Wealthy states provide an immense
amount of developmental aid to poor. In 2013, members of the Organization
of Economic Cooperation and Development’s (OECD), Development Assistance
Committee (DAC) provided over $135 billion in aid to poor countries around the
world. [1] The OECD, a group of rich
countries, promotes this kind of developmental state-to-state aid known as
Official Development Aid (ODA). The OECD defines ODA as aid provided by a
government that is designed to promote the “economic development and welfare of
developing countries” that is concessional in nature.[2]
While changes in developmental aid from year to year
likely have a number of causes, what role does economic growth in the donor states play? In this study, I set
out to uncover to what extent the variance in development aid is an effect of
the past (or present) variance in GDP.
While it may seem obvious that economic
growth might have an effect on the aid that a country provides to developing
states, on closer examination other the picture becomes more complicated. Policy
makers could take a number of different factors into account when deciding how
much aid to send to other states. States could commit to providing a certain
amount of aid per year, regardless of domestic economic conditions. On the
other hand, changes in domestic politics could effect significant policy shifts
in the foreign aid budget. A quick comparison between GDP and
ODA, leaving changes over time to one side for now, demonstrates a positive relationship between the two variables, as shown in Figure 1.
To attempt the estimate the effect of
economic output on rich countries’ decision to release aid monies to developing
states, I employed a cross-sectional time series model that looked across time
and countries for evidence of an effect. Based on the availability of data for
inflation-controlled variables, I focused on the period from 1990 until 2012.
Although ODA is broken up into a many categories I zeroed-in on the aggregate of
all of them, measured in inflation-adjusted 2013 dollars. I also logged both GDP and ODA to estimate a linear relationship.
The OECD provides data on foreign aid
as defined as ODA dating back to 1968 for the members of the Development
Assistance Committee (DAC). While the OECD is a group of mostly wealthy states,
the DAC states focus specifically on promoting development. I therefore decided to
focus on current DAC member states 28 in total, excluding the EU, which is also
a DAC member but not a state. This core group of 28 states does not include all
states that provide large amounts of state-to-state developmental aid.[4] Verily, the OECD is so
skewed towards Western Europe and North America, my model could not account for
regional differences because too many of the states shared the same region.
Still, over the time period in question, the second half of the Twentieth
Century, this sample represents a majority of the worlds wealthy donor states.
Although I originally intended to estimate this effect for a longer time
period, inflation controlled GDP data were not available for these states
before about 1990, so the time frame for this study is from 1990 until 2013.
This also coincides with the fall of the Soviet Union and the addition of a
number of post-Soviet states to the OECD.
To
estimate the effect of variances in economic growth on the changes in foreign
aid, I employed a cross-sectional time series regression model with country
fixed effects, the results of which I present in the table below:
.tiff)
My model attempts to predict current levels of ODA from the
individual interactions of countries’ GDP and a number of control variables.
Importantly, although it seems implausible that ODA outlays could cause changes in GDP, I attempt to
control for such reverse causality by regressing current foreign aid on the
prior year’s GDP.[5]
“Lagging” GAP by one year in this fashion also allows me to test the
proposition that it takes policy makers some time to recognize changes in the
real economy and react to them by changing ODA. My model attempts to control
for variances between states in a number of ways. First, it employs
country-level fixed effects, which is a process that attempts to strip out anything
that is constant across time within units from the analysis. The decision to
employ fixed effects instead of random effects here was also based on the result of the Hausman Test (p =
0.00). The model also attempts to control for changes in domestic politics, a
potential confound, by including two variables: one representing the left,
right, or center orientation of the chief executive of a state, the other representing
the political orientation of the leading party in the legislature. Finally, as
mentioned above, the model also includes control variables for longer term
differences between states that might be confounding variables, namely, the
region of the world and a country’s history as a post-soviet state.
In conclusion, based on the above results, foreign aid and gross domestic
production are highly correlated correlated over time. Moreover, the analysis demonstrates that this correlation
exists in a model where foreign aid could not have caused GDP, eliminating the possibility
of reverse causality. A one unit increase in the previous year's GDP is correlated with a 1.871 unit increase in the amount of ODA, when political orientation of the executive an leading party in the donor state is held constant. A similar effect exists and is also statistically significant at the 0.01 level with GDP and ODA from the same year. Problematically, although the model was able to control for the political orientation of governmental leaders, [6] it was not able to control for differences between regions or post-Soviet status, and left those variables out of the regression. Although these results were statistically significant at a high level, meaning that there is a low probability that they were due to chance, it is not possible to conclude that policy makers only use GDP in making determinations about foreign aid. The results however do indicate that it may be likely that government officials do respond to changes in the economy when making decisions about foreign aid.
Appendix:
Alternative models:
.tiff)
[1] http://www.oecd.org/dac/stats/
[2]
http://www.oecd.org/dac/stats/officialdevelopmentassistancedefinitionandcoverage.htm
[3]
http://www.oecd.org/dac/stats/development-aid-stable-in-2014-but-flows-to-poorest-countries-still-falling.htm
[4] This includes large states like
China and India, which were not relatively wealthy at beginning of the time
frame but are now relatively much wealthier.
This
also leave out states that the OECD does collect ODA data for although they are
not DAC members. For these states, (Bulgaria, Estonia, Kuwait, Turkey*, and the
UAE) ODA data was generally also not available for a long enough time period.
[5] Thus the GDP in 1995 is used to
predict ODA in 1996, under the assumption that ODA changes in 1996 cannot have
cause variance in GDP before those changes took place.
[6] In the dummy variable regression for governmental orientation one variable must be committed from the right, left, center strategy. I elected to set "Center" to the control.
No comments:
Post a Comment