5.3 Data and methodology
5.3.2 Independent variables
There are two main variables of interest: economic performance and the presence of upcoming elections.
5.3.2.1 Political
Variables indicating the various contexts of the political landscape are implemented as dummy variables attached to the relevant monthly cabinet observations. These dummy variables are explained in table 5.2 below.
Table 5.2 – Political Context Dummy Variables
Dummy Variable Description Before
parliamentary/presidential/any election
Assigns a dummy to the cabinet-months that occur in the year before an election. Separate dummies for presidential and parliamentary elections. After
parliamentary/presidential/any election
Assigns a dummy to the cabinet-months that occur in the 12 months after an election, including the month of the election. Separate dummies for presidential and parliamentary elections.
After Election – No Change in Leadership
Assigns a dummy to the cabinet-months that occur in the 12 months after an election, including the month of the election, in which the incumbent leader retained the presidency.
After Election – Change in Leadership
Assigns a dummy to the cabinet-months that occur in the 12 months after an election, including the month of the election, in which there was a change in leader. This includes observations in which the incumbent party retains power but is headed by a different leader.
Change in leader
Assigns a dummy to the cabinet-months that occur in the 12 months after a new person occupies the presidency (or highest executive office). This includes any change in leader including replacement through election, coup, party succession or the implementation of a transitional government.
Democratic Change in Power
Assigns a dummy to the cabinet-months that occur in the 12 months after a new person occupies the presidency (or highest executive office) through electoral victory. This includes observations in which the ruling party remains the same but with a different individual as president.
Non-Democratic Change in Power
Assigns a dummy to the cabinet-months that occur in the 12 months after a new person occupies the presidency (or highest executive office) through non- democratic means such as a coup or rebellion.
Unity government
Assigns a dummy to the cabinet-months that occur in the 12 months after the implementation of a unity government. A unity government is defined as a power-sharing arrangement in which cabinet posts are shared as part of an internationally brokered peace deal. It does not include instances in which the leader decides to incorporate opposition politicians into the cabinet without a peace agreement.
5.3.2.2 Economic
Data on economic performance is taken from the World Bank Development Indicators, specifically the annual GDP growth per capita. Growth is chosen in place of aggregate GDP or GDP per capita, as growth is more commonly used as a measure to indicate the economic health of a country with
recessions and slumps being defined as poor or negative growth rather than low GDP (Thomas and Dimsdale, 2010). The following variables are derived from the economic data.
Table 5.3 – Economic Variables
Variable Description
GDP Growth The annual growth in GDP for that year
Low Growth Year A dummy marking the five lowest years of growth for each country High Growth Year A dummy marking the five highest years of growth for each country
Table 5.4 shows the means, standard deviations and standard deviation in country means for the main variables. Cabinets show a high degree of volatility in size varying from 0 when the leader dissolves the cabinet and leaves all posts vacant to 47. This volatility is mainly driven by expansion and contraction within the outer circle. The average cabinet size also varies significantly across the eighteen countries, with Cameroon having an average cabinet size of 37.44 and Liberia an average of 22.04. The mean level of representation in the data is 75.64 with a standard deviation of 15.25, demonstrating that African cabinets generally represent the majority of relevant ethnic categories within the cabinet and exclusionary governments are rare. The inner circle is generally less representative than the outer circle. The leader’s co-ethnics typically occupy approximately a quarter of the cabinet. The higher average representation of the leader’s co-ethnics within the inner circle reflects the importance of controlling these posts.
The economic data shows a high degree of volatility in terms of GDP growth and changes in growth year on year. There is also a wide variance between countries, with Zimbabwe experiencing an average growth rate of -0.17 percent and Uganda averaging 6.23 percent growth. Economic growth is more volatile in competitive regimes, which have a standard deviation of 7.85 percent, as compared to hegemonic regimes, which have a standard deviation of 3.23 percent. All countries experience positive and negative changes in GDP. Even if certain countries under study escape severe recessions, the volatility in the growth rate means that countries, and their leaders, face varying economic circumstances. Low and high growth years represent a quarter of observations each.
Table 5.4 – Descriptive statistics of main variables
Variable Mean Min Max Standard Deviation Variance Across Countries
Total Cabinet Size 26.523 1 47 6.773 5.159
Inner Circle Size 7.168 1 13 2.040 1.762
Outer Circle Size 19.352 0 36 6.444 4.996
Representation 75.640 6.250 100 15.254 13.047
Inner Circle Representation 45.708 6.250 100 19.146 17.299 Outer Circle Representation 68.643 0 100 18.490 16.212 Disproportion Cabinet 24.920 7.850 51.167 7.893 5.637 Disproportion Inner Circle 27.765 1.167 67.933 8.986 5.802 Disproportion Outer Circle 25.610 7.074 100 8.264 5.340 Leader Co-Ethnic Representation 25.947 0 64.706 15.282 13.499 Leader Co-Ethnic Inner Circle Representation 27.102 0 83.333 20.369 15.090 Leader Co-Ethnic Outer Circle Representation 25.547 0 75 16.228 13.507
GDP Growth 4.993 -36.700 106.280 6.667 2.273
Change in GDP Growth -0.173 -76.072 94.159 7.475 0.630
Low Growth Year 0.213 0 1 0.410 0.042
High Growth Year 0.247 0 1 0.431 0.021
Below Mean Level Growth 0.488 0 1 0.500 0.139
Below Median Level Growth 0.468 0 1 0.499 0.037
Before Any Election 0.238 0 1 0.426 0.044
Before Presidential Election 0.192 0 1 0.394 0.047 Before Parliamentary Election 0.196 0 1 0.397 0.055
Any Reshuffle 0.146 0 1 0.353 0.048
Major Reshuffle 0.056 0 1 0.229 0.019
Minor Reshuffle 0.091 0 1 0.287 0.036
Pre-electoral cabinets represent approximately a quarter of observations but are slightly more common in competitive regimes. General elections, including both parliamentary and presidential elections typically, form 56.72 percent of observations.
Approximately 15 percent of observations involve a major or minor reshuffle, showing that ruling coalitions are reviewed regularly by the leader. Minor reshuffles are more common with major reshuffles occurring less frequently. Competitive regimes engage in minor reshuffles more regularly than hegemonic regimes, perhaps reflecting their greater vulnerability.