This is a preregistration of an extension project for the following study:
Democracy Does Cause Growth by, Daron Acemoglu, Suresh Naidu, Pascual Restrepo, and James A. Robinson
Reference the chosen paper for your project here
(Acemoglu et al., 2014)
In my replication project I am focusing on the following argument made in the study:
Describe scientific rationale for the study extention.
Looking at the study, it gives clear argument and reasoning that the overriding takeaway is that democratization tends results in an average of 20% increase of GDP per capita after 25 years of democratization.
It gives clear reasoning of potential channels that democracy can increase GDP by, being freedom of capital, labour and trade, and education. However it only controls for these covariate x variables, as they are not democracy itself.
There is already a lot of literature out there that looks at differing fields and factors and how they can effect GDP, our focus is simply on democracy, and therefore we place the limit on the covariates at only controlling them.
The study also looks at it on a more zoomed scale, looking at differing regions
Democracy, is explored as a singular measure but there are many different forms of democracy or democratic values, we can extend the study by looking at the regressional effect and correlation between individual democratic factors or values against GDP per capita, to see which factor or value of democracy is most significant to increasing GDP, as countries may obtain such factors withoutb eing a full blown democracy.
Hypothesis:
I hypothesis that looking at the effects of the individual factors of democracy or democratic values will show us a split of the effect of total democracy on GDP per capita, expecting to find certain factors as highly valuable in raising GDP per capita, whilst some factors maybe unexpectedly don’t play a huge role.
Do you expect to get the same or different results from the original study? Why?
Different to a certain degree, as I doubt adding the average effect on GDP per capita of each factor will result in a 20% increase when looked at the 25 year implementation of each factor, as…
Firstly certain factors are likely to be implemented at differing points in time, but also I would expect there to be an increase in GDP per capita coming from the cumulative effect of each factor confounding into an even stronger total effect of democracy when they work in parallel.
Possible that, no factor is significant, but its only when they all work in tangent that democracy is significant.
This dataset also includes GDP growth, GDP per capita growth and GDP per capita values. Alongside all this data for specific years, all the way from 1789 to 2019.
Also, I’ve used the World Bank data depository, which has a wide array of datasets with economic and demographic, variables. Specifically, World Development Indicators. I’ve selected a few more related control variables and variables of interest to use in our analysis. (World Bank, 2023)
I will test the hypothesis using the same regression model applied in the Does Democracy Cause Growth study..
Like the model I have chosen to extend from the paper, the regression used is a linear panel regression, giving a time-spatial view of how the variables interact with eachother over time and the relationship is at different time points.
Many of these variables are given on a rating scale, varying dependent on the variable However many are unitary variables, quite self explantory so, such as GDP per capita, whilst many of our background variables such as Year are continuous.
Not many have been manipulated, trade variable from the V-Dem data set which is an addition of both the value of imports and exports variables together.
Dependent variable(s):
GDP per capita (e_migdppc)
Independent variables (IVs):
Electoral democracy index (v2x_polyarchy)
Liberal democracy index (v2x_libdem)
Participatory democracy index (v2x_partipdem)
Deliberative democracy index (v2x_delibdem)
Egalitarian democracy index (v2x_egaldem)
Investment share of GDP (WORLD BANK DATA) (AKA NET FOREIGN ASSETS INFLOW) (NAME)
Trade % of GDP (WORLD BANK DATA) (NAME)
Tax revenue share of GDP (WORLD BANK DATA) (NAME)
Imports and exports (e_cow_imports) (e_cow_exports)
Natural resource wealth (e_total_fuel_income_pc)
Inflation (e_miinflat)
Primary school enrollment (v2peprisch)
Secondary school enrollment (v2pesecsch)
Infant mortality rate (e_peinfmor)
Population total (e_mipopula)
Rule of law - estimate (By world bank indicators) (e_wbgi_rle)
Political stability - estimate (e_wbgi_pve)
GDP growth (rescaled) (e_migdpgrolns)
Freedom to research and teach (v2cafres)
Rule of law (Absence of war and insurgency and the civil control) (e_fh_rol)
Civil war (e_civil_war)
Year (year)
Country (country_name)
Firstly, we will have use of the simple coefficient magnitude and correlation, which will allow us to see a simplistic view of the correlation to our Y variable with our selected X variables of interest, which are the aspects or features of democracy. We will further this by once controlling for our control variables, looking at the magnitude of the coefficients in our statistic table, to see which is the strongest or weakest, by their positive or negative effect and the size of the magnitude through the coefficient.
Next we will look for statistical significance in our model, and correlation to the Y variable through a p-test, taking p-values of below 0.05 as significant in our model.
Our adjusted R-squared can also give an indication of how much the specified aspects of democracy explain the Y variable.
However like done on the paper, use of different lagged variables or alternate effects test the stability of our results and analysis.
Multiple ways to figure out whether it is worth to exclude outliers or not. Of course it should be done pragmatically to decide whether the outliers would truly distort the data and make our analysis biased, or whether it is just truly representative of the real world as of course there will be in the case of certain factors, like GDP per capita, outliers both in the extreme high and lows.
We may use visual plots, and graphs such as histograms, scatterplots and boxplots, another way is to use Z-tests which will allow us to see how many standard deviations, observations are from the mean and exclude the worst dependent on criteria of whether it is an outlier or not. ## 4.5. Missing data
State your process for dealing with missing data or state not applicable.
Simply, dependent on the test being done, if we have observations which do not have values for a considerable amount of control variables we will exclude it from the data Whilst observations without the key y and x variables of our democracy factors, will just be removed. However to test the analysis further, in this case a dummy indicator variable could be made to show the effect of these observations, and whether there is a correlation or reasoning behind why these observations happen to be missing certain variables.
Firstly, earlier on, some stability tests will be done such as breusch-pagan tests
But also as we already had stated in previous literature and within the paper, many of the control variables are channels allow the effects of democracy to cause growth.
Also whilst we are going to find a global analysis, just like the paper looked at, it may be prominent to subset and conduct our analysis repeated but at differing regions, to see whether results differ dependent on region, and from there we can test why this may be if results differ, culture, religion, political geography, etc.
To find out the true effect of the factors of democracy and whether some are more significant in increasing growth dependent on the region.
This preregistration form was completed in the following R environment:
## R version 4.3.0 (2023-04-21)
## Platform: aarch64-apple-darwin20 (64-bit)
## Running under: macOS Monterey 12.1
##
## Matrix products: default
## BLAS: /Library/Frameworks/R.framework/Versions/4.3-arm64/Resources/lib/libRblas.0.dylib
## LAPACK: /Library/Frameworks/R.framework/Versions/4.3-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.11.0
##
## locale:
## [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
##
## time zone: Europe/London
## tzcode source: internal
##
## attached base packages:
## [1] stats graphics grDevices utils datasets methods base
##
## loaded via a namespace (and not attached):
## [1] digest_0.6.31 R6_2.5.1 fastmap_1.1.1 xfun_0.39
## [5] cachem_1.0.8 knitr_1.42 htmltools_0.5.5 rmarkdown_2.21
## [9] cli_3.6.1 sass_0.4.6 jquerylib_0.1.4 compiler_4.3.0
## [13] rstudioapi_0.14 tools_4.3.0 evaluate_0.21 bslib_0.4.2
## [17] yaml_2.3.7 rlang_1.1.1 jsonlite_1.8.4Bartlett J. (2021). OSF preregistration template.Rmd. GitHub (accessed 03/05/2023)
Bowman, S. D., DeHaven, A. C., Errington, T. M., Hardwicke, T. E., Mellor, D. T., Nosek, B. A., & Soderberg, C. K. (2020). OSF Prereg Template. https://doi.org/10.31222/osf.io/epgjd. OSF (accessed 03/05/2023)
Reference list
Acemoglu, D., Naidu, S., Restrepo, P. and Robinson, J.A. (2014). Democracy Does Cause Growth. SSRN Electronic Journal, 127(1). doi:https://doi.org/10.2139/ssrn.2411791.
World Bank (2023). World Bank Open Data | Data. [online] Worldbank.org. Available at: https://data.worldbank.org.
www.v-dem.net. (2023). Home | V-Dem. [online] Available at: https://www.v-dem.net.