Alex Williams- Q-step essay. Does consensus democracy improve economic outcomes?

Transcription

1 Does consensus democracy improve economic outcomes? Lijphart (2012, p295) argues that consensus democracies have the better record when compared to majoritarian democracies (Lijphart, 2012). He argues that they are superior because of their kinder and gentler (Lijphart, 2012, p294) qualities, and that they outperform majoritarian systems economically on 14 of 15 variables (Lijphart, 2012). I will test the hypothesis that consensus democracy improves economic outcomes against my own theory that the duration of democratic governance improves economic performance. I will assess the impact each of these factors has on GDP per capita, CPI inflation and LFS unemployment. I conclude that more consensual democracy explains lower unemployment, whilst the duration of continuous democratic governance explains lower inflation and higher GDP per capita. Consequently, I will reject the claim that consensus democracy improves economic outcomes. i. Definitions Lijphart (2012), distinguishes between consensus and majoritarian democracies on the basis of two, five variable dimensions which are each operationalised through indexes to give an overarching index to describe countries on two linear dimensions. His claims regarding economic outcomes are regressed against his executive-parties dimension. As I am attempting to replicate his analysis as closely as possible, I shall only consider this dimension. Executive-parties consensus democracy is characterised by multi-party coalitions in the executive, a balanced share of power between the executive and the legislature, a multi-party system, a proportional representation electoral system and a compromising, co-ordinated interest group system (Lijphart, 2012). A fully majoritarian system on this dimension will have, a single party executive, executive dominance over the legislature, a two party system, a majoritarian, disproportional electoral system and competitive non-cooperative interest groups (Lijphart, 2012). ii. Theoretical Argument One argument in favour of Lijphart s (2012) hypothesis is that their position in policy space is likely to be more stable. This is because consensus democracies, tend to have more veto players (Tsebelis, 2002), as they are characterised by power sharing and a desire for super-majority governance. The consensual end of each of the variables on the executive-parties dimension is symptomatic of additional veto players, so the more strongly consensual democracy is, the more veto players can impact policy decisions. As Tsebelis (2002) illustrates, each additional player can only reduce the winset, so with an executive body like the Swiss Federal council, which is made of seven people representing four parties, radical policy change is highly unlikely. Wilson (1975) claims that this allows consensus democracies to manage their economy more effectively. Firstly, he argues that as major policy deviation is unlikely, long-term economic strategies can form, and a steady hand will guide macro-economic policy decisions. In a majoritarian system policy continuity is far less likely, the competitive nature of the disproportional two-party system means that the second party opposes rather than works with the governing party, and the governing party often works to keep office rather than to use power wisely, inhibiting the formation of a cohesive long-term strategy. In the 1950s, majoritarian Britain experienced a period of stop-go economics where chancellors cut interest rates immediately before elections in order to increase consumption. This gained electoral support, but led to unsustainable borrowing and an economic strategy that revolved around winning elections. Secondly, majoritarian governments are almost unable to take a necessary but unpopular economic decision during the campaign. Take devaluation for instance, despite helping to reduce unemployment and increase exports, it is regarded negatively by the public. Hence, a balance of payments crisis could not be eased in a majoritarian system as the government would be held accountable at the election. In a consensual system, like Switzerland, blame is more evenly

2 distributed so no one party risks being held accountable. It seems that having more veto players and less direct accountability actually makes consensus democracies abler to form long term economic plans. So Lijphart s (2012) hypothesises is that consensus systems will have lower unemployment, lower inflation and higher growth rates. However, macroeconomic success may originate in states with a longer duration of democratic government which have more experienced and embedded institutions. These systems give the executive more past experience to draw on and also tend to see parties strategically converging on the centre ground. For instance, Butskellism dominated British economic policy in the post-war period as both parties endorsed similar economic policies. This means that each were close in policy space and reduced the amount of deviation a change in governing party would bring. Thus, despite majoritarian government, British macroeconomic policy was also characterised by stability. Whilst Thatcher s government represents a radical shift away from the status quo, the move right by New Labour, led to macroeconomic continuity under Blair s government. Thus, whilst Lijphart (2012) claims that consensus democracies perform better on account of their stability, I argue that this stability comes from embedded institutions that form based on the length of time a regime has been democratic. However, there are several major caveats to my argument. Firstly, globalisation, which works against both my and Lijphart s analysis, has led to greater macroeconomic interdependence between states. Thus, domestic government policy is not the sole mechanism affecting macroeconomic performance. Secondly, my analysis seems vulnerable to the argument that newer democracies can draw on the historical experiences of older democracies and enact policy accordingly, so there is no inherent advantage for older democracies. However, the political culture and sociological structure of each state is different. One could not argue that linguistically divided Belgium could be governed in the same manner as Britain, or learn massively from the history of Westminster democracy. Thus, whilst this argument may have some gravity, it seems to imply that after twenty years each democracy would be performing similarly and perusing identical policies. As this is not that case my argument holds some validity. iii. Empirical Evidence My empirical analysis uses the same thirty-six democracies as Lijphart s (2012). This allows me to test his hypothesis against mine using the same countries over the same period. Whilst Lijphart, excludes the smallest five countries from his analysis of economic factors, I have included them. This is because every country is vulnerable to external shocks, as the 2008 financial crisis showed, and to maintain as much diversity as possible. To operationalise the duration of continuous democratic rule, I have used a proxy measure based on the year Lijphart (2012, p49) claimed a country became democratic for each country that was not democratic before In all other cases I have used either 1945 or the year in which women became fully enfranchised, unless the country was not independent at that point, as the point at which the regime became democratic. So I have claimed that Finland has been democratic since 1918 when it gained independence from Russia. Likewise, the Republic of Ireland became democratic in I then took the number of years before 2010 as my measure of the duration of continuous democratic governance. Whilst imperfect this methodology will give an indication of whether this is an explanatory variable. Lijphart s (2012) study, controls for both the logarithm of population size and the HDI when assessing the impact of his executive-parties dimension on economic outcomes. I control for the

3 logarithm of population size to maintain consistency, but am unable to control for HDI as it is strongly correlated with the age of democracy, the relationship is significant at the 0.1% level. Inflation Methodology Following Lijphart s (2012) example I have excluded Uruguay, Costa Rica and Jamaica from my analysis of inflation as each of these countries experienced hyperinflation in this period, and are statistical outliers. Botswana is also a statistical outlier, but as it did not experience hyper-inflation and is included by Lijphart in his longer period, it has been included. I have compiled data from UN data s data base on average annual CPI inflation and added it to Lijphart s (2012) dataset. This data is given as an index where 2005 = 100, so I have taken the index number for 2010 and divided it by the 1991 value to produce a number describing how much 1991 prices must be multiplied by to reach 2010 prices. Empirical results Table 1: Regression between the duration of continuous democratic government up to 2010 and CPI inflation between 1991 and 2010 as measured by the UN. Variables Executive-parties dimension Duration of democratic governance Logarithm of population size Intercept N 33 Adjusted R Level of statistical significance: p<0.001***, p<0.01**, p<0.05* Model 1 Model 2 Model 3 Estimate (S.E) ** (0.129) *** (0.309) (0.127) ** (0.005) 2.761*** (0.329) (0.129) ** (0.047) (0.135) 2.845*** (0.639) The duration of democratic governance chart shows a much clearer relationship than the executive parties-dimension chart in which no relationship is clear. My analysis demonstrates that Lijphart s (2012) claim that the relationship between the executive-parties dimension and CPI inflation is statistically significant at the 5% level, only holds as long as we do not control for the duration of democratic governance. Once we do we no longer have evidence to reject the null hypothesis that consensus democracy has no effect on inflation. However, the relationship between the duration of democratic governance

4 and CPI inflation in this period is statistically significant at the one percent level even if we control for the logarithm of population size. If we multiply the regression coefficient by the standard deviation, we find that a country that has been democratic for an additional standard deviation would have seen inflation increase by a multiple of less than a similar country that had been democratic for one less standard deviation. Hence, we have a regression equation Y= X. This relationship is significant at the 1% 10 level so is strong enough for us to accept the hypothesis that the longer a regime is democratic for; the lower inflation it has. Unemployment Methodology In order to complete Lijphart s dataset and give a full picture of unemployment across each of the thirty-six democracies, I have taken labour force survey unemployment rate data from the ILO between 1991 and My variable takes the mean of the available data in each of the thirty-six democracies.

5 Empirical results Table 2: Regression between the duration of continuous democratic government up to 2010 and mean unemployment as measured by the ILO s Labour Force Survey between 1991 and Model 4 Model 5 Model 6 Variables Executive-parties dimension Duration of democratic governance Logarithm of population size Intercept Estimate (S.E) **(0.572) 8.140*** (1.549) N 36 Adjusted R Level of statistical significance: p<0.001***, p<0.01**, p<0.05* * (0.596) (0.022) *** (1.518) * (0.593) (0.022) (0.643) *** (2.934) Unlike my inflation data, these results suggest that there is a relationship between the executiveparties dimension and an economic variable, when controlling for both the duration of continuous democratic governance and the logarithm of 2009 population size. This relationship is statistically significant at the 5% level, with the regression equation y= x which is enough for us to reject the null hypothesis that there is no relationship between the executive-parties measure of consensus democracy and unemployment. Hence, we can conclude that there is such a relationship and our regression coefficient of 1.526, and the standard deviation is one, we should expect unemployment to be 1.53% lower in a country that became one standard deviation more democratic. However, I cannot reject the null hypothesis that there is no relationship between the length of time a regime has been democratic and unemployment, especially as several newer democracies, like Korea, India and Botswana differ significantly from expectation as shown above. Similarly, with the executive-parties index Botswana s inflation is unexpectedly high and Korea s unexpectedly low, but the positive correlation is more evident. GDP per capita Methodology As economic growth data is biased towards less developed rapidly industrialising economies, as opposed to those already close to potential output, it is not a good variable to make long-run economic comparisons from. Consequently, I have considered GDP per capita in 2010, which will give a long term impression of how a regime has performed. I have sourced my data from UN stat. In this regression I have controlled for a binary dummy variable where European countries are assigned a one, which does not directly affect either of the outcomes of interest. Empirical Results Table 3: Regression between the duration of continuous democratic government up to 2010 and GDP per capita in 2010 Model 7 Model 8 Model 9 Variables Executive-parties dimension Duration of democratic governance Logarithm of population size Estimate (S.E) 11226*** (3440) 6132* (2862) 511*** (105) 6223* (2907) 512*** (107) (3152) Intercept (3390) 2976 (7291) 8045 (14383) N

6 Adjusted R Variables Executive-parties dimension Duration of democratic governance Logarithm of population size Europe Intercept Model 10 Estimate (S.E) 3235 (2999) 459*** (102) -691 (2959) 13904* (5883) 2194 (13677) N 36 Adjusted R Level of statistical significance: p<0.001***, p<0.01**, p<0.05* Whilst the relationship between the executiveparties dimension and GDP per capita is statistically significant at the 0.1% in absence of control, the duration of democratic governance is again the explanatory variable. However, once we control for the duration of democratic governance, the logarithm of population size and our dummy Europe variable, there is no relationship between the executive-parties index and GDP per capita. The relationship between the duration of democratic governance and GDP per capita is statistically significant at the 0.1% level when we control for the executive-parties index, the logarithm of 2009 population size and whether a country is in Europe. This result allows me to reject the null hypothesis that there is no relationship between the duration continuous democratic governance and the level of GDP per capita. The consequent regression equation is y= x, thus if the mean aged democracy, had been democratic for an extra standard deviation, its GDP per capita would increase from $ to $ Hence we can accept the hypothesis that the longer a regime is continuously democratic for the higher its GDP per capita will be. Although notably, Luxembourg, Switzerland and Norway have far higher GDP per capita, whilst New Zealand s is far lower than we would expect, as is shown above. iv. Conclusion My empirical analysis has allowed me to conclude that the longer a regime has been continuously democratic, the higher GDP per capita and the lower inflation it will have. Although this variable does not adequately explain unemployment rates. Here, Lijphart s (2012) executive-parties dimension succeeds and allows us to conclude that consensus democracies have lower levels of unemployment. Overall this evidence suggests that whilst consensus democracy explains lower levels of unemployment, it cannot explain levels of GDP per capita or inflation as well as the duration of continuous democratic governance. Thus we cannot conclude that consensus democracies improve economic outcomes in general. So we must reject Lijphart s (2012) overarching claim that consensus democracy is the best form of democracy to improve economic outcomes. Word count (excluding title, bibliography, appendix and tables as instructed) 2200

7 Bibliography Lijphart, A., Patterns of Democracy: Government forms and performance in thirty-six democracies, New Haven, 2012 Tsebelis, G., Veto Players How Political Institutions Work, Princeton Unviersity Press, Wilson, T., The Economic Costs of the Adversary System, printed in Finer, S.E., Adversary Politics and Electoral Reform, 1975 Data sourced from: UN Stat: unstat.un.org The ILO, via Appendix 1 And the Guardian: Villani, L., Hilaire, E., and Provost, C., Duration of democratic governance CPI inflation GDP per capita in 2010 Mean ILO LFS Country unemployment ARG AUS AUT BAH BAR BEL BOT CAN CR DEN FIN FRA GER GRE ICE IND IRE ISR ITA JAM JPN KOR

8 LUX MAL MAU NET NOR NZ POR SPA SWE SWI TRI UK URU US Appendix 2 Here is a selection of the relevant R code that I used in my statistical analysis, not all of these results are shown in the actual essay, but I thought it would be useful to show my workings. data200<-read.csv("q-step data.csv") data200 names(data200) dummyexecparties <- data200$x.executive.parties execpart <- dummyexecparties*-1 plot(execpart, data200$x.my_cpi_inflation_ , xlab = "Executive-party dimension", ylab = "Multiple of CPI inflation increase ") identify(data200$x.my_cpi_inflation_ , labels = data200$x.country., cex=0.7, pos=3) boxplot(data200$x.my_cpi_inflation_ , ylab = "Multiple of CPI inflation increase ") boxplot(data200$x.my_cpi_inflation_ , ylab = "Multiple of CPI inflation increase ") text(data200$x.my_cpi_inflation_ , labels = data200$x.country., cex=0.7, pos=3) y <- data200$x.my_cpi_inflation_ boxplot(y, ylab = "Multiple of CPI inflation increase ") identify(rep(1, length(y)), y, labels = data200$x.country.) data223<-read.csv("year.csv") data223 names(data223) suffrage<-data223$x.universal.suffrage.year. plot(suffrage, data223$x.my_cpi_inflation_ , xlab = "Suffrage", ylab = "GDP per capita increase ") text(suffrage, data223$x.my_cpi_inflation_ , labels = data223$x.country., cex=0.7, pos=3) model240 <- lm(data223$x.my_cpi_inflation_ ~ suffrage) summary(model240) abline(lm(data223$x.my_cpi_inflation_ ~ suffrage)) summary(suffrage)

9 sd(data223$x.my_cpi_inflation_ ) mean(data223$x.my_cpi_inflation_1991.) data223$x.my_cpi_inflation_ ) data224<-read.csv("year2.csv") data224 names(data224) suffrage2<-data224$x.universal.suffrage.year. plot(suffrage2, data224$x.my_cpi_inflation_ , xlab = "Suffrage", ylab = "GDP per capita increase ") text(suffrage2, data224$x.my_cpi_inflation_ , labels = data223$x.country., cex=0.7, pos=3) model241 <- lm(data224$x.my_cpi_inflation_ ~ suffrage2) summary(model241) abline(lm(data224$x.my_cpi_inflation_ ~ suffrage2)) model242 <- lm(data224$x.my_cpi_inflation_ ~ suffrage2+data224$x.executive.parties ) summary(model242) model243 <- lm(data224$x.my_cpi_inflation_ ~ data224$x.executive.parties suffrage2) summary(model243) #Statistically significant #relationship between age of democracy and CPI at the 0.01% level when controlling #for the exec-parties index. data225<-read.csv("year3.csv") data225 names(data225) suffrage3<-data225$x.universal.suffrage.year. plot(suffrage3, data225$x.my_cpi_inflation_ , xlab = "Duration of continuous democratic government up to 2010 ", ylab = "Multiple of CPI inflation increase ") identify(suffrage3, data225$x.my_cpi_inflation_ , labels = data225$x.country., cex=0.7, pos=3) model244 <- lm(data225$x.my_cpi_inflation_ ~ data225$x.executive.parties ) summary(model244) abline(lm(data225$x.my_cpi_inflation_ ~ suffrage3)) model245 <- lm(data225$x.my_cpi_inflation_ ~ suffrage3) summary(model245) model246 <- lm(data225$x.my_cpi_inflation_ ~ suffrage3+data225$x.executive.parties ) summary(model246) data223<-read.csv("year.csv") data223 names(data223) suffrage<-data223$x.universal.suffrage.year. plot(suffrage, data223$x.mean.ilo.labour.force.survey.unemployment., xlab = "Duration of continuous democratic government up to 2010", ylab = "Mean unemployment rate between 1991 and 2010") identify(suffrage, data223$x.mean.ilo.labour.force.survey.unemployment., labels = data223$x.country., cex=0.7, pos=3)

10 model247 <- lm(data223$x.mean.ilo.labour.force.survey.unemployment. ~ data223$x.executive.parties ) summary(model247) abline(lm(data223$x.mean.ilo.labour.force.survey.unemployment. ~ suffrage)) summary(suffrage) model248 <- lm(data223$x.mean.ilo.labour.force.survey.unemployment. ~ suffrage+data223$x.executive.parties data223$x.logarithm.of.2009.population.) summary(model248) model1000 <- lm(data223$x.hdi.2010.~suffrage) summary(model1000) plot(suffrage, data223$x.hdi.2010.) model271 <- lm(data223$x.mean.ilo.labour.force.survey.unemployment. ~ suffrage+data223$x.executive.parties ) summary(model271) model272 <- lm(data223$x.mean.ilo.labour.force.survey.unemployment. ~ suffrage+data223$x.executive.parties data223$x.logarithm.of.2009.population.) summary(model272) model250 <- lm(data223$x.executive.parties ~ suffrage + data223$europe) summary(model250) data223<-read.csv("year.csv") data223 names(data223) suffrage<-data223$x.universal.suffrage.year. cap<-data223$x.tiger. plot(suffrage, cap, xlab = "Duration of continuous democratic government up to 2010 ", ylab = "GDP per capita in 2010") identify(suffrage, cap, labels = data223$x.country., cex=0.7, pos=3) model250 <- lm(cap ~ data223$x.executive.parties ) summary(model250) abline(lm(cap ~ suffrage)) model251 <- lm(cap ~ data223$x.executive.parties suffrage) summary(model251) model252 <- lm(cap ~ suffrage+data223$x.executive.parties data223$x.logarithm.of.2009.population.) summary(model252) model252 <- lm(cap ~ suffrage+data223$x.executive.parties data223$x.logarithm.of.2009.population.+data2 23$europe) summary(model252) model252 <- lm(cap ~ suffrage+data223$x.executive.parties data223$x.logarithm.of.2009.population.+data2 23$europe+data223$X.EIU.democracy.index.) summary(model252) model252 <- lm(cap ~ suffrage+data223$x.executive.parties data223$x.logarithm.of.2009.population.+data2 23$europe+data223$X.EIU.democracy.index.) summary(model252) options(scipen = 5) summary(suffrage)

11 summary(lm(suffrage~data223$x.hdi.2010.)) e <- data223$x.executive.parties *-1 plot(data223$x.executive.parties *-1, cap, xlab = "Executive-party dimension ", ylab = "GDP per capita in 2010") identify(data223$x.executive.parties *-1, cap, labels = data223$x.country., cex=0.7, pos=3) model253 <- lm(cap~data223$x.executive.parties data223$europe+data223$x.logarithm.of.2009.po pulation.) summary(model253) abline(lm(cap~e)) data223$europe = data223$x.country. %in% c('aut', 'BEL', 'DEN', 'FIN', 'FRA', 'GER', 'ICE', 'IRE', 'ITA', 'LUX', 'MAL', 'NET', 'NOR', 'POR', 'SPA', 'SWE', 'SWI', 'UK') data223$europe data223$medianage = data223$x.country. %in% c('swi', 'JPN', 'ITA', 'ICE', 'IRE', 'US', 'SWE', 'NET', 'LUX', 'BEL', 'UK', 'FIN', 'AUT', 'CAN', 'DEN', 'NOR', 'AUS', 'NZ') data223$medianage = data223$x.country. %in% C('SWI', 'JPN', 'ITA', 'ICE', 'IRE', 'US', 'SWE', 'NET', 'LUX', 'BEL', 'UK', 'FIN', 'AUT', 'CAN', 'DEN', 'NOR', 'AUS', 'NZ') medianage <- data223$medianage model544 <- lm(cap~medianage) summary(model544) medianage model543 <- lm(cap~suffrage, data = data[!medianage,]) summary(model543) plot(cap~suffrage, data = data[!medianage,]) medianage1 = C('SWI', 'JPN', 'ITA', 'ICE', 'IRE', 'US', 'SWE', 'NET', 'LUX', 'BEL', 'UK', 'FIN', 'AUT', 'CAN', 'DEN', 'NOR', 'AUS', 'NZ') mean(suffrage) sd(suffrage) mean(data200$x.my_cpi_inflation_ ) sd(data200$x.my_cpi_inflation_ ) IQR(data200$X.my_cpi_inflation_ ) model201 <- lm(data200$x.my_gdp_ ~ execpart) summary(model201) abline(lm(data200$x.my_gdp_ ~ execpart)) model232 <- lm(data200$x.my_gdp_ ~ execpart+data200$x.logarithm.of.2009.population.) summary(model232) sd(execpart) plot(execpart, data200$x.my_cpi_inflation_ , xlab = "Executive-party dimension", ylab = "CPI inflation multiplicator") identify(execpart, data200$x.my_cpi_inflation_ , labels = data200$x.country., cex=0.75, pos=3) model202 <- lm(data200$x.my_cpi_inflation_ ~ execpart)

12 summary(model202) abline(lm(data200$x.my_cpi_inflation_ ~execpart)) plot(execpart, data200$x.mean.ilo.labour.force.survey.unemployment., xlab = "Executive-party dimension ", ylab = "Mean unemployment rate ") identify(execpart, data200$x.mean.ilo.labour.force.survey.unemployment., labels = data200$x.country., cex=0.75, pos=3) model203 <- lm(data200$x.mean.ilo.labour.force.survey.unemployment.~execpart+data200$x.logarithm.of population.+data200$X.HDI.2010.) summary(model203) abline(lm(data200$x.mean.ilo.labour.force.survey.unemployment.~execpart)) #Even if we control for the log of population size and the level of development we still get a statistically significant result. model221 <- lm(data200$x.mean.ilo.labour.force.survey.unemployment.~execpart+data200$x.logarithm.of population.) summary(model221) model222 <- lm(data200$x.mean.ilo.labour.force.survey.unemployment.~execpart) summary(model222) names(data201) plot(execpart, data201$cpi_1991_2009, xlab = "Executive-Parties Dimension", ylab = "Multiple of CPI inflation increase ") text(execpart, data201$cpi_1991_2009, labels = data200$x.country., cex=0.5, pos=3) model204 <- lm(data201$cpi_1991_2009~execpart+data201$pop_in_thousands_2009) summary(model204) abline(lm(data201$cpi_1991_2009~execpart)) #Dataset that excludes Uruguay, CR and Jamaica data202<-read.csv("dataminusurujam.csv") data202 names(data202) execpart2 <- data202$x.executive.parties *-1 plot(execpart2, data202$x.my_cpi_inflation_ , xlab = "Executive-party dimension ", ylab = "Multiple of CPI increase ") identify(execpart2, data202$x.my_cpi_inflation_ , labels = data202$x.country., cex=0.7, pos=3) model205 <- lm(data202$x.my_cpi_inflation_ ~execpart2+data202$x.logarithm.of.2009.population.+d ata202$x.hdi.2010.) summary(model205) abline(lm(data202$x.my_cpi_inflation_ ~execpart2)) model223 <- lm(data202$x.my_cpi_inflation_ ~execpart2+data202$x.logarithm.of.2009.population.) summary(model223)

13 model224 <- lm(data202$x.my_cpi_inflation_ ~execpart2) summary(model224) data204<-read.csv("my dataset.csv") data204 plot(data204$x.executive.parties , data204$x.mean.ilo.labour.force.survey.unemployment., xlab = "Plenary Agenda", ylab = "Mean unemployment") text(data204$x.executive.parties , data204$x.mean.ilo.labour.force.survey.unemployment., labels = data204$x.country., cex=0.5, pos=3) model211 <- lm(data204$x.mean.ilo.labour.force.survey.unemployment.~data204$x.executive.parties ) summary(model211) abline(lm(data204$x.mean.ilo.labour.force.survey.unemployment.~data204$x.executive.parties )) data205<-read.csv("economicdata.csv") data205 names(data205) dummyexecpart6 <- data205$x.executive.parties execpart205 <- dummyexecpart6*-1 plot(execpart205, data205$x.my_gdp_ , xlab = "Executive-party dimension", ylab = "GDP per capita increase ") text(execpart205, data205$x.my_gdp_ , labels = data205$x.country., cex=0.5, pos=3) model212 <- lm(data205$x.my_gdp_ ~execpart205) summary(model212) abline(lm(data205$x.my_gdp_ ~execpart205)) plot(execpart205, data205$x.mean.ilo.labour.force.survey.unemployment., xlab = "Executiveparty dimension", ylab = "mean UNEMPLOYMENT ") text(execpart205, data205$x.mean.ilo.labour.force.survey.unemployment., labels = data205$x.country., cex=0.5, pos=3) model213<- lm(data205$x.mean.ilo.labour.force.survey.unemployment.~execpart205) summary(model213) abline(lm(data205$x.mean.ilo.labour.force.survey.unemployment.~execpart205)) names(data200) model214<lm(data200$x.my_cpi_inflation_ ~data200$x.index.of.central.bank.independence ) summary(model214)

14 model2016 <- lm(data207$x.change_in_gdp_per_capita_ ~execrisis+data207$x.logarithm.of.2009.pop ulation.) summary(model2016) model2017 <- lm(data207$x.change_in_gdp_per_capita_ ~execrisis+data207$x.logarithm.of.2009.pop ulation.+data207$x.hdi.2010.) summary(model2017) lm(data200$x.mean.ilo.labour.force.survey.unemployment.~execpart+data200$x.logarithm.of population.+data200$X.HDI.2010.)