Political economy of conflict
:
Social welfare cuts and social unrest in the
OECD
Demis Iossifidis
10468811
27/06/2014
Introduction
Only twenty-five years ago, Europe still had a Berlin Wall, a Yugoslavia and a Soviet Union. In 2010, a revolutionary wave of demonstrations and protests led to the Arab Spring. At the beginning of March 2014, Crimea was still part of Ukraine; a couple of tumultuous weeks later, it is a part of Russia and as this paper is being written, in Eastern Ukraine pro-Russian separatists are engaged in an armed conflict with the Ukrainian army . In the course of history, manifestations of social unrest have succeeded each other at a high rate and have led to important transitions in regime, borders and religion. With every new manifestation of social unrest there seem to be new mechanisms and different complexities underlying the onset and the intention of the social unrest. And often we see scholars, analysts and journalists attempting to map and even predict the onset of social unrest. An interesting example is the ‘’Risk of social unrest’’ table published by The Economist. This journal published a prediction for 2014 with various places all over the world where social unrest is likely to occur. While such ‘’trendwatchers-like’’ predictions result in smooth and –seemingly- insightful tables, their value is doubtful.
Source: www.economist.com (search for: social unrest 2014)
It is doubtful because who could have predicted that, a few weeks or even days before the start of some small protests in Tunisia, those protests would spread like a wildfire to Egypt and eventually to the whole North African and Arab region? Even the US intelligence services were confident that the Mubarak regime was safe enough not to be crumbled by a handful of demonstrators who appeared in the streets of Cairo every so often (Norton-Taylor, 2011). If we go further back in time, we see that the same goes for the collapse of the Soviet Union (Wallerstein, 2002) and for recent events such as in Crimea. It is therefore safe to say that what leads to social unrest is often far from clear; the direct cause or causes are often complex and can lead to unexpected changes. The
phenomenon of social unrest has been attributed to a variety of social, political, economic and environmental causes; from food scarcity to economic shocks. In this study, the focus will be on one specific potential cause: cuts in social welfare expenditure. Since the economic crisis of 2008, there have been drastic cuts in social spending in most of the OECD countries.1 The disturbances in various parts of Europe that arose during the
economic crisis have been linked to these cuts –both by popular opinion and by scholars. In this study, I contribute to the limited existing literature on this topic, by looking at the relationship between social welfare cuts and social unrest. I have analysed the period between 1980 until 2011 for 23 countries of the OECD, but with special attention to the European Union.
Earlier studies, most notably by Ponticelli & Voth (2011) and Taydas & Peksen (2012), have shown clear negative correlation between cuts and social unrest, but never has there been a study on the two specific variables used in this study. While the outcomes in this study are roughly in line with some of the earlier findings, the story appears to be more ambiguous than earlier studies suggest; as the results section will show. In this thesis I also intend to get a more detailed picture of how the relationship works. As will become clear in this piece, I consider the possibility that specific branches of social welfare expenditure such as unemployment, old-age, health, family and housing expenditures might have distinct implications for social unrest uprising. In addition to this I make one last contribution in this thesis, which is also the most important one. I pose the question whether Southern European countries have different experiences than Northern European countries, with respect to the effects of social expenditure on unrest. It is my claim that the relationship works out differently across different kinds of welfare states and in different political settings; and that one important distinction that we can make is that of Northern Europe and Southern Europe. The results, explained in the results section, strongly support this claim. An insightful way to think about this claim is to compare it with the occurrence of a forest fire. Most of the time, you can throw away a burning cigarette in the middle of the forest and nothing severe will happen. The flame of the cigarette will simply fade away. See such conditions as Northern European conditions, like in The Netherlands or Germany. But in other, particularly dry conditions everything will go up in flames. Such conditions we find in for instance Greece or Spain. The burning cigarette might be a small incident like in Athens 2008, when a 15-year old student was killed by two policemen. In the ‘’dry forest’’, Greek society, it led to the worst riots in 40 years. But –to give an example closer to this research- it can also be the implementation of austerity measurements by the government. There are numerous factors that can lead to a ‘’dry forest’’ and I am explaining these factors in this paper.
The economic crisis of 2008
From 2008 until 2012 Europe suffered from, as what most scholars describe, the deepest economic crisis since the Great Depression (Eichengreen & O’Rourke, 2009). Various American and European banks declared to have suffered massive financial losses at the beginning of 2008. Soon, panic on the financial markets led to severe downturns on the stock markets. Various banks were bailed out in order to prevent them from collapsing;
other banks collapsed. In Greece, Prime Minister George Papandreou revealed a black hole in the country’s accounts in late 2009 and the BBC reported that Greece’s debts had reached 300 billion euros.2 Within weeks ratings agencies started downgrading Greek
banks and the government debt. Meanwhile concerns started to build about other indebted countries in Europe; most notably Portugal, Ireland and Spain. The European Union was facing the deepest economic crisis in its existence; budget deficits, unemployment rates, bank runs and downgrades by ratings agencies became Europe’s new reality. A European rescue package was fastened for Greece in May 2010 (Ruffert, 2011). Ireland, Portugal, Spain and Cyprus followed. The German ‘’obsession’’ with keeping budget deficits down was imposed upon the other members of the European Union which led to rigorous measures of budget cuts that were introduced in order to reduce deficits (Engelen, 2014). Especially in Southern Europe, these policies led to tough economic times. In Greece for instance, the cumulative reduction in GDP went just under 20% in the period 2008-2012.3
Literature review
But the chosen path of austerity was not an undisputed one. An important discussion arose about the question whether austerity measures are the best remedy during economic crises. Some scholars publicly took a stand against the overwhelming political support for these measures. One of them was Ewald Engelen, professor at the University of Amsterdam, who stated that budget deficits are nothing less than a necessity for economic growth and that a complete focus on cuts –like in the economic crisis- was undesirable and destabilising (Engelen, 2014). But other scholars, like Alesina & Adragna, supported the austerity measures and claimed that austerity actually triggers economic growth (Alesina & Ardagna, 2009).
But the actual roots of this debate in political sphere can be found in a paper published after Alesina & Ardagna, called “Growth in a Time of Debt”. Harvard economists Carmen Reinhart and Kenneth Rogoff published it in 2010 and it had acquired touchstone status in the debate over economic policy amongst European policy makers. Looking back, we can conclude that this paper changed the course of policy, as Paul Krugman describes in his paper How the case for Austerity has crumbled. He goes even further, stating that the Reinhart-Rogoff piece may have had more immediate influence on public debate than any previous paper in the history of economics. The paper made a strong case for austerity measures and suggestions that, as John Maynard Keynes once argued, “the boom, not the slump, is the right time for austerity’’ were refuted by Reinhart and Rogoff, showing that waiting would be disastrous; that economies fall off a cliff once government debt exceeds 90 percent of GDP (Krugman, 2013). The 90 percent claim was cited as the decisive argument for austerity by figures ranging from Paul Ryan, the former vice-presidential candidate in the US, who chairs the House budget committee, to Olli Rehn, the top economic official at the European
2
BBC (2009). ‘Greece’s debt reaches 300bn euros’ obtained through: http://news.bbc.co.uk/ (search for: ‘’Greece’s debt reaches’’)
3
The Guardian (2012). ‘Greek economy to shrink 25% by 2014’ obtained through: http://www.theguardian.com/ (search for: ‘’Greek economy to shrink 25% by 2014’’)
Commission, to the editorial board of The Washington Post (Krugman, 2013).
In April 2010, Alesina and Ardagna –who made a case for austerity as well, as I have mentioned- made a presentation to the Economic and Financial Affairs Council of the European Council of Ministers. Their analysis quickly made its way into official pronouncements from the European Commission and the European Central Bank. Thus in June 2010 Jean-Claude Trichet, the then president of the ECB, dismissed concerns that austerity might hurt growth:
‘’As regards the economy, the idea that austerity measures could trigger stagnation is incorrect…. In fact, in these circumstances, everything that helps to increase the confidence of households, firms and investors in the sustainability of public finances is good for the consolidation of growth and job creation. I firmly believe that in the current circumstances confidence-inspiring policies will foster and not hamper economic recovery, because confidence is the key factor today.’’ 4
The influence of both the Alesina-Ardagna and the Reinhart-Rogoff paper resounds through the opinion of Trichet in this interview. Politicians, policy makers, the ECB, the IMF –they all were biased in their decisions during the economic crisis, as the IMF later admitted.5 Did this bias make policy makers insensitive for possible social consequences?
This question is not easy to answer, but we can say with certainty is that, similar to the more fiscal focused debate about whether austerity leads to growth or to more economic decline, there is a discussion with regard to the social consequences.
Most papers on the topic come from the 90s and those were focused on case studies and on developing regions, mainly in Africa and Latin America. Paldam (1993) examines the relationship between austerity measures and unrest in nine countries in South America, using high-frequency data, and he finds that the run-up to new austerity measures can be linked to higher levels of unrest, but that actual implementation of austerity measures is followed by fewer disturbances. Haggard, Lafay and Morrison (1995) focused on developing countries, looking at cuts imposed by external parties, particularly the IMF. They found that IMF interventions and monetary cuts led to greater instability in those countries. Furthermore, Woo (2003) showed that countries with higher levels of unrest are likely to be more indebted. A more recent contribution comes from Taydas & Peksen (2012), called ‘’Can states buy peace?’’. Their study is the first thorough empirical analysis of whether social welfare spending is a significant predictor of civil war onset raised the question: how does social welfare provision affect the likelihood of civil war onset? The first part of their argument follows the logic of raising opportunity costs: social welfare efforts by the state can shape the preferences of citizens in ways that discourage the use of violence for political goals. The second part of the argument is the claim that government commitment to redistributing in favor of the weak has an important impact on the perceptions and preferences of social actors towards the state. In return for productive social welfare policies, political leaders gain public loyalty, compliance and support (Taydas & Peksen, 2012 p.276). Taydas & Peksen
4
European Central Bank: ‘Interview with Jean-Claude Trichet and La Repubblica’, obtained through: http://www.ecb.europa.eu/ (search for: ‘’Interview with La Repubblica’’)
5
The Guardian: ‘IMF admits: we failed to realise the damage austerity would do to Greece’, obtained through: http://www.theguardian.com (search for: ‘’IMF admits damage austerity’’)
use the rather high 25 battle-related deaths per country-year threshold to examine whether social welfare spending has an impact on civil conflict, and with significant results. Even with a higher threshold, of 1000 battle-related deaths, their findings remain significant. This concept of ‘’buying peace’’ is also supported by Blau (2000 p.298), who states that social unrest is an important trigger for policy makers to make changes in welfare expenditure. Blau describes that while manifestations of social unrest are often condemned by government officials, spending more on social welfare is an effective method to calm things down. But Blau finds that once the tumult dies down, the spending is usually cut back again.
A very relevant and recent contribution comes from Ponticelli & Voth (2011). With their study they take a clear stand against the austerity measured introduced during the economic crisis. In their Europe based research they looked at the relationship between changes in national government expenditure and social unrest. They took a very broad period; from 1919 to 2009, asking the question: to what extent do societies become unstable after budget cuts? Their results support the claim that fiscal consolidation leads to social unrest. With this finding, they contradict the findings of Alesina et al. In 2010 Alesina published a study together with Carloni and Lecce (2010) in which the conclusion was that ‘’cutting our way back to growth’’ is not only desirable but also politically feasible. Their findings showed that that there is no real punishment at the polls for governments pursuing cut-backs. Ponticelli & Voth (2011) reject this claim. According to them, Alesina and the earlier mentioned Reinhart & Rogoff do not take into account the possibility of social and political unrest.
The disagreement amongst scholars can partly be explained by the fact that the approaches scholars choose to look into this relationship differ widely. Alesina et al (2010) for instance use government re-elections as an indicator for dissatisfaction in the population, Taydas & Peksen look at battle-related deaths as an indicator for conflict and Ponticelli & Voth focus on events of social unrest such as riots and demonstrations. Not only the interpretation of the dependent variable varies from study to study, also the way that austerity policies are measured is different. Ponticelli & Voth for instance, choose the more general ‘’budget cuts’’ as an indicator which is more an indicator for state capacity and Taydas & Peksen look into the more specific ‘’social welfare spending’’, which says more about the willingness to redistribute resources. I believe that an important contribution in the field can be made by looking into a combination of those last two stories. I am looking at the specific kind of expenditure (social welfare expenditure) that Taydas & Peksen choose instead of budget spending in general. And I am looking at an – for the OECD- more relevant social phenomenon (social unrest) instead of the–less common in developed countries- civil conflict variable. As I have shown in this literature review there is no consensus on the effect of cuts on social unrest, and it even seems that during the recent economic crisis, policy makers have chosen for the camp of scholars that would legitimate their actions most.
But the burning barricades in downtown Barcelona or at Syntagma square in Athens are an important reason for me to expect something different. For that reason my first hypothesis is:
H1 Cuts in social welfare spending lead to social unrest
In this main hypothesis we find the two variables that constitute the core of this research. The dependent variable is social unrest, which is a concept that covers the following expressions of social unrest:
Political assassinations
Demonstrations
Riots
Government crises
These indicators of social unrest come from the Cross-National Time-Series Data Archive. This extensive database gives us the data on social unrest from 1980 up until 2011. While initially I had intended to look at the effect of social welfare cuts on each of those indicators of unrest, this is not meaningful. There are no convincing theoretical grounds on which I could expect that cuts in social welfare (branches) would lead to a specific kind of unrest. In order to keep the focus on a meaningful and theoretically convincing claim instead of some vague expectations about possibly stronger effects, I will limit the social unrest variable to one. This one variable is called the weighted
conflict index and it constitutes a multiplied number of the five above mentioned
indicators.6This means in short that we will be looking at all forms of social unrest, while
each of those forms is being weighted by its impact.
Why do I focus specifically on social welfare spending as the independent variable, why not on general government expenditure? The reason is that social welfare spending is more than simply a statistical figure; it is an indicator that the state cares about its citizens. By investing in social safety nets, in-kind transfers and valuable goods that are underprovided by the private sector or that are only available to certain segments of society, the government can make an immediate, direct impact on people’s conditions. As Taydas & Peksen (2012) explain, such welfare policies can mitigate the effects of poverty, lower societal grievances originating from absolute and relative deprivation, improve the conditions of the poor and decrease socio-economic inequality. Bueno de Mesquita et al (2003) add to this that government spending decisions are of decisive importance for the political survival of governments and the stability of the system. Given the fact that citizen attitudes are informed by self-interest, redistributing state resources and allocating revenues to projects that benefit a large segment of society, can enhance the ties between citizens and the state. And it enhances the power base of the regime. Higher government social spending shows citizens that the government cares about their constituency and that it has an interest in improving their life (Taydas & Peksen, 2012 p.276). When focusing on government spending in general, this specific feature of social
6
The weighted conflict index is calculated in the following manner: Multiply the value of the number of assassinations by 24, general strikes by 43, guerrilla Warfare by 46, government crises by 48, purges by 86, riots by 102, revolutions by 148, anti-government demonstrations by 200. Sum the 8 weighted values and divide by 9. The result is the value stored as the Weighted Conflict Index.
General strikes
Attempted revolutions
Guerrilla warfare
welfare expenditure gets lost in the more general mixture of other expenditures such as defense or recreational facilities for instance. The explanatory power of this research stays high by choosing for such a delineated variable.
The CPDS (Comparative Political Database) will be used for data on social welfare expenditure. It covers the years between 1980 and 2009 (except for Australia, Germany, Ireland, New Zealand and United States: until 2010). Given the particular interest that I have expressed in the recent economic crisis, it is relevant to obtain expenditure data that is as recent as possible. For that reason, I am using a second dataset, extracted from the EUROSTAT dataset on social expenditure, for the period 2002-2011. In a separate table in the results section, I am presenting the results from that more recent period of time. The way in which the CPDS dataset is constructed, gives us the opportunity to look at more than just social welfare spending in total. The dataset provides various specifications of welfare spending. Why is it meaningful to look at specifications of the welfare state and their effects on social unrest? This is because each specification can be seen as a category of policies providing redistribution for a specific group within society. We need to recognize that the welfare state has many faces: every change in one of the specifications of welfare expenditure may have an effect for a specific group of people. Unemployment policies for instance, will mostly affect younger and middle aged people. Pensions on the other hand affect only the elderly. It would be a missed chance, an act of carelessness, to simply throw these different groups within society on one pile and to assume that every social policy is meant for society as a whole. This is obviously not the case. The CPDS dataset distinguishes various specifications of social welfare expenditure. These include:
Total public and mandatory private expenditure on old age as a percentage of GDP
Total public and mandatory private expenditure on survivor benefits as a percentage of GDP
Total incapacity-related benefits (public and mandatory private) as a percentage of GDP
Total public and mandatory private expenditure on health as a percentage of GDP
Total public and mandatory private expenditure for families as a percentage of GDP
Total public and mandatory private expenditure on active labour market programmes as a percentage of GDP
Cash expenditure for unemployment benefits as a percentage of GDP (public and mandatory private)
Total public and mandatory private expenditure on housing as a percentage of GDP
There are reasons to look closer at four of those specifications which I believe capture best the experience of ‘’feeling protected by the government’’ through social programmes. I am going to look whether cuts in those specific branches have a stronger or weaker (depending on my expectation per branch, which I will motivate here below) effect on social unrest, compared to the effect of the total social expenditure variable.
While all of these specifications contribute to the social safety net provided by the welfare state, I choose to leave some branches out. The active labour market
programmes for instance, capture this idea of protection by the government less. Also,
I choose to leave out survivor benefits, incapacity-related benefits and family
benefits because they form a very small cost in comparison with the four I did choose
and they are focused on very specific and small groups within society. In Germany for instance, 9,1% of GDP is spent on old age benefits and only 2,1% on survivor benefits. Cuts in these branches are likely to have less societal impact. The four specifications I choose are almost all big in terms of percentage of GDP and relevant to bigger groups in society. This brings us to the following independent variables for the second part of the thesis:
Old age; the largest share of social welfare expenditure within OECD countries
goes to the elderly, so it is insightful to look at the effect that cuts in this branch have on social unrest. Also, the living standard of many elderly depends solely on the level of old age expenditure. This indicates that this variable highly contributes to the earlier mentioned ‘’experience of protection’’. My expectation is that the elderly are less likely to hit the streets and to participate in protest movements, because of the simple reason that elderly are limited in their mobility and they are less likely to initiate conflict.
Unemployment; this specific branch of social welfare spending will be tested
because I expect it to have a particularly high impact on social unrest. Given the value that people attach to their jobs, unemployment is generally perceived as a painful situation. As early as the 1930s scholars have investigated the phenomenon of ‘’social isolation’’ as a consequence of unemployment (Tazelaar & Sprengers, 1984). Once a person is unemployed, he or she is dependent on unemployment benefits and this creates a similar dependency position as that of the elderly; you need government support to survive. Since anyone who is employed is a potential candidate for unemployment benefits in times of crisis, I expect that cuts in this branch will have a high effect on social unrest.
Health; people attach high value to their health, as appears from the many
sayings existing about it (‘’the greatest wealth is health’’, to name one). Health is the core element of human welfare. It is my expectation that people that depend on government support for affordable health, attach high value to this government service. For this reason, I have a similar expectation for health as I have for unemployment: a high effect on social unrest.
Housing; an affordable place to live is a high priority for people. Especially the
weaker groups in society will experience support for a roof above the head as an obligation that their government has. Deprivation of this support will be experienced as a breach of people’s basic needs. Thus I expect, similar to what I expect for unemployment and health benefits, that this effect will be stronger. My explanation above for choosing these specifications of social expenditure should be read with the knowledge that there is no thorough theoretical frame that supports these choices, simply because this has not been done before by scholars when looking at social unrest. Having said this, I present to you the hypotheses:
H2a Cuts in old age benefits have a lower effect on social unrest
H2b Cuts in unemployment benefits have a high effect on social
unrest
H2c Cuts in health benefits have a high effect on social unrest
H2d Cuts in housing benefits have a high effect on social unrest
I have no theoretical reasons to expect a difference in how strong the effect is, between the last three branches; unemployment benefits, health benefits and housing benefits. So that is why I express hypotheses H2b H2c and H2d in the same manner.
Southern European exceptionalism?
There are reasons to believe that the relationship between social welfare and unrest is country-specific and dependent on specific assets of the welfare states -and the level of development of that welfare state. As I have mentioned briefly in the introduction, I expect the effect of social welfare cuts to be stronger in Southern Europe and the results following from this expectation form the most important part of my contribution.7 In this
part I am explicating the reasoning behind this expectation.
First it is important to ask: can we speak of a Southern European scheme? Yes, the political, economic and social structures of the Southern European countries are more similar to each other than to those of other members of the European Union (Magone, 2003). As Rhodes (2009, p.62) points out, we can indeed speak of a Southern European model. The most important characteristic that the Southern European countries share is that the welfare state model is more modest than that of the Northern European countries (Ferrera, 2005 p.3). Mingione describes the Southern European welfare model as a model based on a weak and inefficient state, with strong regionalism and localism, a persistent diffusion of small family businesses and a relatively high level of family responsibility for welfare services (Mingione, 2001). Consequently, in times of low economic growth, or severe economic decline like during the economic crisis of 2008, there is limited resilience in the system that should protect the people during those
7
In the regression analyses the countries that have been marked as Southern Europe are: Portugal, Greece, Italy and Spain
difficult times.
Attempts to reform the state institutions are heavily constrained by the features of the ‘southern syndrome’: the strength of vested interests, clientelistic collusion, an absence of political consensus and the weakness and fragmentation of administrative structures (Rhodes, 2009 p.15). These systemic flaws have consequences for the political perception in those countries: in Greece 12% of the population stated to have trust in the national parliament, in Spain this was 8%.8 This is a sharp contrast with for instance
Sweden, where the trust percentage was 70%. There is a persistent level of negative expectations and distrust in these Southern European systems (Rhodes, 2009 p.15). In other words: these regimes enjoy less legitimacy (Andreotti et al, 2010).9 But there seems
to be a paradox. Because despite the weaknesses of the Southern European system and despite the lack of trust in government officials, still the support for government intervention in a ‘’welfare state manner’’ is bigger in Southern European countries than in Northern European countries.10 In a survey from the European Social Survey in 2008,
Greece, Spain and Portugal were all in the top 10 of countries in a questionnaire asking people whether government intervention was desirable on various issues (see footnote). The question is: why? Why would you want to be taken care of by a government that you do not trust and that you do not find capable of doing so? The answer is that people demand the government to take on this task, because –much more than in other welfare models- the legitimacy that the regime enjoys stands or falls with its ability to provide social welfare benefits. It is important to emphasize the difference with Northern European state models, where the regime in question obtains legitimacy through many more ways than just providing social benefits (infrastructure and education for instance). But in Greece or Portugal, there are barely any other legitimating policies and interventions that the government is providing in, in a successful matter. So when an economic crisis hits the region, like in 2008/2009, it is in social policy that the regime can derive its legitimacy from. Add to this the fact that the dependency on such policies has been much higher there than in Northern Europe (if we look at the economic crisis, we see that, for instance, unemployment rates in Greece hit 27,3% compared to 6,7% in the Netherlands11) and one can understand how the provision of social services works
through a different mechanism in these countries. Because of this bigger social policy dependency and the already limited legitimacy that these regimes enjoy, the idea of providing social services becomes a part of a fragile balance between legitimacy and dissatisfaction, which leads to a society that is much more sensitive to changes in social welfare expenditure. In such a particular setting, changes in social welfare expenditure are more noticed and more socially controversial acts.
8
Standard Eurobarometer 80, Automn 2013 version 9
I define legitimacy here as a set of positive attitudes of a society towards its democratic institutions, which are considered by the people as the most appropriate form of government (Gunther et al, 1995)
10
European Social Survey Round 4, 2008 version. Respondents were asked how far they thought it should be the government’s responsibility to do each of the following: ensure a job for everyone who wants one; ensure adequate health care for the sick; ensure a reasonable standard of living for the old; ensure a reasonable standard of living for the unemployed; provide paid leave from work for people who temporarily have to care for sick family members.
11
Eurostat unemployment rate 2002-2013
Lastly, I want to bring forward an argument that is not socio-economic and that might be a rather unconventional one, but just as relevant. I want to bring forward a geographical argument: the weather. In the literature, theories of geographical determinism and stereotypes suggest a relationship between climate and emotional intensity and frequency. As early as the first century B.C., Poseidonos contrasted the Norse (Germanic or Celtic tribes living in a cold climate) with the Mediterranean or southerners (living in a moderate climate). In the eighteenth century, Montesquieu came up with a reverse climate theory for emotional experience: warmer climates made southerners more sensitive to emotion (Jahoda, 1992). Recent anthropology holds that among Southern Europeans, in comparison to Northern ones, an emotional (e.g., violent and passionate) culture prevails (Basabe et al, 2002 p.107). Furthermore, violent crime studies support the hypothesis that a warm climate promotes aggression (Basabe et al, 2002 p.107).
My claim is that all of these conditions I have just described are distinctive, Southern European conditions. And together these conditions lead to a much more tumultuous socio-political arena. the dependency on these policies is bigger and the legitimacy of governments is lower. Hence the expectation that cuts in social welfare spending have a stronger effect on unrest in Southern Europe. The hypothesis that follows from this is:
H3: Cuts in social welfare spending have a stronger negative effect on social
unrest in Southern Europe than in Northern Europe Control variables
The relationship between social welfare cuts and social unrest is associated with ‘’hard times’’. We have seen that in times of economic difficulties, Europe has chosen for various measures of fiscal retrenchment. People will be unhappy because of these hard times, so the correlation between cuts in social welfare spending and social unrest might not be causal but simply accidental. That is why we need to control for several factors that influence this situation. The first control variable is changes in the GDP level. By controlling for growth, you take out the component of unrest that is driven by ‘’times being hard’’.
In the regression models I am also controlling for other variables that possibly influence the onset of social unrest. These variables are (abbreviations used in the models is in parentheses):
Amount of leftist parties (gov_left): I am controlling for social-democratic
and other left parties as a percentage of total cabinet posts. Leftist governments tend to invest more in social welfare policies and thus might have an influence on the level of social unrest.
Openness of the economy (openc): the expectation is that while an open
economy may have positive influences due to the economic growth it generates, it also leads to a situation in which a country with more openness is more vulnerable to economic changes in the global economy.
Debt (debt): we control for the gross government debt as a percentage of GDP.
The expectation is that the effect of debt on social unrest is comparable to the effect of GDP growth. It adds to the more general condition of the government.
Net union membership (netu): net union membership is likely to have an
effect on social unrest because of its power to assemble people.
Unemployment (unemp): unemployment can lead to social unrest through a
clear mechanism of discontent about a person’s position within society. In general, governments are usually being blamed for high unemployment, which could lead to anti-government demonstrations.
Population (lpop): the log of total Population is used to control for the expected
positive association between higher population rates and the likelihood of social unrest (Taydas & Peksen, 2012 p.279). While this control variable is commonly used in the literature and might apply on a global scale, I expect that in this particular study the logic works the other way. Because when we look at the members of the OECD, we see that the most developed countries (such as the United Kingdom, the United States and Germany) are also the countries with the highest number of inhabitants. Nonetheless, this control variable will be included.
Years (year): in addition to these control measures, a measure of time is also
added to the models. In a cross-national time-series analysis like ours, there is a risk that the estimation is biased due to time dependency (Beck et al. 1998). Time is likely to influence the observations of social unrest onset: when a country has recently experienced social unrest it is more likely to experience it again in the future. Conversely, a country that has enjoyed peace for a long time, is less likely to undergo a civil war.
Results
Testing the effect of total social welfare expenditure on social unrest
In Table 1 I present the first regression model.12 In this first model I analyse the
relationship between social welfare spending and social unrest. As I described in the theoretical part, this is the first –broad- step towards explaining the relationship. I am using the total social welfare spending variable as the independent variable and the
weighted conflict index –which is a weighted number of all the types social unrest- to test
the first hypothesis. As it appears, the results in Table 1 show that there is an effect in the expected negative direction, but it is weak and not significant. This means that (and this is important to keep in mind) for the period 1980-2009 the relationship between social welfare expenditure and social unrest does not stand. Therefore we reject H1. We do see that a big share of the effect is predicted by the control variables that I used. Particularly GDP growth and unemployment have a highly significant negative effect on social unrest.
Testing the effect of the social welfare branches
The next step is to look at the specifications within the welfare expenditure that I found most relevant to use. The results are also in Table 1 and it is clear that for three out of the four social welfare specifications -unemployment, old age and health- the effect on social unrest is weak and non-significant. The only effect that does show significance is that of housing expenditure. It is extremely strong and significant. This means that we reject hypotheses: H2a, 2b and 2c. Hypothesis H2d is accepted. A possible explanation
Testing the effect on Southern European countries
As I have thoroughly explained, I have the expectation that the relationship between social welfare cuts and social unrest plays out differently -more negatively- in Southern Europe. My claim is that this relationship is a more sensitive one in that region because Southern European regimes enjoy less legitimacy amongst their people, and the legitimacy that they do have derives mostly from what the regime does (or doesn’t do) with regard to social policies. In other words, citizens of these countries watch their government’s actions with more strictness (more as in: more than in Northern Europe) with regard to social policies. I formulated this expectation in hypothesis 3.
In order to test this claim I have made an interaction variable. This interaction variable needs some explanation which I am going to give here. For each of the four specifications of welfare expenditure (the same four that I have chosen earlier), I have made an interaction variable which is called ‘’Notseur’’. The variable ‘’Notseur’’ is a dummy variable, with 0 being a Southern European country and 1 being a Northern European country. This way, we are able to observe the effect of social welfare expenditure separately for Southern Europe and Northern Europe.13 In Table 2, only one
12 The model being used is a multilevel mixed-effects regression.
13 The model here is an estimate of all conflict, using a random intercept maximum likelihood estimator, with standard errors allowing for unstructured clustering.
of the specifications shows a significantly higher effect in Southern Europe than in Northern Europe, namely unemployment spending. I am going to use this outcome to explain how the interaction variable works. In Table 2, behind ‘’Unempl exp’’ we see that the effect is -337.191**. So, more spending on unemployment benefits leads to less social unrest, and this effect is significant at 5%. But the interaction variable underneath, ‘’Unempnotseur’’ shows us a positive relationship of 413.981***. So when you move from 0 (Southern European country) to 1 (not a Southern European country) the effect becomes positive and significant. Keep in mind that a positive relationship in this second row is exactly what we were expecting. Because it means that, when you’re not a
Southern European country, this effect of changes in social welfare on conflict
becomes significantly more positive (or less negative). So the result tells us that social welfare cuts have a significantly smaller impact in Northern European countries.
We see a similar, but less significant result for old age spending. For both health expenditure and housing the effect is in the expected direction –if you’re not a Southern European country the relationship turns out to be more positive- but not significant. This means that out of the four types of social welfare expenditure that we have tested (and of course the total social welfare expenditure variable) only the unemployment spending variable gives us a significantly smaller effect in Northern Europe. This brings us to the conclusion that cuts in social welfare expenditure in Southern Europe do not have a stronger negative effect on social unrest thus we reject H3; but with the side note that the Southern Europe argument is true for unemployment expenditure.
Model 1 Model 2 Model 3 Model 4 Model 5
Total social unrest Total social unrest Total social unrest Total social unrest Total social unrest
Total social exp -6.402 (16.289)
Unempl exp 10.151 (68.333)
Old age exp -20.281 (36.119)
Health exp -69.293 (46.148) Housing exp -607.844*** (194.955) gov_left -4.578*** (0.789) -4.568*** (0,796) -4.544*** (0,794) -4.620*** (0.782) -4.344*** (0.831) openc 4.530** (2.298) 4.523* (2.331) 4.298* (2.468) 4.600** (2.299) 3.361 (2.285) debt 6.362*** (2.026) 6.314*** (1.974) 6.687*** (2.148) 6.562*** (1.958) 5.038** (2.453) realgdpgr -30.525** (11.897) -30.405*** (11.669) -30.559*** (11.633) -36.152*** (12.239) -20.509 (12.632) netu 0.321*** (0.042) 0.326*** (0.043) 0.325*** (0.043) 0.321*** (0.042) 0.394*** (0.052) unemp -51.861*** (14.424) -57.449*** (17,832) -53.361*** (13.836) -57.343*** (13.692) -39.831*** (14.838) lpop -394.422** (180.045) -402.239** (182.773) -408.901** (182.41) -379.329** (180.735) -140.893 (143.483) year -18.676*** (5.908) -20.358*** (5.153) -18.394*** (5.905) -15.922*** (5.699) -18.962*** (5.445) Table 1
Significance levels: *** significant at 1%, ** significant at 5%, * significant at 10% (Standard errors in parentheses)
Table 2
Model 6 Model 7 Model 8 Model 9 Model 10
Total social unrest Total social unrest Total social unrest Total social unrest Total social unrest
Notseur -2,220.582*** (798.277) -1,928.658*** (657.972) -2,183.469*** (746.075) -1,599.533** (757.348) -790.606 (501.433)
Total social exp -23.574 (21.366)
Socexpnotseur 35.605 (22.489) Unempl exp -337.191** (138.178) Unempnotseur 413.981*** (137.478) Oldage exp -63.53 (42.503) Oldagenotseur 84.583* (46.406) Health exp -85.269 (67.821) Healthnotseur 34.393 (72.125) Housing exp -955.356 (620.584) Housingnotseur 454.529 (641.96) gov_left -4.386*** (0.796) -4.211*** (0.797) -4.414*** (0.794) -4.555*** (0.792) -4.300*** (0.83) openc 4.561** (2.273) 4.303* (2.296) 4.775* (2.438) 4.643** (2.285) 3.594 (2.287) debt 6.131*** (2.011) 5.092** (1.982) 6.398*** (2.134) 6.412*** (1.951) 5.292** (2.578) realgdpgr -27.229** (11.958) -32.017*** (11.579) -28.140** (11.673) -34.783*** (12.323) -19.849 (12.702) netu 0.339*** (0.043) 0.329*** (0.042) 0.351*** (0.044) 0.329*** (0.044) 0.418*** (0.055) unemp -56.910*** (14.564) -54.501*** (17.799) -53.705*** (13.755) -56.724*** (13.656) -41.421*** (14.879) lpop -482.820*** (171.967) -453.949*** (174.808) -512.041*** (175.807) -456.181*** (175.988) -228.37 (152.848) year -18.319*** (5.849) -17.436*** (5.146) -17.297*** (5.835) -15.706*** (5.682) -18.024*** (5.502)
Significance levels: *** significant at 1%, ** significant at 5%, * significant at 10% (Standard errors in parentheses)
Taking the economic crisis of 2008 into account
The results we have seen thus far, covered the period from 1980-2009 and it is safe to say that most of the expectations I expressed in my hypotheses turned out not to be true in the analyses. But as I have started this paper, one of the main starting points for this research was the economic crisis of 2008. Given the specific interest that I have expressed in the recent economic crisis, I merged data from EUROSTAT on social welfare expenditure from 2002 until 2011, with the social unrest data I already used, namely that of the CNTS. The reason is that I want to look at the most recent period available in datasets, because I want to see how the relationship works out in what has been called the worst crisis since the Great Depression. If even in this ‘’most likely’’ scenario the relationship between social welfare cuts and social unrest does not stand, it is safe to say that there is no such relationship. So since the recent economic crisis is the most likely scenario in which this relationship will occur, it is very important to look at it.
The regressions analysis you see underneath is the same as in Table 1, but done with more recent EUROSTAT data and with slightly different country coverage, since not all OECD members are included; only European Union countries.
Table 3
Model 11
Total social unrest Total social exp -46.576** (18.15) gov_left -3.317** (1.395) openc -2.468 (2.097) debt 0.777 (2.72) realgdpgr -14.228 (18.058) netu -0.071 (0.047) unemp 4.891 (22.578) year -20.324 (17.506) lpop 245.413*** (76.386)
Interestingly, we now see that the effect of social expenditure as a percentage of GDP has a significant negative effect (-46.576**) on social unrest. While in our first analysis this was not the case, H1 is accepted here. This tells us that social welfare cuts – most likely as a result of the economic crisis of 2008- indeed led to severe social unrest.
While I would have been glad to do so, I am not able to look at specifications of the welfare state as the independent variable in this analysis, because of the simple reason that the EUROSTAT data does not provide these figures. So I am moving on to the next important question that I posed: do social welfare cuts have the same effect in Southern Europe as in Northern Europe? By the same fashion as before, I created an interaction variable: social expenditure in countries that are not Southern European. So just like before, this variable tells us what the effect of changes in social welfare cuts is on social unrest, when we leave the Northern European countries out. And to be clear: just like in the analysis in Table 2, we expect this number to be positive and significant because only in that case we can say that the effect is significantly less negative in
Northern Europe than in Southern Europe (or if you turn it around: more negative in Southern Europe than in Northern Europe).
Table 4
Model 12
Total social unrest
notseurB -5907.494*** (1078.55)
Total social exp -269.974*** (44.159) Socexpnotseur 240.655*** (43.324) gov_left -3.979*** (1.254 openc -2.796* (1.578) debt 3.849 (2.5) realgdpgr -13.193 (16.924) netu -0.005 (0.034) unemp 16.273 (18.123 lpop 138.376** (56.721) year -0.2 (17.114)
This table shows us that social welfare spending has a significantly negative effect on social unrest, but this effect becomes positive and significant when moving from 0 (Southern European country) to 1 (not a Southern European country). This means that the effect of the social welfare cuts is significantly less severe in Northern Europe – or significantly more severe in Southern Europe. The fact that we get these results when adding the recent crisis years is interesting and insightful, since this period has not yet been subjected to an analysis of this kind. These results also indicate that, when using the time period 2002-2011, our conclusions change drastically. H3, which states that Cuts in social welfare spending have a stronger negative effect on social unrest in Southern Europe than in Northern Europe, was rejected in our first analysis but when using the EUROSTAT data the results are overwhelmingly significant.
Conclusion
Social unrest is a phenomenon with disruptive consequences for societies. The images of burning barricades in downtown Barcelona or at Syntagma square in Athens are still fresh in the collective memory of European citizens and politicians. Social unrest, in the severity that we have witnessed during the crisis, can pose threats to the stability of not only the domestic sphere but also to that of neighbouring countries -or an entire continent for that matter. In the situation of Europe, it added extra pressure to the discussion whether some Southern European countries should leave the Eurozone or not (a ‘’Grexit’’ almost became reality).
While the existing literature on social unrest is limited, it regained attention since the economic crisis of 2008. Both researchers and policymakers do not agree with each other on which economic policy is preferable, but it is clear that ‘’cutting our way to
growth’’ was the more dominant paradigm during this crisis. However, others have condemned the strict policies of mainly Northern European countries such as Germany, which kept the focus on keeping budget deficits down. According to those scholars, policy makers have neglected the social consequences of austerity measures. With this paper, I have attempted to contribute to the field, by looking into the relationship between social welfare spending and social unrest. While implied by various forms of media and politicians, extensive research on this relationship has never materialized. In addition to this, I focus on the effect of cuts in various specifications of the welfare state and I look into the relationship between Southern Europe and Northern Europe.
The findings are not completely irrefutable, since they do not all point in the same direction. For the period 1980-2009 the results are rather weak and ambiguous: there is no significant effect for the total social welfare expenditure variable and we only observe a significant effect for housing expenditure. Also, there seems to be no difference between the effect in Northern Europe and Southern Europe, only when we look at unemployment benefits the effect of cuts seems to be significantly stronger in Southern Europe. But the results become much more interesting –and convincing- when doing the same analysis with EUROSTAT data which reaches up until 2011. In that analysis, not only the relationship between total social welfare expenditure and social unrest is negative and significant, but also the ‘’Southern Europe claim’’ gets overwhelming support. By making an interaction variable, I looked at the effect of social welfare cuts in countries that are not Southern European. Since the effect was significant and positive, this should be interpreted as: cuts in Northern Europe have a significantly smaller impact. Scholars have not looked at this possibility yet, so this could be called the big
news of this paper.
So it is clear that the results of the analysis of the period 2002-2011 indicates that cuts social welfare spending do increase the risk of social unrest in a severe crisis like crisis that started off in 2008. But, it also tells us that there is a lot more to explore with regard to the level of cuts that lead to social unrest. Since this economic crisis, that lasted roughly five years, showed us more interesting results than the 30 years before (with regard to this particular relationship), this period will most likely become a rich source for scholars to do further research on social unrest. Seen from that perspective, this study can be seen as a first step towards understanding this mechanism in the exceptional crisis of 2008-2013.
Clearly, the findings of this article have implications for policy making. They go against the idea of ‘’cutting our way to growth’’ expressed by Reinhart & Rogoff (2010) and thus against the policy decisions made in light of the European economic crisis. Based especially on the analyses done with the EUROSTAT data, this study adds to the criticism that other scholars like Paul Krugman have expressed towards the claims of Alesina & Ardagna; that not only cuts are desirable for economic growth, but that cuts do not lead to any real punishment at the polls for governments pursuing cut-backs. Indeed, as the findings in this piece tell, we need to take into account the social consequences that social welfare cuts have for societies, because the relationship is present and it is fair to say that these consequences have been underestimated both by scholars and policy makers.
References
Alesina, AF., Ardagna, S. (2010). Large changes in fiscal policy: taxes versus spending. Cambridge: National Bureau of Economic Research
Alesina, AF., Carloni, D. & Lecce, G. (2010). The electoral consequences of large fiscal
adjustments. Chicago: University of Chicago Press
Andreotti, A., Garcia, S.M., Gomez, A., Hespanha, P., Kazepo, Y. & Mingione, E. (2010).
Does a Southern European Model Exist? London: Routledge
Basabe, N., Paez, D., Valencia, J., Gonzales, JL., Rimé, B. & Diener, E. Cognition &
Emotion. London: Routledge
Beck, N., Katz, J. & Tucker, R. (1998). Taking Time Seriously: Time-Series Cross-Section
Analysis with a Binary Dependent Variable. American Journal of Political Science
42(4): 1260-1288
Blau, J. (2000). The dynamics of social welfare policy. New York: Oxford University Press, Inc.
Bueno de Mesquita, B., Smith, A., Siverson, RM. & Morrow, J. (2003). The logic of
political survival. Cambridge, MA: MIT Press
Eichengreen, B. &O’Rourke, K. (2009). A tale of two depressions. Lexington, MA: Advisor Perspective
Engelen, E.(2014). Europese mythes: boekje voor burgers. Amsterdam: Uitgeverij Leesmagazijn
Ferrera, M. (2005). Welfare state reform in Southern Europe: Fighting poverty and
social exclusion in Italy, Spain, Portugal and Greece. London: Routledge
Gunther, R., Diamandouros, N & Puhle, H.J. (1995). The politics of Democratic
Consolidation: Southern Europe in Comparative perspective. Baltimore: JHU Press
Haggard, S., Lafay, JD. & Morrison, C. (1995). The political feasibility of adjustment in
developing countries. Washington D.C.: Development Centre of the Organisation for
Economic Co-Operation and Development
Jahoda, G. (1992). Crossroads between culture and mind. New York: Harvester/Wheatsheaf
Krugman, P. (2013). How the case for austerity has crumbled. New York: The New York Review of Books
Magone, JM. (2003). The politics of Southern Europe: integration into the European
Union. Westport: Praeger Publishers
Mingione, E. (2001). The Southern European welfare model and the fight against
Norton-Taylor, R. (2011). Why Do Revolutions such as Tunisia’s Come by Surprise? www.theguardian.com (search for: revolutions surprise Tunisia)
Paldam, M. (1993). The Socio-Political Reactions to Balance of Payments Adjustments in
LDCs: A study of Nine Cases from Latin America. Paris: School of Economics and
Management
Ponticelli J., & Voth, H.J. (2011). Austerity and Anarchy: Budget cuts and social unrest
in Europe, 1919-2009. London: CEPR (Centre for Economic Policy Research)
Reinhart, C. & Rogoff, K. (2010). Growth in a time of Debt . Pittsburgh: American Economic Association
Rhodes, M. (2009). Southern European Welfare States: Identity, problems and
prospects for reform. London: Routledge
Ruffert, M. (2011). The European Debt Crisis and European Union Law. United Kingdom: Kluwer Law International
Sprengers, M. & Tazelaar, F. (1984). Werkloosheid en sociaal isolement. De Sociologische Gids
Taydas, Z. & Peksen, D. (2012). Can states buy peace? Social welfare spending and civil
conflicts. Journal of Peace Research
Wallerstein, I. (2002). The eagle has crash landed. Foreign Policy (06/01/2002) (http://www.frankendesign.com/pdf/EagleCrashLanded.pdf)
Woo, J. (2003). Economic, political and institutional determinants of public deficits. Journal of Public Economics
