive information that is not already contained in standard economic indicators. For standard economic indicators he used labor income growth, the change in the real stock price (S&P500) and the change in the three-month Treasury bill rate. He regressed the five categories of expenditures on consumption growth itself and these economic indicators. The evidence showed that including the interest rate variables reduces the statistical significance.
An older study came to a slightly different conclusion. Leeper’s (1992) article looked at the role that consumer confidence plays in foreshadowing economic activity. He tested if the data from the consumer confidence index is informative, and if the data is independent of other information about future economic conditions. He first explained the different approaches researchers have had against consumer confidence. The first approach is that consumer confidence is a causal variable. Pigou and Keynes explained that on a constant basis of facts, men do not form a constant judgement. The second approach is that consumer confidence is a catalyst. Von Haberlor agrees with Pigou and Keynes and states that optimism and pessimism are regarded as causal factors which tend to induce or intensify the rise and fall of investment. The last approach is that surveys of consumer confidence predict economic conditions. Whatever one’s view is, the survey data are taken seriously by a wide range of economic observers. Leeper explained that some changes in the consumer confidence index are predictable but some changes are unexpected; these unexpected changes cannot be predicted using historical values of economic variables. He observed whether these unexpected changes are related to industrial output and unemployment. He concluded that there is indeed a correlation between these three variables. After this, he included stock prices and interest rates. The evidence showed that as soon as he included the financial variables, the correlation diminishes. Real stock prices and interest rates appear to absorb the predictive power of confidence. He overall found that confidence adds very little predictive power to economic forecasts, as it only slightly predicts the forecasts of output and unemployment.
2.3 Conclusion and hypothesis
Prior studies have shown that terrorism affects the economy of a country in multiple ways. Arin et al. showed that terrorism has a significant effect on the financial market of the six
countries studied. Garner showed the little effect 9/11 had on consumer confidence. He concluded that the dummy variable that represents the terrorist attack of 9/11 was not statistically significant and that the consumer confidence was already declining because of worsening economic conditions in the third quarter. Another explanation for this insignificance could be that the sample of Garner (2002) is too small since his study was not even a year later.
Other researchers studied the consumer confidence in more detail; their conclusions had some similarities and some differences. Ludvigson (2002) and Leeper (1992) both concluded that interest rates reduce the significance of consumer confidence as an economic indicator. Interestingly, Garner left the interest rates out of his research. In contrast to this, Acemoglu and Scott (1994) did found the interest rates a significant variable. The more recent studies (Acemoglu and Scott (1994) and Ludvigson (2002)) also concluded that consumer confidence is the only indicator that predicts future consumption growth and that confidence is a predictor of future economic growth. Leeper (1992) founded less convincing results. His results showed that consumer confidence is a weak indicator of the economy and only slightly forecasts industrial output and unemployment. Besides that, Leeper (1992) founded that real stock prices as well as real interest rates reduce the statistical significance. However, all studies experience a variation in the consumer confidence index that remains unexplained.
Besides the studies that measure the economic effects of terrorist attacks, only one study has been done to observe the effect of terrorist attacks on the consumer confidence of a country. This study will contribute to the debate and will analyze if there is a significant effect of terrorist attacks on consumer confidence. This research will give a more generalized answer to this question by observing 5 European countries in an almost 17-year period. As terrorist attacks do not only effect the home country but also other countries, this study will also analyze the effect of a terrorist attack that occurred in one country on one another. Therefore, we will have two hypotheses.
Garner’s (2004) study shows that the consumer confidence of the US after 9/11 was quite resilient. Furthermore, the economy of European countries shows to be durable after big shocks. This is why I expect that the terrorist attacks will only have a slight effect on the consumer confidence. Therefore, the first hypothesis will be that terrorist attacks that happened in the home country, do have a significant statistical effect on the consumer confidence of the home country. The second hypothesis will be that terrorist attacks in a foreign country, do have a statistical significance effect on the consumer confidence of the home country. To test these hypotheses, the P-value of the coefficients will be used as a decision criterion.
H0: P-value of dummy terrorist attack in home country = 0.05
H1: P-value of dummy terrorist attack in home country < 0.05
H0: P-value of dummy terrorist attack in foreign country = 0.05
3. Data collection
In this section the data used for this research will be explained and analyzed.
In this research we have two types of variables. Firstly, we have the dependent variable, and secondly we have the explanatory variables. All the data will be quarterly calculated in a time frame from January 2000 till October 2017 for all the five European countries. The five countries will be Belgium, France, Germany, Spain and the United Kingdom. All the countries are numbered from 1 till 5. The model is unbalanced and will give us 339 observations. A table for all the data explained, is added in the appendix.
Starting with the data for the dependent variable, the consumer confidence index. The consumer confidence indexes are retrieved from Datastream. Datastream obtained the consumer confidence indexes from each country from their National Banks. These national banks release new indexes each month. Like already explained, the consumer confidence index is based on households’ plans for major purchases and their economic situation, both current and estimated for the immediate future. Opinions compared to a normal state are collected and the difference between positive and negative answers provides a qualitative index on economic conditions. The data is quarterly adjusted.
Besides this, we have the explanatory variables. Previous research shows that simple statistic models explain a large part of the variation in the consumer confidence. However, many researches use different explanatory variables and have different conclusions which variables are significant. This research will add more explanatory variables than other researchers to be able to improve the predictive power of the regression equation. All the explanatory variables are quarterly retrieved from the Organization for Economic Co-operation and Development (OECD). The OECD is an organization that publishes annual overviews and comparative statistics. The explanatory variables are as follows.
Firstly, we start with the consumer price index (CPI). The CPI measures the inflation of a country and is defined as the change in the prices of a basket of goods and services that are typically purchased by specific groups of households. The CPI is measured in terms of the annual growth rate and in index, with 2010 as base year. This variable is balanced for all countries. When there is a large increase in the inflation, the monetary unit will deflate. This will have bad consequences for the economy and so on will let the consumer confidence index decrease. This why a negative relationship is expected.
Secondly, we have the unemployment rate. The OECD defines unemployment rate as the number of unemployed people as percentage of the labor force, where the latter consists of the unemployed plus those in paid or self-employment. Unemployed people are those who report that they are available for work and that they have taken active steps to find work in the last four weeks. The unemployment rate is a well know indicator of the economy. If the unemployment rate increases, the state of the economy decreases and so on the consumer confidence index will decrease as well. This is why a negative relationship is expected.
Thirdly, we use price index. These indices are calculated from the prices of common shares of companies traded on national or foreign stock exchanges. Normally, they are determined by the stock exchange, using the closing daily values. The indices are quarterly adjusted and measured with 2010 as a base year. Higher share prices reflect a growing economy. A higher state of the economy will let the consumer confidence index increase as well. This is why a positive relationship is expected.
Fourthly, we have the real house price. The real house price shows the indices of residential property prices over time. It gives the ratio of the nominal price to the consumers’ expenditure deflator in each country and is measured in terms of an index with 2010 as base year. A higher real house price, reflects a growing economy. A higher state of the economy will let the consumer confidence index increase. This is why a positive relationship is expected. Even though Garner (2002) doesn’t use housing prices, Acemoglu and Scott’s (1994) results showed that housing wealth does has a significant effect on the consumer confidence index.
Fifthly, we include the short term real interest rates. These are rates at which short-term borrowings are effected between financial institutions or the rate at which short-short-term government paper is issued or traded in the market. These rates reflect the average of daily rates and is measured as a percentage. The short-term interest rates are also known as Treasury Bill Rate. When the confidence falls, the reaction of households is to increase savings and reduce borrowing. The reaction of the government will be to lower the interest rates to make spending more attractive. Sixthly, we have the long-term interest rates. These rates reflect the prices at which the government bonds are traded on the financial market. They only refer to bonds whose capital repayment is guaranteed by governments, because of this the long term rates are an important indicator in the political stability of a country. These rates are measured as a percentage. The long-term interest rates reflect the safest investment. Also for long-term interest rates counts that when consumer confidence and the economy is low, the interest rate will be low as well to encourage investment. This is why a positive relation is expected. Although Ludvigson (2002) and Leeper (1992) concluded that interest rates reduce the significant of the consumer confidence as an economic indicator, Acemoglu and Scott (1994) did find statistical significance. Also the study of Arin et al. (2008) shows that there is a significant causality effect of terror attacks on stock market return and volatility. To improve the predictive power, these interest rates will be added in this analysis.
Seventhly, we take in the Gross Domestic Product (GDP). This is a standard measure of the economy of a country. It is calculated by the total value of final goods and services produced by a country during a period minus the value of imports. The difference with NGP is that GDP does not make a deduction for the depreciation of machinery, buildings and other capital products used in production. The GDP is measured as a percentage change from the previous quarter. The GDP is also a well-known economic indicator. An increase in the GDP reflects a growing economy. A higher state of the economy will let the consumer confidence index increase as well. This is why a positive relationship is expected. In the research of Leeper (1992), industrial output was highly significant. Because industrial output can only be calculated by month, GDP is chosen.
Additionally, dummy variables will be added that will represent possible temporary influences from the attacks. To observe the effect of a terrorist attack that happened in the home country on the consumer confidence of the home country, a dummy variable will be added. This dummy variable will take the value of 1 in the quarter were the attack on the home country happened. To observe the effect of a terrorist attack in one country, on the consumer confidence of another country, an extra dummy variable will be added. This dummy variable will take the value of 1 in the quarter were an attack happened in one of the other five countries analyzed. The data from the terrorist attacks are retrieved from the Global Terrorism Database. This research will only focus on large terrorist attacks that made it to the headlines and who could possibly influence the economy. Specifically, we focus on terrorist attacks that caused more than 5 fatalities and more than 10 injured. The table underneath shows the terrorist attacks that will be used in this research.
Date Country City
Perpetrator
group Fatalities Injured Target
11-03-04 Spain Madrid Abu Hafs al-Masri Brigades 62 450 Transportation 07-07-05
United
Kingdom London al-Qaida 56 784 Transportation
07-01-15 France Paris al-Qaida 12 12 Journalists
13-11-15 France Paris Islamic State of Iraq 135 321 Business, Private citizens & Property 22-03-16 Belgium
Brussels and
Zaventem Islamic State of Iraq 18 135 Private citizens & Property, Transportation, Airports 19-12-16 Germany Berlin Jihadi-inspired extremists 12 48 Private citizens & Property 22-07-16 Germany Munich Right-wing extremists 10 27 Private citizens & Property 14-07-16 France Nice Jihadi-inspired extremists 87 433 Private citizens & Property 22-03-17 United Kingdom London Islamic State of Iraq 5 49 Private citizens & Property 22-05-17 United Kingdom Manchester Islamic State of Iraq 22 512 Private citizens & Property 03-06-17 United Kingdom London Islamic State of Iraq 8 48 Private citizens & Property
4. Method and research design
In this section the research design and chosen method will be described and explained.
4.1. Method
To investigate the consumer confidence index of five countries with multiple explanatory variables including two dummy variables, panel data will be used. Panel data means we will be pooling observations on a cross-section of countries over several periods. This method is the best fit for this research because of the following reasons. Firstly, because panel data gives us an extra opportunity, a statistical treatment of individual heterogeneity. Studies using time-series or cross-sectional data who do not control for this heterogeneity, run the risk of obtaining biased results. Secondly, because panel data gives more informative data, more variability, less collinearity among the variables, more degrees of freedom and more efficiency is expected. Other time-series studies experience a lot of multicollinearity in their research. Panel data adds a lot of variability and adds more information, because of this multicollinearity is less likely. Also, the variation in the data can be decomposed into variation between the countries. Thirdly, panel data studies are better able to study the dynamics of adjustment. Cross-section studies do not take these dynamics in their research, even though their results look very stable, they hide a multitude of changes.
In this research we have a lot of economic indicators as explanatory variables. Especially these indicators are better studied with panel data. Including dynamic effects in this research allows us to identify and measure effects that are not possible in other cross-section or time-series studies. To get dynamic effects of the dummy variables, lagged variables from the last 5 quarters will be used. The model with and without dynamic effects will be compared to see what the effect of dynamic variables is.
For both models an important assumption has to hold. The assumption that the error term is independent from all the other explanatory variables for all countries and for each quarter, E(
it,Xit) = 0. If this assumption does not hold, the estimators become biased andinconsistent.
4.2. Fixed effects model
Panel data has two different models, the fixed effects model and the random effects model. When the fixed model is used, the equation will look as follows.
• CCIit =
i + X+
itA fixed effects model tries to describe the consumer confidence index with all the explanatory variables available. The
i catches the average deviation from the different countries. This isa fixed effects model the assumption
it IID(0,σ2) has to hold. This means that the error termis independently and identically distributed.
4.3 Random effects model
When the random effects model is used, the equation changes to the following equation:
• CCIit =
+ X+
it •
it =
it +
iThis model includes two changes. Firstly, the alpha term changed to
.
Secondly, the error term changed to
it =
it +
i. Where
it reflects the random error term who isindependently and identically distributed
it IID(0,σ2) and where
i reflects the individualspecific components. This means that the effect size varies across all different countries.
4.4 Hausman test
To determine which model we can use, a Hausman test should be performed. A Hausman test compares the estimators of the fixed and random effects models and evaluates whether the results are the same. If the null hypothesis holds, the estimators of both models are consistent but the random effects model is a better fit due to higher efficiency. Under the alternative hypothesis, only the estimator of the fixed model is consistent. So the fixed effects model is preferred due to more consistency.
H0 H1
Random effects estimator Consistent, Efficiënt Inconsistent Fixed effects estimator Consistent, Inefficiënt Consistent
The results of the Hausman test proved that the null hypothesis holds, so we will continue with the random effects model. This means that there are individual country effects that are part of the error term.
4.5 The model
The random effects model with lagged dummy variables and the random effects model without lagged dummy variables will be compared. The equations will look as follows.
The random effects model with 5 lagged dummy variables
• CCIit =
+ 1*CPIit + 2*UMPRTit + 3*SPIit + 4*RHPit + 5*STIRit + 6*LTIRit +7*GDPit + DTHCit + DTHCit-1 +DTHCit-2 +DTHCit-3 +DTHCit-4 +DTHCit-5 + DTFCit + DTFC it-1 +DTFCit-2 + DTFCit-3 +DTFCit-4 + DTFCit-5 +
itThe random effects model without lagged dummy variables
• CCIit =
+ 1*CPIit + 2*UMPRTit + 3*SPIit + 4*RHPit + 5*STIRit + 6*LTIRit +7*GDPit + DTHCit + DTFCit +
it •
it =
it +
iThe regressions include shorten variable names and symbols. An explanation for these variable names and symbols:
➢ The i = 1, 2...,5 reflects the number of the countries. The countries are categorized as Belgium = 1, France = 2, Germany = 3, Spain = 4 and United Kingdom = 5.
➢ The t = 1, 2…, 214 reflects the quarters in the time frame from January 2000 till October 2017.
➢ CCI stands for the consumer confidence index ➢ CPI stands for the consumer price index ➢ UMPRT stands for the unemployment rate ➢ SPI stands for the share price index
➢ RHP stands for the real house price ➢ STIR stands for the short-term index rate ➢ LTIR stands for the long-term index rate ➢ GDP stands for the gross domestic product
➢ DTHC stands for the dummy variable when the terrorist attack happened in the home country
➢ DTFC stands for the dummy variable when the terrorist attack happened in a foreign country. This will be one of the other five European countries.
➢
it stands for the error term4.6 Long term effects
To gain a further understanding of the effect of the terrorist attacks in home and foreign country, we will also have a look at the long term effect of the attacks on the consumer confidence index. The Wald-test of linear restrictions will provide us with this information. This test will combine the coefficients and will investigate if the sum is statistical significant. The sum of the coefficients reflects the long term effect of the terrorist attacks on the consumer confidence index. If the sum is statistical significant, there will be a long term effect. The coefficients are the lagged variables of the last 5 quarters of the dummy variable of the terrorist attacks in home country and the dummy variable of the terrorist attacks in foreign country.
5. Results
5.1 Random effects model with lagged dummy variables
First of all, we will look at the model in general. The F-test of the model shows that all the coefficients in the model are different from zero. Also, the correlation between the error term and the regressors is assumed to be zero. If this assumption holds, then the differences across the countries are uncorrelated with the regressors. The results of the R-squared shows that 54% from the variance of the consumer confidence index is explained by the model.
Secondly, we will focus on the regressors. The coefficients show that the CPI, unemployment rate, real house price index and long-term index rates have a negative relation with the consumer confidence index. This result is partly aligned with the results from Ludvigson (2002) and Leeper (1992), who concluded that interest rates do not have a significant effect. Besides this, the lagged variables of period 1, 3, 4 and 5 of the DTHC also have a negative relationship with the consumer confidence index. The P values reflect the statistical significance of the regressors. It shows that the CPI, unemployment rate, share price index, real house price index, short-term interest rates and GDP all have a significant effect on the consumer confidence index. The long-term interest rate and all the lagged dummy variables do not have a significant effect on the consumer confidence index. It does show that the P values of the lagged DTFC are intensely lower than the P values of the DTHC. This means that the terrorist attacks that occurred in another country have more impact on the consumer confidence index of the home country than the terrorist attacks that ensued in the home country. Both dummy variables are not smaller than 0.05 and thus do not have a significant effect on the CCI. Ultimately, from this the hypothesis for both dummy variables holds.
5.2 Random effects model without lagged dummy variables
First of all, we will have a look at the model in general. The F-test of the model shows that all the coefficients in the model are different from zero. This reflects that the model is ok. Moreover, the correlation between the error term and the regressors is assumed to be zero. This shows that the differences across the countries are unrelated to the regressors. The results from the r-squared shows that 53% from the variation of the consumer confidence index is explained by the model.
Next, we will have a look at the regressors. The coefficients show that CPI, unemployment rate and real house price index have a negative effect on the consumer confidence index. The P values of the regressors show that CPI, unemployment, share price index, real house price index, GDP and DTFC have a significant effect on the consumer confidence index. Hence, the short-time interest rates, long-term interest rates and the DTHC do not have a significant effect on the consumer confidence index. This result is in alignment with the results from Ludvigson (2002) and Leeper (1992), who concluded that the short-term and long-term interest rates do not have a significant effect. The dummy variable DTHC is not smaller than 0.05 and so does not have a significant effect on the consumer confidence index.
The hypothesis for DTHC holds. The dummy variable DTFC is smaller than 0.05 and so does have a significant effect on the consumer confidence index. Therefore, the hypothesis for DTFC does not hold.
5.3 Long term effects
The results show that the combined coefficients of the lagged DTFC is significant. So there is a long term effect of the terrorist attacks that happened in a foreign country on the consumer confidence index of the home country. The coefficient shows a positive effect of the attacks on the consumer confidence index.
The test also shows that the combined coefficients of the lagged DTHC is not significant. Thus, there is no long term effect of terrorist attacks that occurred in the home country on the consumer confidence index of the home country. In contrast to the positive relation of DTFC on the consumer confidence index, the coefficient of DTHC shows a negative relation with the consumer confidence index.
6. Conclusion
The hypothesis explained on page 17 illustrated that a slight effect of the terrorist attacks on the consumer confidence index was expected. Two hypotheses were drawn. The first hypothesis was that terrorist attacks that happened in the home country, do have a significant statistical effect on the consumer confidence index of the home country. The second hypothesis was that terrorist attacks that occurred in a foreign country, do have a statistical significance effect on the consumer confidence of the home country.
H0: P-value of dummy terrorist attack in home country = 0.05
H1: P-value of dummy terrorist attack in home country < 0.05
H0: P-value of dummy terrorist attack in foreign country = 0.05
H1: P-value of dummy terrorist attack in foreign country < 0.05
For the random effects model with lagged dummy variables, both hypotheses hold. This means that for both dummy variables, no significant effects on the consumer confidence index were detected. For the random effects model without lagged dummy variables, the hypothesis for the dummy variable terrorist attack in home country holds. The hypothesis for the dummy variable terrorist attack in foreign country does not hold. This means that there is only a significant effect of the terrorist attack that happened in a foreign country on the consumer confidence index.
The results of the test for the long term effects contribute to the results of the model without lagged dummy variables, and show that only the long term effect of the terrorist attacks that happened in a foreign country is significant. There is no significant long term effect of the terrorist attacks that occurred in the home country.
7. Discussion
Garner (2004) was previously the only one who observed the effect of terrorist attacks on the consumer confidence of a country. This research has contributed to the discussion and has provided new results. These results allow for a deeper understanding of how our economy nowadays deals with shocks like these. This is important to know so that we know how to anticipate on these shocks. It seems that our economy is very resilient and that terrorist attacks do not have a large impact on the feelings and trust of civilians.
This research has focused on western European countries that have recently suffered from large-scale terror attacks. To be able to give a more generalized answer, more countries from varied locations and a longer time frame should be taken in account. Also, these regression models only explain 54% and 53% of the variation of the consumer confidence index. To make an even more efficient model, a deeper investigation on the chosen
explanatory variables is needed. Therefore, one can see that there are some limitations to this research.
Interestingly, the results highlight that the short term and long term dummy variable terrorist attack that happened in the home country have a negative relationship with the consumer confidence. Besides this, only the dummy variable terrorist attacks that happened in a foreign country have a positive significant effect on the consumer confidence. It is unclear what the reason could be for these results, that’s why further research could be taken in consideration.
8. References
Abaide, A., & Gardeazabal, J. (2008). Terrorism and the world economy. European Economic Review, 51(1), 1-27.
Acemoglu, D., & Scott, A. (1994). Consumer confidence and rational expectations: are agent’s beliefs consistent with the theory? The Economic Journal, 204(422), 1-19.
Arin, K.P., Ciferri, D., and Spagnolo, N. (2008). The price of terror: The effects of terrorism on stock market returns and volatility. Elsevier, 101(3), 164-167.
Baltagi, B. (2005) Econometric analysis of Panel Data. West Sussex, England: John Wiley & Sons, Ltd.
Christelis, D., & Georgarakos D. (2009). Household Economic Decisions under the shadow of terrorism. CFS Working Paper, 2008(56).
Consumer confidence. 2017. In oxforddictionaries.com. Retrieved November 15, 2017, from
https://en.oxforddictionaries.com/definition/consumer_confidence
Garner, C.A. 2002. Consumer confidence after September 11. Federal Reserve Bank of Kansas City.
Global Terrorism Database. (2017). GTD data. Retrieved from
https://www.start.umd.edu/gtd/
Leeper, E.M. 1992. Consumer attitudes: King for a day. Federal Reserve of Atlanta.
Ludvigson, S.C. 2004. Consumer confidence and consumer spending. Journal of Economic perspectives, 18(2), 29-50.
The Organization for Economic Co-operation and Development. (2017). OECD data. Retrieved from http://www.oecd.org
Wooldridge, J. M. (2001) Econometric analysis of Cross Section and Panel Data. Cambridge, Massachusetts: The MIT Press
9. Appendix
➢ Consumer confidence index
➢ Consumer Price Index (Inflation)
➢ Share price index
➢ Real house price
➢ Long-term interest rates
➢ The Hausman test
(V_b-V_B is not positive definite) Prob>chi2 = 0.9745
= 2.72
chi2(9) = (b-B)'[(V_b-V_B)^(-1)](b-B) Test: Ho: difference in coefficients not systematic
B = inconsistent under Ha, efficient under Ho; obtained from xtreg b = consistent under Ho and Ha; obtained from xtreg DTFC .3651322 .3712022 -.0060699 . DTHC .2200452 .173642 .0464033 . GDP .7243378 .7868417 -.0625039 .0139316 LTIR .044823 .0153744 .0294487 .0151196 STIR .0623468 .0976278 -.0352809 .0163887 RHPI -.0306826 -.0267261 -.0039565 .0015907 SPI .0181502 .0164077 .0017425 .0007154 UMPRT -.0757612 -.0541771 -.0215841 .0149941 CPI -.1708224 -.1419537 -.0288687 .0139268 fixed random Difference S.E.
(b) (B) (b-B) sqrt(diag(V_b-V_B)) Coefficients
➢ Panel data regression
rho 0 (fraction of variance due to u_i)
sigma_e .80457971 sigma_u 0 _cons 101.0223 .4981972 202.78 0.000 100.0458 101.9987 DTFC .3712022 .1715344 2.16 0.030 .0350009 .7074034 DTHC .173642 .2772821 0.63 0.531 -.3698209 .7171048 GDP .7868417 .0759976 10.35 0.000 .6378891 .9357943 LTIR .0153744 .0751237 0.20 0.838 -.1318655 .1626142 STIR .0976278 .0522144 1.87 0.062 -.0047107 .1999662 RHPI -.0267261 .0038897 -6.87 0.000 -.0343499 -.0191024 SPI .0164077 .0024809 6.61 0.000 .0115453 .0212701 UMPRT -.0541771 .0131987 -4.10 0.000 -.0800461 -.0283081 CPI -.1419537 .04219 -3.36 0.001 -.2246446 -.0592628 CCI Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000 Wald chi2(9) = 372.56 overall = 0.5310 max = 71 between = 0.3123 avg = 67.8 within = 0.5422 min = 58 R-sq: Obs per group:
Group variable: Countrynum~r Number of groups = 5 Random-effects GLS regression Number of obs = 339
➢ Panel data regression with lagged dummy variables for 5 years
.
rho 0 (fraction of variance due to u_i)
sigma_e .78056982 sigma_u 0 _cons 101.1964 .5379821 188.10 0.000 100.142 102.2508 L5. .0031303 .2237657 0.01 0.989 -.4354424 .4417031 L4. .3144117 .2158309 1.46 0.145 -.1086092 .7374325 L3. .2675603 .1942871 1.38 0.168 -.1132355 .6483561 L2. .2730464 .1862313 1.47 0.143 -.0919604 .6380531 L1. .2973911 .1816506 1.64 0.102 -.0586376 .6534198 --. .1671423 .1825068 0.92 0.360 -.1905644 .524849 DTFC L5. -.0756974 .3960203 -0.19 0.848 -.851883 .7004881 L4. -.0158135 .395768 -0.04 0.968 -.7915046 .7598776 L3. -.0004281 .3386854 -0.00 0.999 -.6642393 .6633831 L2. .0239482 .3065013 0.08 0.938 -.5767834 .6246798 L1. -.1150909 .2959361 -0.39 0.697 -.695115 .4649333 --. .0265862 .2873267 0.09 0.926 -.5365638 .5897362 DTHC GDP .755045 .077105 9.79 0.000 .603922 .9061681 LTIR -.007351 .0788195 -0.09 0.926 -.1618345 .1471324 STIR .1148849 .0531414 2.16 0.031 .0107297 .2190402 RHPI -.026386 .0044767 -5.89 0.000 -.0351602 -.0176119 SPI .0147873 .00255 5.80 0.000 .0097894 .0197853 UMPRT -.055597 .0133842 -4.15 0.000 -.0818296 -.0293645 CPI -.1683746 .043566 -3.86 0.000 -.2537625 -.0829868 CCI Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000 Wald chi2(19) = 351.00 overall = 0.5400 max = 66 between = 0.3144 avg = 63.8 within = 0.5534 min = 58 R-sq: Obs per group:
Group variable: Countrynum~r Number of groups = 5 Random-effects GLS regression Number of obs = 319
➢ Wald-test
.
(1) -.1564956 .7757138 -0.20 0.840 -1.676867 1.363876 CCI Coef. Std. Err. z P>|z| [95% Conf. Interval] ( 1) DTHC + L.DTHC + L2.DTHC + L3.DTHC + L4.DTHC + L5.DTHC = 0 . lincom DTHC+l.DTHC+l2.DTHC+l3.DTHC+l4.DTHC+l5.DTHC (1) 1.322682 .4178291 3.17 0.002 .503752 2.141612 CCI Coef. Std. Err. z P>|z| [95% Conf. Interval] ( 1) DTFC + L.DTFC + L2.DTFC + L3.DTFC + L4.DTFC + L5.DTFC = 0
![Formation of dinuclear palladium(I)Hydride [Pdsub2(?-H)(?-CO){(S,S)-BDPP}sub2]Cl by methanolysis or hydrolysis of Pd(COMe)(Cl){(S,S)-BdPP}{(S,S)-BDPP=(2S,4S)-2,4-bis(diphenylphosphino)pentane} - 3997y](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACH5BAEAAAAALAAAAAABAAEAAAICRAEAOw==)