Kingmaker’s State of States 2016 – How They Ranked

The overall State of States ranking from Kingmakers shows how each of the 36 Nigerian states, plus the Federal Capital Territory Abuja, ranked in 34 indicators across seven categories.

In calculating the rankings, each of the seven categories were assigned equal weightings:

Business Environment: 14 percent
Economy: 14 percent
Education: 14 percent
Financial Health: 14 percent
Healthcare: 14 percent
Living Standards: 14 percent
Wealth Generation and Distribution: 14 percent

Note: Weights don’t add to 100 due to rounding.

The elements of the categories and the weightings assigned to each were:

Business Environment

Ease of Doing Business: 50 percent
Enterprise Performance: 50 percent

Economy

Economic Growth: 33 percent
Labour Participation: 33 percent
Unemployment Rate: 33 percent

Education

Adult Literacy: 25 percent
Educational Attainment: 25 percent
Education Quality: 25 percent
School Attendance: 25 percent

Financial Health

Budget Deficit: 33 percent
Debt Management: 33 percent
Revenue Generation: 33 percent

Healthcare

Infant Mortality: 25 percent
Child Mortality: 25 percent
Child Nutrition: 25 percent
Life Expectancy: 25 percent

Living Standards

Improved Drinking Water: 25 percent
Refuse Collection: 25 percent
Improved Sanitation Facilities: 25 percent
Use of Solid Fuels: 25 percent

Wealth Generation & Distribution

GDP Per Capita: 33 percent
Poverty Rate: 33 percent
Income Inequality: 33 percent

Kingmakers created an index score for each metric for each state. In each metric, the top state based on the raw data was given 1 point and the bottom state was given 37 points. States between these were indexed proportionally. For the overall rankings, Kingmakers created an average of the seven category rankings, and then ranked the outcome. We chose this method for the overall ranking so that it would not be skewed by large differences in scores at a metric level.

The Data

In order to select metrics for the project, we consulted experts in each of the categories for guidance. These are the criteria we used in choosing metrics:

Metrics that measure citizen outcomes in a state were favoured over inputs or outputs. For example, we selected the percentage of state residents who had finished primary school education or higher instead of the amount spent on education per capita or the number of students enrolled at state schools.
The set of metrics within each category should work together to provide an overview of that category.
Standardized data needed to be available across all or most states.

The data behind the rankings were sourced from

Kingmakers.com Calculations

Population

The data for the population for the states was acquired from the National Population Commission and National Bureau of Statistics. We got the total population for each of the states of the years 2006, 2008, 2010 and 2011. We then calculated the compound average growth rate (CAGR) from 2006 to 2011 and used that to extrapolate the population count for the years 2012 to 2015.

V(t₀): start value, V(t_n): finish value, t_n – t₀: number of years.

The same method was used for calculating for the values for the male and female population as well as the distribution across age groups.

There are a few limitations to using CAGR to forecast metrics and they include:

CAGR calculates the smooth average of growth over a period, it ignores volatility and implies that the growth during that time was steady. Yet, this is never actually the case. As such, we cannot take CAGR at face value.
CAGR is mainly a historic measurement and , no matter how steady the growth of a metric has been over a period of time we cannot safely assume that the growth will continue at the same rate during the following year or years, as other factors may come into play and affect that rate of growth.
Lastly, CAGR has a problem with representation. Say for instance a metric’s CAGR could be an impressive 50% over the past three years. However, due to poor or negative growth in the two years preceding those three years, the CAGR over the past five years could come down to more modest 2.73%.

However, still we adopt a culture of collecting and collating data more periodically, CAGR is one of the tools are our disposal to make calculated guesses to fill in the void for the years no data was collected.

Adult Literacy Rates

The data used for adult literacy rate was got from the Demographic and Health Surveys (DHS) Program. There were surveys were carried out in the following years: 1999, 2003, 2008 and 2013.

The results from 1999 were excluded from our calculations as we considered the data was too old. The 2003 and 2008 surveys only contained results for the six geopolitical zones in which the states were located, while only the 2013 survey contained results for both the geopolitical zones and the individual states.

We calculated the CAGR from each of the geopolitical zones from 2008 to 2013 and 2003 to 2013.

Since we had only data at state level from 2013, we used the CAGR from 2008 to 2013 to retrogressively get the values for 2012, 2011 and 2010.

V₀ = V₁ x (100%-CAGR(_{2008, 2013}))

V₀: Retrogressive value to be calculated, V₁: value from the succeeding year, CAGR(_{2008, 2013}): compound annual growth between 2008 and 2013 for the geopolitical zone in which the state is located.

For example:

Abuja FCT
2013 Adult Literacy Rate (V₁) = 84.6%
CAGR(_{2008, 2013}) for North Central Zone = 2.1%

2012 Adult Literacy Rate (V₀) = 84.6 x (100% - 2.1%) = 82.8

To get the values for the years succeeding years from 2013 to 2015, we got the value for the by multiplying the value of the preceding year with the following formula

V₀ = V₁ + (V₁ x CAGR(_{2003, 2013}))

V₀: Progressive value to be calculated, V₁: value from the preceding year, CAGR(_{2003, 2013}): compound annual growth between 2003 and 2013 for the geopolitical zone in which the state is located.

By using the longer period of 2003 to 2013 for the CAGR, we were looking to have a much better representation of forecasting what the value might be as we do not have any data for the years succeeding 2013.

Same method used for: Labour Participation, Infant & Child Mortality, Primary & Secondary School Attendance, Child Nutrition

WAEC Results

For the 2015 WAEC Results, while we got the rankings for all the states, we were unable to get the pass rate for two thirds of the states. The states were Abia (1) - 63.94%, Anambra (2) - 61.18%, Edo (3) - 61.05%, Lagos (6) - 48.02%, Ekiti (11) - 41.97%, Ogun (19) - 32.91%, Kano (24) - 25.44%, Borno (25) - 24.65%, Oyo (26) - 21.03%, Niger (27) - 19.66%, Adamawa (28) - 18.08%, Osun (29) - 18.03% and Yobe (37) - 4.37%

We used linear regression equations, similar to the one below, to get an approximate value for the missing pass rates based on their ranks after listing the states according to their ranks.

y = a + bx

y: the projected pass rate for state, a: the y-intercept, b: the slope and x: the state rank.

The y-intercept and slope were recalculated using the reported pass rates before the missing information, and those just after.

Unemployment Rate

The International Labour Organisation’s (ILO) recommendation for calculating unemployment is cover people who are: out of work, want a job, have actively sought work in the previous four weeks and are available to start work within the next fortnight; or out of work and have accepted a job that they are waiting to start in the next fortnight. The calculation should exclude those within the labour force who do not have a job and are also currently not seeking employment.

This is the method that the National Bureau of Statistics currently uses for calculating unemployment on a national level. The previous simply looked at those who were currently unemployed in the labour regardless of their job seeking status and expressed that as a percentage of the total labour force.

Looking at the national data, we found that there was a correlation between the percentage of people currently not employed regardless of their job seeking status and the unemployment rate which uses the recommendations stipulated by the ILO. We developed a linear regression equation which gives an approximate value for the unemployment rate using the percentage of people not actively participating in the labour force as its input.

We took data from the DHS Program to get the percentage of people not actively involved in the labour force by adding the number of people who were not currently employed and those who had not been employed in the last 12 months of the survey. Using the same technique as stipulated earlier for calculating the adult literacy, we forecasted the non-participatory labour rates for all the states from 2011 to 2015, and using those values as the input for our linear regression equation, we derived the unemployment rates.

UR = ( ( NCE + NE12 ) x 0.743947 ) - 11.1697

UR: Unemployment rate, NCE: Percentage of labour force not currently employed, NE12: Percentage of labour force not employed in the last 12 months.

Adult Life Expectancy

Looking the national data for infant mortality rates and the life expectancy at birth since 1980, we noticed that was a negative correlation between the two metrics. As infant mortality rates drop, life expectancy at birth increased. The data was used to develop a linear regression equation which had the following statistics

The infant mortality rates were calculated using data from the DHS Program and the applied to the linear regression equation to get the adult life expectancy.

Life Expectancy = 59.87237 – ( 0.111 x 'Infant Mortality Rate' )

Cooking Fuel

We got the data for the type of cooking fuel used by Households in Nigeria from the Nigerian Bureau of Statistics which had surveys from 2007 and 2011. The CAGR for each individual states was calculated and then used to forecast possible values for 2010 and 2012 – 2015. The drawbacks of using CAGR as a method as forecasting has already been discussed and its limitation in the area of proper representation was particularly pronounced as in some case, the metric measurement forecasted presented as a percentage of the household using solid fuels for cooking was more than 100 percent. These values had to be adjusted to the maximum possible value.

We did this technique for the data for access to sanitary facilities, improved drinking water and improved refuse collection.

Poverty Rate

To come up with the rankings for our poverty we decided to use the multidimensional approach which takes into account several factors that constitute poor people’s experience of deprivation – such as poor health, lack of education, inadequate living standard, lack of income (as one of several factors considered), disempowerment, poor quality of work and threat from violence.

A multidimensional measure can incorporate a range of indicators to capture the complexity of poverty and better inform policies to relieve it. Different indicators can be chosen appropriate to the society and situation.

For the purpose of the 2016 rankings we used the indicators and the weights as stated below. The indicators were chosen to reflect lack of education, poor health and inadequate living standards which are all directly under the purview of the state governments. The weightings were similar to those used by Oxford Poverty & Human Development Initiative (OPHI) and the United Nations Development Programme in drawing up their up poverty reports.

LACK OF EDUCATION INDICATORS
- Educational Attainment: This took into account the proportion of the population who had not completed primary school education. (17 percent)
- School Attendance: The average of the proportion of school aged children that were not attending both primary and secondary school. (17 percent)
POOR HEALTH INDICATORS
- Child Mortality Rate: The proportion of children who died within 5 years from birth. (17 percent)
- Child Nutrition: The proportion of children who had developed stunted growth. (17 percent)
INADEQUATE LIVING STANDARDS
- Use of Solid Fuel: The proportion of the population that use solid fuels to cook. (8 percent)
- Unimproved Sanitation Facilities: Proportion of population with no access to improved sanitation facilities. (8 percent)
- Unimproved Drinking Water: Proportion of population with no access to improved drinking water sources. (8 percent)
- Unimproved Refuse Collection: Proportion of population with no access to improved refuse collection. (8 percent)

The sum of the indicators gave us the proportion of the population that were likely to be facing multidimensional poverty in each of the states.

GDP Per Capita

The National Bureau of Statistics in 2012 released the Gross Domestic Product (GDP) estimates for each of the states in Nigeria and the Federal Capital Territory, Abuja for the year 2010. With those figures we were able to get an estimate of the GDP Per Capita for each state by dividing the GDP by the estimated population for the year 2010.

After going through several models in order to forecast the GDP for each of the state, we settle on a multiple regression model that forecasted the change in the GDP Per Capita of each of the state.

We derived the model by looking at the change in the GDP per capita for the whole of Nigeria from 2010 to 2015 and then examined all the socioeconomic indicators we had available to us. We settled on the following 5 independent predictors when their relationships were looked at gave us the highest statistical relevance possible in determining the rate of change in GDP per capita.

The indicators included

Adult Life Expectancy
Infant Mortality Rate
Female Labour Participation Rate
Unemployment Rate
Adult Literacy Rate

The model used for calculating is shown below.

Y = b₀ + b₁X₁ + b₂X₂ + b₃X₃ + b₄X₄ + b₅X₅

Y: Percentage in GDP per capita
b₀: Intercept
b₁: Adjusted percentage change in adult life expectancy
X₁: Coefficient for percentage change in adult life expectancy
b₂: Adjusted percentage change in infant mortality rate
X₂: Coefficient for percentage change in infant mortality rate
b₃: Adjusted percentage change in female labour participation rate
X₃: Coefficient for percentage change in female labour participation rate
b₄: Adjusted percentage change in unemployment rate
X₄: Coefficient for percentage change in unemployment rate
b₅: Adjusted percentage change in adult literacy rate
X₅: Coefficient for percentage change in adult literacy rate

The value of each of variables were adjusted in order to deal with the wide variation in the changes to the value of the indicators on the national level as compared to those experienced on the state level. While the changes on the national level tended to be in small units, that on state level tended to be much larger, sometimes in the tens of units.

We adjusted the values being passed to the model to deal with the wide variations by multiplying the input value of the state with an indicator factor. The indicator factor was derived from finding the value of the percentage of in the indicator at a particular state level that gave a unit change in value at the national level.

Once we had the GDP per capita for each of the state, we were able to determine the GDP itself and the GDP growth from 2011 to 2015.

Dr. Obi Igbokwe
Founder
Kingmakers