RESEARCH UNDERPINNING THE GRADE MODEL
Government revenue, quality of governance and child and maternal survival
Highlights
This paper considers the effect of government revenue and the quality of governance on the under-five and maternal survival rates.
A non-linear panel data study was undertaken using annual data for every country in the world.
Significance of the Study
The study’s broad conclusion is that while government resources, measured as total revenue received by a country’s government, is essential, the quality of governance is even more critical in determining a good outcome for both mother and child.
Research Approach
We define our two dependent variables, under-five survival rate and maternal survival rate.
A standard linear model or a log-log model with constant elasticities would be inappropriate for this data.
Therefore, we need a model with a broadly defined ‘S’ shape that starts with minimal effects for very low government revenue levels per capita and then has a period of rapid growth followed by a falling off as saturation is reached.
The ideal model is described by a broad family of functional forms called sigmoids, and within this family, the most widely used function is the logistic function.
f(x)=Min+(M−Min)/(1+e−((α+χw)(x−(β+δw)))
Where Min is the non-zero minimum for the dependent variable, which we set at 95%, 𝑤 is a x vector of exogenous variables and 𝜒and𝛿 are 1xk vectors of parameters. This allows the shape of the logistic curve to vary for each country depending on the variables in the 𝑤 vector (our governance indicators).
Study Results
Some of the Worldwide Governance Indicators (WGI) are significant in every case demonstrates that the shape of the logistic curve does indeed vary between countries in a significant way.
It is not possible to interpret these parameters in the usual way as elasticities or partial derivatives. They are simply parameters which govern the shape of the ‘S’ shaped curve, that is, how steep it is and where the sharp rise occurs.
For this reason, we show the actual shapes of the curves for some sample countries, which illustrates exactly how important both the basic two parameters are and the role the governance indicators play in varying the shape of the curve in each country.
Findings
We find that increased government revenue is associated with an increase in both child and maternal survival rates. For low-income countries, this effect is dominated by the effect of improving the quality of governance.
When government revenue reaches a level of around 5000 USD per capita, governance becomes much less critical.
This modelling offers the ability to model the impact of a change in government revenue in the individual country while accounting for governance.
It is a valuable addition to the armoury of those who advocate for reduced corruption and increased government revenue.