Last week, my blog post concerned Chapter 4 of Dubner and Levitt’s Freakonomics, which discussed the affect of legalized abortion on decreased crime rates.  This was a very interesting chapter, as it compared the affect of the novel idea of legalizing abortion with much more traditional thoughts such as increased policing, improving policing strategies, gun control laws, increased jail time, stronger economy, and decreased drug trade/trafficking. 


This week, I was given a research article by Donahue and Levitt regarding The Impact of Legalized Abortion on Crime.  This article is eerily similar to the chapter in Freakonomics,since it also makes the argument that legalized abortion was the most significant factor in reducing crime in the 1990s.  They discuss the idea that the first five states that legalized abortion in 1970 saw the first major drops in crime.  Eventually, the rest of the nation followed suit after case of Roe v. Wade in 1973.  The two conclude from their findings that up to 50% of the drop in crime can be attributed to the fact that abortion was legalized.  Other aforementioned factors such as a stronger economy and better policing strategies were merely dampening crime, and not having nearly the same effect that the legalization did.  Donahue and Levitt believe that their statistical analysis proves a causal link between legalized abortion and decreased crime, based on historical data.  The argument is that since legalized abortion essentially reduces the amount of babies born into poverty/ crime-ridden household situations, the groups or “cohorts” for criminal activity are reduced–there are fewer people on the streets who are likely to commit (violent) crimes. 


Now, while this article and the Freakonomics chapter both seem to coincide quite well in thought and ideology, there was also a critique of their findings that was produced by Foote and Goetz.  The criticism is data-specific, and shows that the errors made by Donahue and Levitt prove the effect to be less significant.  They argue that the data was not specific enough, and a better per-capita specification should’ve been used. However, if this is corrected, the results suggest a much weaker correlation between legalized abortion and crime rates.  Furthermore, they suggest that variation within the state was not accounted for like it should have been.  Robustness should have been taken into account to correct for differences among the trends, state by state.  


The Freakonomics chapter and the original article can surely be reconciled; they are making the same point more or less. However, the difficulty lies in the critique, which provides specific error-specific corrections that apparently need to be made to improve result accuracy.  Can they be reconciled? Perhaps.  However, if the regression results turn out to be much less significant than the originals, a new data set may need to be used to try and obtain a more accurate result. Otherwise, Donahue and Levitt may need to rethink their idea that legalized abortion had a significant and causal affect on the decrease in crime.


Chapter 4 focuses on the sharp drop in crime rates in the 1990s, which was caused principally by legalization of abortion.   Furthermore, it examines other explanations that have been offered for the crime-drop, including innovative policing strategies, reliance on prisons, and so on. This phenomenon due to abortion comes from the idea that most children who were born before legalization to poor women (who would’ve had abortions if able) turned out to be the criminals causing problems.


Some of the criticisms I would have deal with where the data has been pulled from regarding crime statistics (especially specific crimes) as they compare to things such as innovative policing strategies, tougher gun-control laws, and etcetera.  Where were these stats taken from? And how accurate is the measurement of “innovation” in policing strategies.  Furthermore, although the majority of crime is in urban areas, the chapter does not show any  type of studies regarding rural crime.


I think on the whole this is a very interesting chapter and one that would interest.  After all, a drop in crime would be of interest to almost anyone.  However, I think despite these different crime-drop explanations, there are many exceptions to each of them.  Most of the explanatory variables are statistically significant in explaining the outcome of crime-drop, but are specific to a certain geographic area, demographic, and etcetera.   This means that it is tough to make one broad, sweeping statement about what does cause crime-drops in the country.  The fact remains that it is many factors aggregated together–not just one or two.

For my research paper, the topic that I chose to explore was the following: does net FDI significantly influence net income in Sub-Saharan Africa?  I chose this topic because it correlates with another course I am taking in the economics department regarding African economic history.  My data set comes from the World Bank, which provides a very large database regarding African Development Indicators.  There are a multitude of different data sets to choose from and analyze to decipher which are most pertinent for my particular research.  Although I am starting with a single regression of net FDI as a percentage of GDP on net income as a percentage of GDP, this test alone most likely will not suffice. I will need to incorporate several dummy variables into my model in order to improve accuracy. Furthermore, other variables will need to be added in order to full account for changes in net income in Sub-Saharan Africa.  One of the biggest challenges that I will face includes the issue of having endogenous variables. Also, since many countries in Sub-Saharan Africa have informal economies, their statistics either aren’t provided for certain years, or they are simply not very accurate.

This chapter discusses how population policy and control is such an integral part of the development programs in poor countries.  The authors, through research, find several different factors that account for the problems of population control, including family dynamics, social norms and economic implications.  The argument presented goes against the traditional school of thought saying that a lack of access to contraception is the reason for poor population control.


The stat that I will use to make a hypothesis is the fertility rate of Ethiopia, which is 6.12 children/women.  I would like to hypothesize that the higher the fertility rate in a country, the more poor the country will be.  I would test this by compiling stats including GDP for different countries, average family income, and dummy variables such as whether or not there is access to health care of any kind (yes, D=0, no, D=1), and perhaps whether it is a single parent family or not. The regression model would obviously account for all of these variables, including the two sets of dummy variables for health care access and single parent family.  So, my dependent variable would be GDP, with fertility rate, avg family income, and then health care and family status all being independent variables.

These dummy variables, especially whether or not there is a single family, could be very significant in determining an effect on GDP and fertility rate.  Health care may not have as big of an effect, especially since many people in poor countries probably do not have this access. 

In Brad Pitt’s latest movie, “Moneyball” one can see how Billy Beane (Pitt) went against the grain as an MLB General Manager in terms of forming his team.  The Oakland Athletics are a very small-market team when compared to the household names such as the Boston Red Sox and the New York Yankees.  Instead of using traditional methods to figure out which players would be best to replace the stars that the A’s had to release (due to money and trade restrictions), he relied upon the statistical formulas and methods written about by Bill James, a baseball historian and statistician.

The idea behind Beane’s decisions was based on the idea that wins are produced by scoring runs.  Runs, as he quickly realized, were not just simply produced by big name, expensive, home run-hitting players.  Instead, the same amount of runs that were scored by these heavily-valued players he found could be produced by highly undervalued players that had high base on balls, on base percentages, and etcetera. With these things said, variables such as age, throwing defects, off-field problems, and EVEN fielding in general should NOT be significant in determining whether to pick a player or not.  Of course, these variables can all have some sort of effect. However, what is statistically significant are the stats that are compiled to figure out the runs created by certain individuals.

An example of Beane’s method from the movie occurs in a meeting decided how to replace Jeremy Giambi, a big hitting first baseman.  While the men at the table are trying to find an exact replacement for Giambi (a type of player that would be hard to find and too expensive), Beane suggests that they instead find 3 guys for less money that, when aggregated, have their on base percentages equal to that of Giambi’s.  This, in essence, reproduces Giambi.

An important factor when discussing all of this is the sample size.  It would probably take a season or two (there are 162 games in a season, excluding playoffs) to see a statistically significant increase in wins due to run production.  There are also many other factors that would be involved.  However, Beane’s logic based on Bill James’ work changed the way baseball teams ought to be managed, especially considering the amount of money each team has to work with.


The article I decided to review for this post is entitled, “The Role of Exports, FDI and Imports in Development: New Evidence from Sub-Saharan African Countries.”


This study was very interesting and presents thorough statistical analysis to argue its case.  The study begins by addressing the history of Sub-Saharan Africa (SSA) and its economic struggles in the 1980s.  The text then continues stating that several knew policies were enacted or at least proposed to help solve this problem, which included the halting of exports and FDI in SSA.  After conducting research, the authors of the paper promote “market-oriented policy changes” that essentially liberalize trade and deregulate laws that will allow increased FDI and export growth to stimulate the economies in SSA.


My specific project is concerned with FDI and its affect on net income in SSA.  With that said, I think this paper is extremely relevant because it is stating that once market policy changes are put into place to increase exports and FDI specifically, economic growth will occur. With economic growth, one would hope that jobs would be created, the labor force would increase, and income would increase as well. 


The only suggestions that the paper makes with regard to potential issues regarding the classic linear regression model is in reference to unbiasedness or random sampling.  The only way these two properties could potentially be violated is if the data is too lopsided into one country or region.  Although the study is for all of SSA, oftentimes the data (or at least the perception of it) can be skewed due to the extreme disparity between countries such as Somalia and South Africa.  This just reinforces the fact that when I run my regression, I need to account for multiple variables and errors such that my data reflects as accurately as possible the true effect of the independent variables upon the dependent variable.


A link to the article can be found here:




The article pertaining to my research is entitled “South Africa 2011 FDI More Than Triples, Africa Overall Down.”

The link to the article is here:
This article, although succinct and not overly detailed, was very useful for me.  As the title suggests, FDI in South Africa (a country in Sub-Saharan Africa)  more than tripled last year, which is a very good statistic for the country.  However, what is more important is the fact that on the whole, the continent of South Africa received less inflow of products than usual.  This distinction is important because my research is based on all of the countries associated with Sub-Saharan Africa in order to take a more general position on whether or not net FDI affects net income.
The article continues and highlights how investor confidence is starting to slip in South Africa, a country that has been experiencing unemployment spikes and a widening gap between the rich and poor.  It is scary to think that they are worried about South Africa, because compared to other countries in SSA (Sub-Saharan Africa), South Africa is doing much better economically.  If confidence is on the downgrade in South Africa, it’s hard to imagine if there is any confidence at all in countries such as Somalia or Sierra Leone.
What is most important for me as a takeaway from this article is the realization that there are MANY errors and extra variables that I need to take into account when trying to determine the correlation between net FDI and net income in SSA.  Countries are so different in terms of population, growth, development, civil strife, and etcetera. With that said, I will need to account for these inconsistencies in my data report before I run the appropriate tests to make a statement about SSA as a whole.  This will be a good amount of work but I think I’m up for the challenge.