Resilience of the EU and Leverage of the European Neighbourhood Policy: Good News and Bad News

Abstract

Since her very beginning, the European Union (EU) has been subject to various challenges, even called crises by common parlance. In this chapter, we argue that the EU shows remarkable stability and significantly strong resilience to conflicts. Testing this statement, we deal with time series analysis, that is, Vector Auto Regression Modelling is used to test this statement for the purpose of impulse response function interpretation. It is shown that the onset of much debated challenges the EU faced has even statistically significantly increased the overall positive opinion of Europeans on the EU, thereby strengthening her legitimacy and political significance. Findings prove significant characteristics of strong resilience to shocks. Implications for the leverage of the European Neighbourhood Policy (ENP) point to low leverage of ENP due to a biased use of instruments. To improve the leverage of ENP, its benchmark indicators have to meet concerns about the dynamics of hybrid democracies.

Appendices

Annex 7.1: VAR Models

As the data are compiled in time series format, this chapter draws on time series econometrics. The impact of the above listed crises will be evaluated based on the responses they cause in the two time series. This is accomplished by calculating the impulse response functions for 12 vector auto-regression models (VAR) of which each corresponds to one crisis and one of the two above-mentioned variables. The crises are represented by dummy variables, which take the value 1 in the year of the onset of the crisis and 0 otherwise. The impulse response functions allow tracing the dynamic impact of the crisis impulse on the respective time series.

The vector auto-regressions methodology has found broad application in macroeconomics. Even though this methodology has found some application in the field of politics (e.g. Enders and Sandler 1993), it is comparatively uncommon yet immensely useful (compare, Freeman et al. 1989). In contrast to mere eyeballing, this method of time series econometrics allows to distinguish between statistically significant and statistically insignificant responses, factoring in the standard deviation in the time series in question.

The following formula represents a basic, bivariate VAR model in any of the data points of a time series that is potentially affected by the values of itself or the other time series from up to six time periods ago.

Hence, for instance, the time series of the social stability index or the net public approval time series will be explained on the basis of their own weighed lags as well as the weighed crisis dummies, that is:

$$ {y}_{1,t}={c}_{1,t}+{\varphi}_{1,1}^1\ast {y}_{1,t-1}+{\varphi}_{1,2}^1\ast {y}_{2,t-1}+\dots +{\varphi}_{1,1}^p\ast {y}_{1,t-p}+{\varphi}_{1,2}^p\ast {y}_{2,t-p}+\dots +{\varphi}_{1,1}^6\ast {y}_{1,t-6}+{\varphi}_{1,2}^6\ast {y}_{2,t-6}+{e}_{1,t} $$

The impulse responses were calculated on the basis of recursive VARs in differences, in which the dummy time series is ordered before the respective response time series. A model in differences uses the first differences of the time series in question to account for non-stationarity exhibited in the time series. For a detailed mathematical treatment of VARs and impulse response functions, refer to Stock and Watson (2001) and Hamilton (1994).

Annex 7.2: Data

1.1 Public Approval

The presented values for public approval and public disapproval are taken from the since 1973 published Eurobarometer. Public approval is defined as the ratio of people answering the following question ‘Generally speaking, do you think that (OUR COUNTRY)’s membership of the European Union is…?’ with ‘A good thing’. On the contrary, public disapproval is defined by the ratio of those who answer the same question with ‘A bad thing’. The values until 2011 could be taken from the public opinion webpage of the European Commission (2016), subsequent values were taken from two publications from the European Parliament (2013, 2015a, b). Until 2011, the Eurobarometer reports half-yearly values and yearly values thereafter. The time series was collapsed to yearly data points. Missing data points were dealt with by means of averaging of the data point before and after it. On the basis of these values for public approval and disapproval, the net approval rates were calculated by subtracting the disapproval time series from the approval time series. The resulting time series is shown hereafter.

The presented values for public approval and public disapproval are taken from the Eurobarometer (published since 1973). Public approval is defined as the ratio of people answering the following question ‘Generally speaking, do you think that (OUR COUNTRY)’s membership of the European Union is…?’ with ‘A good thing’. On the contrary, public disapproval is defined by the ratio of those who answer the same question with ‘A bad thing’. The values until 2011 could be taken from the public opinion webpage of the European Commission (2016), subsequent values were taken from two publications from the European Parliament (2013, 2015a, b). Until 2011, the Eurobarometer reports half-yearly values and yearly values thereafter. The time series was collapsed to yearly data points. Missing data points were dealt with by means of averaging of the data point before and after it. On the basis of these values for public approval and disapproval, the net approval rates were calculated by subtracting the disapproval time series from the approval time series. The resulting time series is shown hereafter.

1.2 Construction of the Social Stability Index

The Social Stability Index (SSI = ln (GDP per capita/political terror scale) used in the time series on which the models are based is constructed on the basis of the real GDP capita value measured in constant US dollars of 2010 as reported by the World Bank (2016). For every year, the real GDP per capita and political terror scale values are the averages of the values for the countries with EU membership, that is, 6 in the beginning and 28 in the end. Every enlargement is accounted for by including the respective countries into the arithmetic mean. The political terror scale links up to the values reported by the US state department and available as download from the webpage of the project The Political Terror Scale (2016).

Annex 7.3: Charts

Fig. 7.6

Components of SSI (Source: Author’s representation)

Fig. 7.7

Impulse responses for public approval for the EU (Source: Author’s representation)

Fig. 7.8

Impulse responses for the social stability index for the European Union (Source: Author’s representation). For the interpretation of Fig. 7.8, mind that the response variable is log-transformed

Annex 7.4: Data—Descriptive Statistics

Variable	Min. Year	Max. Year	Min.	Max.	Average	Std. Dev.
Social stability index	1983	2000	9.719225	10.3164	10.00716	0.1317246
Political terror scale	multiple	2007	1	1.62963	1.19919	0.1986472
GDP/capita (2010 $)	1976	2008	18076.04	34601.46	26841.73	5592.251
Public approval	2011	1991	0.4658	0.68495	0.5489961	0.0538752
Public disapproval	1991	2011	0.0704	0.185	0.1312479	0.0276633
Net approval	2011	1991	0.2808	0.6146	0.4177526	0.0790581

1. Time Series Data is available on demand
2. Source: Author’s representation

Annex 7.5: Sketch of Stability conditions for a Negotiation Model

A stability analysis of such a process hints at formal mathematical implications for resilience. With

$$ \overline{x}:\mathrm{benchmark}\ \mathrm{equilibrium}\ \mathrm{of}\ \mathrm{the}\ \mathrm{system}\ (EU) $$

$$ {x}_t:\mathrm{state}\ \mathrm{of}\ \mathrm{the}\ \mathrm{system}\ (EU) $$

$$ {b}_t:\mathrm{negotiating}\ \mathrm{efforts} $$

$$ \alpha, \beta >0 $$

denoting variables and parameters of a potential model, interaction between nations can be described by a set of differential equations:

$$ {b}_{t+1}=\alpha {\left({x}_t-\overline{x}\right)}^2 $$

(Eq. 7.1)

and

$$ {x}_{t+1}=-\beta {b}_t\left({x}_t-\overline{x}\right)+{x}_t $$

(Eq. 7.2)

where t denotes (points in) time.

Once the state of the system deviates from its equilibrium due to a symmetric shock, governmental negotiations start (Eq. 7.1). On the grounds of institutional settings and vis-a-vis tit-for-tat rules of the game political reforms (Eq. 7.2) will be implemented.

Local stability around the steady state (Jacobi-Matrix) would reveal Eigenvalues between 1 and 0 which denote Grenzstabilität, a stable cycle. Chaos seems possible face to global (large) deviations because of non-linearities involved.