Transition Report 2013 Stuck in transition?

CH2 180sq

Facts at a glance

70% Global proportion of countries which had democratic institutions in 2012, compared with 30 to 40 per cent from 1960 to 1990.

INCOME IN 1992 is correlated with levels of democracy in 2012 in a global sample.

Cover 180sqV2

 

94% of countries with average per capita income above US$ 10,000 held free and competitive elections in 1999.

BY 2000 all constituent democracies of the former Yugoslavia had become full democracies.

Markets and democracy

Economic development and democracy: evidence

Does economic development encourage democratisation in transition countries? The literature suggests at least four reasons why average per capita income might influence a country’s propensity to democratise.

  • At higher average income levels, high-income voters will be more willing to accept the redistributional consequences of democracy, especially if the costs of repression are considered excessive.
  • Development is generally correlated with lower levels of inequality, at least in the long term.1
  • Development is linked to a shift in the nature of wealth ‒ that is to say, from fixed assets, such as land, to mobile capital.
  • Higher per capita income is associated with education and secularisation, with educated citizens being more likely to demand political participation and to embrace democratic beliefs.

Table 2.1 presents the results of a simple regression of the level of democracy on economic development for a global sample over the period from 1800 to 2000. It looks at the impact of economic growth on the development of democracy with lags of 5, 10 and 25 years. This shows that GDP per capita has a strong positive impact on the emergence of democracy globally. Levels of democracy will be higher today for countries that were richer 5, 10 or 25 years ago (see the three left-hand columns of the table).

Table 2.1

Historically, higher per capita income has been a predictor of democratisation – but that is not the case in today’s transition region

  Countries outside transition region
Lag length τ(years)
Countries in transition region
Lag length τ(years)
  5 10 25 5 10 25
Polity at t-τ
0.66*** 0.39*** 0.21** 0.58*** 0.07 0.55**
Log of GDP per capita at t-τ
0.04** 0.14*** 0.23** -0.04 0.1 -0.04
Observations 2007 911 269 163 78 26
Countries 143 137 78 11 11 8
R2 0.81 0.67 0.55 0.82 0.69 0.37

Source: Polity is taken from the Polity IV dataset. Per capita GDP, which is in 2000 US dollars, is taken from Gleditsch (2002) for the period 1950-2004 and Maddison (2008) for earlier years, as merged in Boix et al. (2012).
Note: The table shows regressions for a sample period of up to 200 years (1800-2000). The dependent variable is the Polity index of democracy at time t. * = p<0.10; ** = p<0.05; *** = p<0.01.

Unsurprisingly, the relationship between economic growth and democracy does not hold true in the countries that make up today’s transition region, many of which were part of non-democratic states or empires for much of their recent history – the Habsburg, Ottoman and Russian empires prior to the First World War, and then the Soviet Union or one of its satellites in eastern Europe.

During these periods many countries in the transition region experienced rapid development led by industrialisation, but remained undemocratic. Consequently, the rest of this analysis focuses on the period between 1989 and 2012 to examine the relationship between economic development and political regime outcomes in the post-communist period.

Table 2.2 shows the results of a panel regression that is analogous to that of Table 2.1, except that it also includes measures of natural resource endowments (as a share of GDP) and income inequality (measured by the Gini coefficient). These are variables that should, based on the preceding analysis, influence the propensity to democratise for a given level of per capita income.

As in Table 2.1, the regression considers the relationship between democracy in year t and lagged GDP per capita (as well as income inequality and natural resources). Because of the shorter length of the sample, the lag length is always five years.2

Three variants of the dependent and lagged dependent variable are considered: first, the level of democracy as expressed by the Polity index (as in Table 2.1); second, the highest Polity score over the preceding five years (Max5Polity), which effectively restricts the analysis to cases where there has been an increase in democracy; and lastly, the minimum score over the previous five years (Min5Polity), which restricts the analysis to cases where there has been a decline in democracy.

The rationale for analysing these variants in addition to the Polity index at time t is that the effect of some of the explanatory variables may not be the same when it comes to promoting or delaying democratic improvements and when it comes to defending or undermining a level of democracy that already exists.

Table 2.2

Determinants of democracy in the transition region and in all other countries, 1989-2012

Dependent variable Polity Max5Polity Min5Polity Polity Max5Polity Min5Polity Min5Polity
   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Transition region All other countries Transition region All other countries Transition region All other countries Transition region Transition region Transition region Transition region
Polity at t-5 0.139*** 0.679*** 0.397*** 0.635*** 1.003*** 0.944*** 0.095* 0.386*** 0.961*** 0.950***
-0.01 0 0 0 0 0 -0.07 0 0 0
Log of GDP per capita at t-5 0.545 0.391** 1.204** 0.356** 0.474** 0.276*** 0.343 1.191** 0.375 0.275
-0.36 -0.01 -0.01 -0.02 -0.05 0 -0.57 -0.02 -0.15 -0.31
Natural resource rents at t-5 -0.027* -0.037*** -0.080*** -0.030*** -0.002 -0.012** -0.036** -0.081*** -0.006 -0.006
-0.09 0 0 -0.01 -0.83 -0.04 -0.03 0 -0.54 -0.55
Inequality at t-5 0.026 0.021 -0.041 0.025 0.045* -0.004 -0.046 -0.051 0.018 0.031
-0.49 -0.26 -0.27 -0.16 -0.07 -0.67 -0.32 -0.3 -0.57 -0.35
Transition indicator at t-5             0.732*** 0.117 0.608*** 0.544**
            0 -0.67 0 -0.01
EU membership                   0.503
                  -0.24
Constant -1.027 -1.794 -4.698 -1.234 -6.022** -2.183*** 1.082 -4.537 -5.712** -5.213*
-0.86 -0.29 -0.33 -0.46 -0.01 -0.01 -0.85 -0.39 -0.03 -0.05
N 103 376 103 376 103 376 95 95 95 95
chi2 17.6 849.8 181.5 793.4 1816 6475.4 25.4 161.6 1822.2 1849.5
p 0.001 0 0 0 0 0 0 0 0 0
ll -235.3 -916.2 -217.7 -838.5 -182.8 -664.8 -216 -204.2 -166.3 -165.6
p_c 0 0.158 0 0.001 1 0.377 0 0 1 1

Source: Polity IV, EBRD (for transition indicators), World Bank World Development Indicators.
Note: The table shows the results of a panel regression involving observations at four different points in time – 1995, 2000, 2005 and 2010. The estimation technique is a multi-level mixed (fixed and random) effects, maximum likelihood model, using Stata’s xtmixed command. Errors are clustered at the country level. Polity refers to the Polity2 index, Max5Polity to the maximum level of the Polity2 index over the preceding five years, and Min5Polity to the minimum value of the index over the previous five years. P-values are shown in parentheses. * denotes that p<0.10; ** that p<0.05; and *** that p<0.01. “Natural resource rents” refers to the share of natural resource production in GDP, and “inequality” refers to the Gini coefficient of income inequality. “EU membership” is a variable taking the value 1 if a country is among the 10 new Member States in central and eastern Europe that joined the European Union in 2004 and 2007, and 0 otherwise. N denotes the number of observations, p the overall significance level of the regression, chi2 the chi-squared statistic, ll the log of the likelihood of the comparison model, and p_c the p-value of the comparison model.

Table 2.2 shows that, when controlling for the type of political regime in place five years previously, for natural resources, and for inequality, the probability of a country becoming more democratic depends strongly on lagged GDP per capita (see columns 1 to 6). The coefficient estimated is larger for the transition region than for the rest of the world, and statistically significantly larger than zero in all specifications except model 1. The effect of lagged GDP growth appears to be larger as regards inducing democratic improvements ‒ columns 3 and 4 ‒ than it is when it comes to protecting countries from democratic reversals ‒ columns 5 and 6.

  1. See Atkinson et al. (2009), Davies and Shorrocks (2000) and Morrisson (2000). [back]
  2. For each country, the dependent variable in the first observation in the sample is the 1995 democracy score, while the lagged dependent and independent variables correspond to 1990; the second observation is the 2000 democracy score, while the lagged dependent and independent variables correspond to 1995, and so on.[back]

icon-toolsTools