Support for Globalisation and the Welfare State (ARI)

Support for Globalisation and the Welfare State (ARI)

Theme: Trade protectionism and popular support for globalisation are controversial topics which raise many concerns. Based on the International Social Survey Programme (ISSP) questionnaire submitted to 44,170 people from 33 OECD countries, this paper shows that countries with the most extensive Social Security systems are those which most support free trade.

Summary: The hypothesis of compensation maintains that citizens demand that their governments spend enough on Social Security so as to protect them from the economic volatility resulting from globalisation. Indeed, some authors assert that popular support for globalisation hinges on maintaining and even extending the Welfare State. This is because if citizens believe that economic integration jeopardises social protection systems, they will oppose globalisation. Here we show that there is a positive correlation between Social Security expenditure and support for free trade. Other factors which also affect opinions on free trade are: education, a country’s human capital, age, gender, occupation and place of residence. Once all these effects are factored in, the correlation between social spending and support for free trade in each country is even stronger. The results suggest that opposition to globalisation may be overcome by maintaining the Welfare State. Citizens seem willing to see globalisation intensify if, in exchange, they are compensated with protection from the increased economic volatility –especially as regards job insecurity– which globalisation involves.

Analysis

Introduction
In ‘Who Fears Globalisation in the EU?’ (ARI nr 100/2006) we showed that insecurity in the labour market is at the heart of the reticence generated by globalisation. Other social factors (such as gender, area of residence or idiosyncratic factors) or political aspects play a very limited role in determining people’s attitudes towards this economic process. As a result, the conclusion was that a social welfare system which compensates workers for the job insecurity generated by the concerns raised by this economic process.

The aim of this analysis is to verify whether, in fact, countries in which the Social Security system is more extensive do register higher levels of support for globalisation. For this purpose, we have used the International Social Survey Programme (ISSP) questionnaire on national identity put to 44,170 people from 33 countries in 2003 (ISSP 2003, ‘National Identity II’). The people surveyed were asked whether ‘[their country] should limit the import of foreign products in order to protect the national economy’. The opinions of those surveyed range from those who strongly agreed (the most protectionist position) to those who strongly disagreed (the opinion most in favour of free trade). The responses to this question on limiting imports will be crossed with Social Security expenditure in the country in which the person interviewed resides, so as to observe whether the existence of the Welfare State increases support for free trade. There is information concerning Social Security expenditure for 22 of the 33 countries included in the ISSP questionnaire, so that the sample for analysing the correlation between the Welfare State and support for free trade includes 27,896 individuals instead of the 44,170 originally surveyed.

The Welfare State and Support for Free Trade
Rodrik (1998) shows that the countries that are most exposed to international trade face greater levels of volatility in income and consumption.[1] In other words, opening a country’s economy increases the risks posed to its citizens. However, this author holds that citizens would be willing to allow their governments to liberalise trade if the increase in economic volatility generated as a result were to be offset by higher spending in Social Security. In fact, Scheve and Slaughter (2006) maintain that popular support for globalisation is linked to the maintenance of the Welfare State.[2] This is why, if globalisation is to prevail, it must be compatible with social protection systems. Therefore, this hypothesis of compensation predicts that support for free trade and, in general, for globalisation, should be higher in countries whose social protection systems are the most generous.

Table 1. Results of International Social Survey Programme Questionnaire in OECD Countries

CountrySupport for free tradeSocial Security expenditure (% GDP):average for 2000-03Pension expenditure (% GDP):average for 2000-01Unemployment expenditure (% GDP):average for 2000-01
Australia2.2517.665.030.98
Austria2.4425.6410.630.75
Canada2.6417.144.770.75
Czech Republic2.6620.726.770.26
Denmark2.3026.658.292.99
Finland3.0021.777.782.05
France2.6127.9210.621.63
Germany2.8327.0611.561.23
Hungary2.2221.477.890.42
Ireland2.5614.872.650.75
Japan2.8617.047.030.50
South Korea2.605.391.310.12
New Zealand2.4718.494.861.26
Norway2.9923.776.650.43
Poland2.2322.388.300.89
Portugal2.4621.697.700.91
Slovakia2.1517.786.700.69
Spain2.4920.278.381.33
Sweden3.1229.949.221.14
Switzerland3.0519.1811.610.50
United Kingdom2.4019.898.170.29
United States2.3515.495.230.27
Simple mean2.5820.567.330.92

Table 1 shows support for free trade in 2003 and public spending on Social Security in the 2000-03 period in the 22 OECD countries for which figures are available. Support for free trade was calculated based on responses to the question of whether ‘[their country] should limit the import of foreign products in order to protect the national economy’, included in the 2003 ISSP National Identity II survey. Specifically, the responses were coded as follows:

            Agree strongly                                      1

            Agree                                                   2

            Neither agree nor disagree                     3

            Disagree                                               4

            Disagree strongly                                  5

A score of 1 indicates a protectionist position, while a score of 5 reflects support for free trade. Figures referring to Social Security expenditure were obtained from the OECD Social Expenditure Database (1980-2003) and are shown as a percentage of GDP. The latest available figures (2000-2001) are also shown for the two most significant items of Social Security spending: pensions and unemployment benefits. The figures show that Scandinavia (Sweden, Finland and Norway) and Switzerland are most in favour of free trade. On the other hand, public opinion in countries that have only recently joined the European Union (Slovakia, Hungary and Poland) is more favourable to protectionism. In Spain the bias is slightly in favour of limiting imports, indeed slightly above the average for the OECD countries analysed (2.49, vs. an average of 2.58). Countries where there is greater support for free trade (Scandinavia) are those which spend the largest percentage of GDP on Social Security. In fact, the correlation between free trade and Social Security spending is positive although not significant (+0.21). In particular, the approach to international trade appears to be more closely linked to pensions, although it is not significant either (+0.25), than to unemployment benefits, with which there is no correlation (0.00).

Figure 1 shows the relation between support for free trade and the three measures used to gauge the size of the Welfare State: (a) total Social Security expenditure and its two largest items (b) pensions and (c) unemployment benefits. The regression line shows an increasing correlation between total Social Security expenditure and pensions expenditure and support for opening a country’s economy. In other words, in countries which spend most on Social Security, especially on pensions, there is less popular support for limiting imports. In any event, there are many countries outside the confidence interval. Austria, for example, spends a high percentage of GDP on Social Security (25.64%) and yet support for free trade among the Austrian public is less than average (2.44, vs. 2.58). South Korea, on the other hand, assigns only a limited percentage of GDP to Social Security (5.39%), but its support for free trade is somewhat higher than average (2.60). Furthermore, unemployment benefits show no kind of correlation with preferences in regard to free trade. In short, the graphs suggest that there is a positive link between the Welfare State and support for free trade, but that the relationship is not perfect. These findings suggest that other factors also impact on people’s attitudes to free trade in developed countries. We shall now examine whether the link between the Welfare State and support for free trade continues once these other relevant factors have been taken into account.

Other Factors Shaping Support for Free Trade
Economic literature suggests that, as well as the size of the Welfare State, there are other factors that determine a person’s attitude towards free trade (see Scheve and Slaughter, 2006). Individual preferences about trade policies depend critically on the mobility of workers between economic sectors. When workers cannot easily change sector, for instance in the context of the Ricardo-Viner model, their biases regarding economic integration policy will depend on the sector in which they are employed. If, on the other hand, workers are able to move freely among sectors, as occurs in the Heckscher-Ohlin model, then workers’ skills become the key variable in their support for globalisation. In this model, economic integration modifies the relative prices of factors in accordance with the Stolper-Samuelson theorem: it increases the relative return of abundant production factors while it reduces it for factors that are relatively scarce. Consequently, in countries where training is widely available, highly-skilled workers will support free trade, whereas low-skilled workers will oppose it. In contrast, in countries where training is relatively scarce, highly skilled workers will be against economic integration, whereas low-skilled workers will support it. This is why the core variable in the explanation of a person’s approach to economic integration is education, which is measured by the variable covering the subjects’ years at school. The subjects’ occupations are also taken into account in controlling the effects of non-formal education on their attitudes towards economic integration. Specifically, the one-digit ISCO-1988 occupation classification standard is used.

Socio-demographic characteristics can also affect preferences in regard to trade. Accordingly, there is empirical evidence of a U-shaped relationship between age and support for economic integration. In other words, people of working age are the most reluctant in regard to globalisation, whereas young and elderly people are less likely to be against it. Some authors also suggest that women are more protectionist than men, which might be because the employment market is segmented in such a way that women are more at risk from economic liberalisation. Some research has also found that support for globalisation in the most rural areas is lower than in other regions. Table 2 describes the characteristics of the individuals interviewed in the ISSP 2003 questionnaire (education, age, gender, occupation and place of residence) which we will use as control variables in the study of the relationship between the Welfare State and support for free trade.

Table 2 Individual characteristics, as well as Welfare State, which impact on the attitudes of the persons surveyed to wards the free trade

Individual characteristicDefinition of the variableRange
EducationYears at school1-26
Education in interaction with Welfare StateThe individual’s years at school in interaction with Social Security expenditure (% GDP) in the country where he/she resides5.4-778.2
Education in interaction with pensions systemThe individual’s years at school in interaction with spending on pensions (% GDP) in the country where he/she resides1.3-301.6
Education in interaction with unemployment benefitsThe individual’s years at school in interaction with spending on unemployment benefits (% GDP) in the country where he/she resides0.1-77.7
Education in countries with abundant human capitalThe individual’s years at school in interaction with average years at school of individuals in the same country8.3-360.6
OccupationNine dummy variables taking value 1 if the occupation of the persons surveyed are: – Members of armed forces- Company directors- Scientific professionals and intellectuals- Mid-level technical experts and professionals- Office workers- Services sector workers and salespeople- Farmers and skilled agricultural and fishing workers- Workers skilled in mechanical arts and crafts- Machine operators and assembly-workers- Unskilled workers0-1
AgeYears15-98
GenderDummy variable taking value 1 if the person surveyed is a woman and 0 if he is a man0-1
Place of residenceFive dummy variables taking value 1 if the person surveyed lives in: – Urban area, major city- Suburban area, outskirts of major city- Small city- Small town- Rural location0-1

Results of the Estimate on Factors Determining Support for Free Trade
Table 3 shows the results of the estimate of factors determining the support for free trade using the responses of the 20,466 individuals from the 2003 ISSP National Identity II survey for whom information was available in regard to all the variables: education, social expenditure in their country, gender, age, occupation and place of residence. An ordered probit model was used since this method of estimating takes into account that the distance between agreeing strongly with free trade and agreeing somewhat cannot be the same as between agreeing somewhat and disagreeing somewhat. Model 1 includes the person’s level of education on its own as well as in interaction with the average education in the country where that person resides. The findings reveal that education has a significant negative impact on the likelihood of supporting free trade. However, an individual’s education in interaction with the average years at school in that person’s country of residence has a significant positive impact. In other words, in countries where the average number of school years is high, education has a positive impact on people’s opinion of free trade, whereas in countries where average education is low, protectionist positions are more common. Specifically, in countries averaging six or more years of schooling, education is positively linked to economic integration, while in countries below this threshold the correlation is negative. The result is in line with the predictions of the Heckscher-Ohlin model: in countries with abundant human capital free trade will benefit the most highly-skilled, while in countries with scant human capital economic integration helps the least skilled. The gender dummy variable is negative and significant, confirming that women are more protectionist than men. Age, on the other hand, does not seem to affect individuals’ preferences in regard to trade. As regards occupation, there is empirical evidence that office workers, services workers, mechanical craftsmen, machine operators, unskilled workers and, above all, farmers, are more opposed to free trade than other workers. Model 2 introduces the individual’s education in interaction with the average number of school years of the region in which that person resides. The results confirm the predictions of the Heckscher-Ohlin model: in regions averaging five or more school years education increases support for free trade, and elsewhere it reduces it. The other variables show very similar coefficients to model 1, indicating that the results are robust to changes in the estimate’s specification.

Table 3. Factors determining opinion on free trade (ordered probit model)

Dependent variable:Should your country limit the import of foreign products in order to protect the national economy?Agree strongly                                     1Agree                                                       2Neither agree nor disagree                               3Disagree                                                4Disagree strongly                                 5
Education-0.037
 (2.36)**
Average education in the country*0.006
 (4.88)***
Gender-0.231
 (14.25)***
Age-0.002
 (0.66)
Age squared-0.000
 (0.90)
Members of armed forces-0.128
 (0.90)
Company managers0.047
 (0.51)
Scientists and intellectuals0.050
 (0.55)
Mid-level professionals-0.108
 (1.19)
Office workers-0.157
 (1.71)*
Services sector workers-0.234
 (2.55)**
Farmers-0.622
 (6.31)***
Workers skilled in mechanical crafts-0.317
 (3.46)***
Installation operators-0.316
 (3.38)***
Unskilled workers-0.271
 (2.90)***
Residence in a major city0.034
 (0.29)
Residence in outskirts of a major city-0.057
 (0.49)
Residence in a small city-0.091
 (0.79)
Residence in a small town-0.163
 (1.40)
Residence in rural location-0.263
 (2.18)**
Observations20466

(*), (**), (***) 10%, 5% and 1% significant, respectively.

The two estimates in Table 3 include country dummy variables. The Austria dummy variable, for example, is a variable taking value 1 for that country and 0 for all others. The dummy variables capture idiosyncratic cultural or historic factors for each country. These are aspects shared by the individuals in a country and affecting their trade preferences. In short, these dummy variables reflect the support for free trade in each country once the individual factors, such as education, gender, age, occupation or place of residence have been discounted. The Australia dummy variable was omitted to avoid perfect multicollinearity, so that the idiosyncratic country effects must be interpreted in relation to Australia. Table 4 shows the findings for the 21 country dummy variables introduced in models 1 and 2 of Table 3. Accordingly, the coefficient of 0.62 associated with Spain indicates that, once the effects of the individual characteristics of the Spanish interviewees have been discounted, residing in Spain increases the likelihood of being in favour of free trade in relation to residing in Australia. It is observed that most of the country dummy variables are highly significant, confirming that, in addition to the individual benefits, nationality also affects people’s attitudes towards economic integration. Denmark, Switzerland, Sweden, Finland and Germany are, in that order, the countries with the highest probability of supporting free trade. Slovakia, the US, Australia and Poland are, in contrast, those which most favour protectionist positions. The ranking is also similar regardless of whether the dummy variables estimated in model 1 or those of model 2 are used. Spain is in the intermediate zone, although it is closer to supporting free trade than to protectionism. In other words, the first impression of Spain that protectionist opinion prevails was due to the fact that the effects of the individual characteristics on trade preferences had not yet been factored in.

Table 4. Support for free trade by country once individual characteristics have been factored in

CountrySupport for free trade
Australia0.00
Austria0.38
Canada0.32
Czech Republic0.48
Denmark1.07
Finland0.88
France0.27
Germany0.88
Hungary0.29
Ireland0.40
Japan0.81
South Korea0.40
New Zealand0.26
Norway0.69
Poland0.16
Portugal0.70
Slovakia-0.15
Spain0.62
Sweden0.98
Switzerland1.02
United Kingdom0.35
United States-0.10

Figure 2 shows the relationship between country dummy variables and the various social expenditures. Because the dummy variables reflect a country’s support for free trade based on idiosyncratic factors, it is reasonable to think that one of these aspects is the varying size of the social protection system. In fact, our initial hypothesis is that social expenditure increases support for free trade because it protects individuals from the volatility of international trade. Indeed, Figure 2 shows that the relationship between social spending and support for free trade (measured via the country dummy variables in Table 4) is now clearer than in Figure 1. In fact, the correlation between support for economic integration and social expenditure, pensions and unemployment is now 0.41, 0.43 and 0.41, respectively, and always 5% significant. It is also interesting to observe that most countries are now in the confidence interval of the regression line in both social spending and pensions and unemployment. The relationship between the Welfare State and support for free trade is not only robust once other factors such as individual characteristics have been taken into account, but it is also even clearer.

Conclusions: The compensation hypothesis maintains that citizens demand that their governments spend enough on Social Security so as to protect them from the economic volatility resulting from globalisation. Indeed, some authors suggest that popular support for globalisation hinges on maintaining or even extending the Welfare State, since if citizens observe that economic integration jeopardises social protection systems they will oppose globalisation. In this paper we present evidence that there is a positive correlation between Social Security spending and support for free trade in 22 OECD countries included in the 2003 ISSP National Identity II survey.

This relationship is robust when taking into account other factors that also affect people’s opinion on free trade, such as: education, the human capital in a country or region, age, gender, occupation and place of residence. Once all these effects are factored in, the correlation between social expenditure and support for free trade in each country is even more robust. The findings suggest that resistance to globalisation can be overcome by maintaining the Welfare State. People seem willing to intensify the globalisation process if, in exchange, they receive compensation to protect them from the greater levels of economic volatility, particularly due to employment insecurity, which it entails.

Ismael Sanz
Professor of applied economics at the Universidad Complutense de Madrid

Ferrán Martínez
Doctor in sociology from the Universidad Complutense de Madrid and member of the Instituto Juan March


[1] D. Rodrik, ‘Why Do More Open Economies have Bigger Governments?’, Journal of Political Economy, nr 106 (5), 1998, p. 997-1032.

[2] K.F. Scheve and M.J. Slaughter, ‘Public Opinion, International Integration, and the Welfare State’, in P. Bardhan, S. Bowles and M. Wallerstein (eds.), Globalization and Egalitarian Redistribution, Princeton University Press, Princeton, 2006, p. 217-260.

Ismael Sanz

Written by Ismael Sanz

Ferrán Martínez i Coma

Written by Ferrán Martínez i Coma