This is the first attempt to build a model that accounts for some of the observed variability in the Brazilian state GDP values. The investigation is attached as a PDF because some of the Minitab output does not retain its original format when it is pasted into WordPress.
Here are the potential predictor variables, their type, measurement, and the source used to obtain the data:
|State contribution to national GDP||Quantitative||Year 2010, measured in R$ 1,000,000||IBGE|
|Population||Quantitative||Year 2010, measured by individuals||IBGE|
|Municipalities||Quantitative||Year 2010, measured by individual municipalities||Fundação Joaquim Nabuco|
|Region||Qualitative||Year 2010, the federal government identifies five regions (more details below)||Fundação Joaquim Nabuco
|Foreign Border||Qualitative||Year 2010, border is defined as “yes” or “no”||Consult any modern day geopolitical map of Brazil.|
|Border with MERCOSUR country||Qualitative||Year 2010, border is defined as “yes” or “no”; MERCOSUR is defined as sharing a border with Argentina, Uruguay, Paraguay, or Venezuela. Venezuela was admitted to MERCOSUR in 2012, but its inclusion was considered important for the economic impact it could have had on border states||Consult any modern day geopolitical map of Brazil.|
|Foreign Presence||Quantitative||This is defined as the number of consulates in a given state. This would probably be more meaningful if we exclude the Federal District.||Wikipedia|
|Ocean Port||Quantitative||The number of ports in the state (measured by various sources between the years 2009 and 2015)||Wikipedia
|River Port||Quantitative||The number of ports in the state (measured by various sources between the years 2009 and 2015). I also considered making the predictor variables for ocean and river port qualitative.||Wikipedia
|Soy||Quantitative||The amount of thousand tons exported during the two harvests of year of the growing year 2010/2011.||CONAB|
|Corn||Quantitative||The amount of thousand tons exported during the two harvests of year of the growing year 2010/2011.||CONAB|
Corn and soy account for about 80% of the grains produced in Brazil. Soy has “immediate liquidity” since it is an international commodity while corn historically was used for internal supply (EMBRAPA).
When I look for significance of “region”, I should be aware that the sample size of each region is very small. There are 3 to 9 observations per group, with 9 for the Northeast being more the exception than the rule. I suspect this variable might have been significant if we coded municipalities per region and used their relative GDP instead of that for the state. If we opted to include this variable in the model, we would have included four dummy variables and chosen one baseline region.
After looking at associations, I could be interesting to see if the deleted residual for the Federal District makes the point influential. This unit is an anomaly in a few categories, especially in regards to the “foreign presence”.
Finally, I understand that with more time I would have consulted original sources instead of using data from second-hand sources.
I begin by a simple matrix plot of each variable regressed against the other.
Visually we can identify a few associations between variables:
- Population and number of municipalities
- Population and state GDP
- Soy and corn exports
Correlation: Population, Municipalities, GDP, Soy, Corn
Population Municipalities GDP Soy
GDP 0.953 0.629
Soy 0.072 0.292 0.064
Corn 0.353 0.607 0.313 0.860
Cell Contents: Pearson correlation
To a lesser degree, associations exist between:
· Foreign Presence and GDP
· Soy and GDP
· Corn and GDP
Correlation: Foreign Presence, GDP, Soy, Corn
Foreign Presence GDP Soy
Soy -0.067 0.064
Corn -0.032 0.313 0.860
Cell Contents: Pearson correlation
The rest of this investigation can be found via this link: 12-00-16-brazilian-gdp-project.