John Donohue and Steven Levitt’s paper The Impact of Legalized Abortion on Crime can be viewed as a more scientific version of chapter 4 of Freakonomics. The paper offers statistical analysis of regressions run with the data mentioned in Freakonomics. In essence, Levitt simplified his paper in order to make it accessible to the public by having the readers accept without any evidence the conclusions made in Freakonomics. By leaving out the technical details in his book, Levitt allows the reader to believe his methods are perfect and for readers to justify explanations in their own way. This may have been a good strategy, as when the methodology and evidence are explained some problems arise.
Christopher Foote and Christopher Goetz identify a few of the unrealistic conclusions found by Donohue and Levitt by exploring variable bias in the original regressions. One case they make is that data should be run on a state by state basis. From these analyses, conclusions can be made about the entire nation, but it avoids the problem of heteroscedasticity that would arise from performing a regression on the nation as a whole. If the error on the number of arrest could be explained by which state the statistic came from, the variance of the error term would not be random.
Foote and Goetz also point to a problem with the dependent variable, saying that arrest per capita would be a less bias indication on if abortion was effective in preventing crime. Everyone accepts that abortion lowers the nation’s population, and unless only children destin to not be criminals were being killed, this would lower the total number of arrests. Therefore, even if aborted children were no more likely to become criminals, Donohue and Levitt’s model would show that abortion helped to stop crime in a significant way; however, stopping crime in this manner is not the conclusion Levitt comes to in Freakonomics.
When Foote and Goetz rerun the regressions replacing arrests with arrests per capita, the numbers are less significant. I claim that this does not disprove the conclusions drawn by Donohue and Levitt. In Freakonomics, the innovative and more successful policies of policemen are noted as a obstensible reason for the decrease in crime, but better police work should lead to more arrests per capita. Because of this, abortion’s possible downward push would have been negated by the upward push of the stronger police forces.
I do not find critiques particularly interesting, but I would like to see a response from Donohue and Levitt to this one. The points in the critique, while they may be over analyzing, deserve explanation, and once a back and forth is established, the intellectual banter becomes a great read.
Chapter 4 in Freakonomics by Levitt and Dubner is about the increasing crime rate in the United States to the point of epidemic proportions and then its ‘unexplainable’ drop. They hypothesize that the ruling in Roe v. Wade, legalizing abortions, was responsible for the drop as children who were more likely to become criminals are more likely to be aborted.
Levitt and Dubner make strong points on the probabilities of a child becoming a criminal. Low income families, families with only one parent, or young parents are far more likely to have a child who chooses a life of crime. These mothers are also more likely to have an abortion, as they may feel they are not ready for a child or cannot provide for one.
I would like to ask the author his opinions on the explanations given by economists of the time. I am not entirely convinced that abortion was the only explanation for such a drastic drop in crime. In the chapter a couple opposing viewpoints are explained and critiqued, but most are not given much more than a mention.
I am suspicious how all other economists and researchers looking into the topic missed this connection. If the crime rate was rising as rapidly as the book states, the issue would have become one of the most widely researched in America. If the sudden drop in crime rate was completely unexplained, more people would have been intrigued and looked into an explanation. Roe v. Wade was a Supreme Court case; therefore, its results and implications would have been well-known, and I find it hard to believe no researchers at the time made the connection.
Hello there, nice music we’re listening to. Since we’re stuck together in this elevator and you cannot escape, let me tell you about a paper I’m writing.
Terrible, bloody wars have grabbed news headlines, in particular conflicts in countries with an abundance of natural resources. Many countries with resources such as diamonds or oil have been exploited by more developed countries or international corporations, but the developing country still receives some economic payment for its natural resources. While wars and other conflicts occur in places all over the world, the added economic incentive of natural resources may lead to an increase in the bloodiness of a conflict. In my paper the correlation between specific natural resources and the total number of deaths, both militant and civilian, is explored and explained. Although it does not test for causation, it is not a stretch to see how natural resources would affect factors that lead to bloodier wars, for example better weaponry and aid from more developed nations. Using statistical analysis leads to some interesting conclusion about how natural resources are related to violence within conflicts and it has far-reaching implications in foreign policy.
George DeMartino’s talk at Gettysburg College was one of the best talks I have been to in my time as an undergraduate, and I am thankful to the department and Professor Madra for setting up and scheduling the talk at a time I could attend. While I do not personally agree with his opinions, I recommend seeing Professor DeMartino talk or reading his book, The Economist’s Oath: On the need for content of professional economic ethics, for the discussions each creates and his skill as a presenter.
DeMartino’s main argument is that the neoliberal revolution led by economist such as Jeffery Sachs was unethical because of the initial harm their recommendations did to the people of developing countries. Whether or not their policies are unethical is a heated debate, but this post will not deal with those issues. DeMartino’s argument, although it had statistics appearing to support his claims, was not very econometrically oriented. Below, a possible econometric model is offered for how he could have made his talk more statistically based, but the drawbacks of the models are also discussed.
First, let us look at the immediate harm felt by countries directly after enacting the policies recommended by American economists during the neoliberal revolution. DeMartino referenced the increase in male deaths in many of these countries as evidence the population was hurt by the policies. He claims economists need to be careful not to over discount the lives of current populations in order to hopefully improve the future. Since he wants to look at male deaths or the increase in deaths after the creation of the policies, let the natural log of male deaths be the dependent variable.
A dummy variable for whether or not a country followed the policies recommended to them should be used, in order to control for an exogenous increase in male deaths throughout the developing world. The ideal data set would have countries near those that choose to enact the policies, as they would likely have similar conditions. Comparing countries enacting the policies with countries from more developed places such as Western Europe would lead to inaccurate conclusions. It might even be more beneficial to create three more dummy variables for Eastern Europe, Africa, and South/Latin America in order to control for geographical location.
Another aspect to consider, as brought up by Professor Baltaduonis, is the policies were enacted, at least in Eastern Europe, around the time government were moving away from strong military presence. The decrease in military created an opportunity for gangs and mafia to take control of the streets, creating a more dangerous environment. If possible, a variable for the change in the size of military and police force would be ideal, but a dummy variable representing a change in military and police enforcement could explain a good amount of change in male deaths. Of course, this would require setting a level of change to warrant being represented by the dummy variable.
One variable often overlooked when viewing deaths or increase in crimes, is the introduction of a new drug or a sharp influx of drugs into a country, one example being the invention of crack cocaine in the United States and the effect it had, especially on black communities. Hoping the countries have drug control policies or records, a variable relatable to increased male deaths would be the natural log in the estimated amount of drugs within a country. While it is possible the increase in drugs could be caused by changes in policy, tests for correlation between the two variables could be a deciding factor on whether or not to include it in the model.
There are some variables that originally seem like good fits for the model, but would actually dilute the statistical significance of the relevant variables discussed above. Change in GDP, unemployment, or the number of doctors could be viewed as reasons for an increase in male deaths; however, both of these are strongly correlated with the policies recommended during the neoliberal revolution that advocated for privatization. Adding these variables would result in multicollinearity, weakening the model.
Running the regression, I would expect to find that enacting the recommended policies did result in an increase in male deaths, but not as drastically as Professor DeMartino leads listeners to believe. Changing from state-run jobs to private sector jobs can be a tough transition with less job security and a higher level of competition. These factors would likely lead to higher unemployment initially, resulting in more desperate citizens more likely to take risks. The real question is, how many people should governments be willing to sacrifice today in order to increase the likelihood of a better tomorrow?
Chapter five in Poor Economics, is about the family size of poor families, and the effect population has on wealth. While most agree that smaller population leads to a wealthier nation, keeping family size small is hard in underdeveloped nations. Women do not have the bargaining power within the family, the education, or occasionally the desire needed to shrink family size.
In some cultures, children are used as a financial safety net once the parents age, so having more children increases the likelihood of one of them being about to support the elderly parents later in life. A statistic introduced by the authors states “in China, for example, more than half the elderly lived with their children in 2008, and that fraction increases to 70 percent for those who had seven or eight children” (Banerjee and Duflo). This statistic implies the belief of parents is correct, more children creates a higher probability they will be cared for in old age, but are there other factors at work?
A family with more children must have been able to afford those children, especially in China where parents are obligated to pay a dowry when a daughter gets married. Wealthier families are more likely to value education and produce wealthier off spring. The wealthier children then have an easier time caring for their parents and may feel more obligated to provide for them, in particularly if parents favored one child with education and monetary gifts.
Also, large families were more common in the past, meaning families with a multitude of children are more likely to be traditional. In traditional China, extended families lived with each other, so there is more of an expectation for the children to take care of their parents.
In order to test the implication that having more children causes parents to be more likely to live with and be supported by their parents I would make a regression using a linear probability model. The dependent variable would be the dummy variable of if the parents live with their children in old age or not. Obviously the sample would have to be limited to families with elderly, living parents. I would use household income as an independent variable to measure family wealth, but seeing if a family was traditional or not would be trickier.
While families vary in how traditional they are, there must be a traditional Chinese event less celebrated in recent years. With some research into what such an event could be, I would introduce a dummy variable as to if a family partook in the traditional event. I expect this dummy variable to tell me if a traditional family is significantly more likely to have parents live with their children in old age. I would know if it was statistically significant by the t-statistic of the coefficient on the dummy variable.
All together, my regression would look as follows:
Livewithchild= α + β1Income + β2 Traditional + β3NumChild + ε.
Income and the number of children a family has normally effect on the margin, so running them as the natural log of their value may yield better results. After running a few regressions, I would be able to conclude if I can reject or not reject the null hypothesis that the number of children does not affect the probability parents live with their children in China.
Michael Ross’s paper A Closer Look at Oil, Diamonds, and Civil War, focuses on the mistakes other papers have made when looking at the correlation and causality between natural resources and civil wars. He offers a new method in an attempt to correct the previous mistakes made in the prevailing literature on the topic. In the end, his findings are consistent with the conclusion reached by researchers using the methods Ross sees as flawed.
This paper is extremely helpful for my research project. Looking at the control variables used by Ross, I saw variables I would not have considered putting into my regression. It also shown light on the fact that I may need to combine more data sets in order to obtain this information, but GDP and population should be easy to find. Of the problems pointed out, my paper will avoid many of them because I am not looking at the initiation or duration of a war; however, new problems may arise not mentioned in this paper when considering bloodiness.
One of the largest flaws pointed to by Ross is the use of dummy variables for having natural resources. The amount of natural resources in a country is a continuous variable, so modeling it with a dummy variable is inherently inaccurate, especially when an arbitrary resource level is set to decide if the country ‘has’ resources or not. The data I am using has actual numbers, so my regression will use a continuous variable to model amount of natural resources. The main reason being for not using a dummy variable, looking at this topic from the perspective of an economist, I would like to see how higher economic incentive (more natural resources) influences how bloody a war becomes. One place where a dummy could be useful in respect to natural resources is the distinction Ross makes between hard to reach resources requiring significant investment, and resources able to be mined by unskilled labor.
It appears the model used by Ross and the models he critiques are fairly good at following the assumptions of the classical linear regression model, except for the correlation between variables. A year is more likely to be bloody if the year before it was bloody. One way to correct this may be to look at wars as a whole, average the statistics of the entire war, and use those averages in the regression. Possibly, sums could be use instead of averages, but this would favor wars that last longer. That may problem because some of the studies cited in Ross’s paper have found that countries with natural resources fight in conflicts significantly shorter than countries without the resources.
Ross’s paper has been helpful to me in many ways, even in the early stages of my research. I am most excited for the insight it will provide in the later stages. An entire section of the paper is dedicated to causal mechanisms used by him and other authors who wrote on similar subjects. This will be extremely useful when I attempt to make inferences from my research.
Ross, Michael. A Closer Look at Oil, Diamonds, and Civil War. 9 Vol. , 2006. Print
The article “Can Africa break its ‘resource curse’” paints a hopeful picture of African countries learning from past mistakes dealing with resource management. Countries that have recently discovered oil or diamonds have successful, mineral rich countries such as Botswana and Ghana to model themselves after. This may be wishful thinking, as leaders of those same countries also have models such as DR Congo, Liberia, Ivory Coast, and Sierra Leone, where bloody conflicts resulted from the natural resources of the countries.
The article also mentions the hazards associated with mineral extractions. Large scale accidents, such as oil spills, can decimate large groups of peoples’ livelihood. Creating large groups of poor people, especially ones whom feel entitled to direct profits from resources, could feasible lead to conflicts. The poor highly discount the future, and this leads to more rash decisions and potentially more violence. Number of accidents or a dummy variable that controls for a large accident occurring just before a conflict could be helpful in my proposed regression for my topic. Additionally, the opportunity for accidents that directly affect people who live off the land may be why literature has found significantly stronger links between conflict and onshore drilling than conflict and offshore drilling. Finding a data source that accurately records accidents would provide interesting insight, but the possibility of one existing is small because the large corporations running the mineral extraction do not want the bad press, which is a topic too deep to delve into now.
If you are interested in that topic I recommend looking at WikiLeaks; here is an article to get you started that I have used in the past.
In the movie Moneyball, the protagonist Pete (Jonah Hill) uses both time series and cross sectional regression and analysis. He uses the abundance of statistics in baseball to his advantage, and appears to run a kitchen sink regression with certain variable weighted more than others. This model results in a singular number, the predicted value a player adds to his team. Pete takes the results of the regression run on each individual player in the league in order to find players undervalued by all other teams in the league, allowing a low budget general manager to build an above average team.
The question Moneyball focuses on more in-depth is whether or not using purely numerical analysis is a complete analysis of baseball players. While the movie paints the picture of the model being nearly perfect for the A’s and the Red Sox, it is unlikely the model by itself would be sufficient to build a championship team (as the A’s found out in the playoffs).
The application of a strictly numerical approach to rate players does have merit, in particularly to players with experience in ‘the show.’ Older players have demonstrated their worth in the major leagues, and their true worth is hidden in the statistics they have accumulated throughout their career. Establish players do not significantly deviate from their career statistics without signals being apparent (at least in hindsight) within their numbers. A real world example of this is Curtis Granderson, whose terrible splits between left handed pitchers and right handed pitchers indicated a problem seeing and identifying pitches from left handed pitchers. The Yankees’ hitting coach saw this and corrected Granderson’s batting stance, resulting in huge returns for the Yankees.
The numerical analysis falls short in aspects described by the scouts as intangibles. Although scouts may exaggerate the importance of these intangibles, a team needs leaders and pressure performers, aspects not incorporated into statistics. The movie touches on this when Billy Beane, the team’s GM, confronts one of the team’s older players, David Justice, and convinces him of the importance of setting a good example for younger players. Leadership creates a higher level on confidence in a team, and helps players perform better in important situations; a reason young teams normally perform poorly in the playoffs. The 2002 Oakland A’s are a good example of a team that folded under the pressure of a playoff appearance because of being such an inexperienced team.
The most blatant disregard for intangibles in the movie happens when Pete tells Billy he would have drafted Billy in the ninth round. No computer run regression can predict the psyche of a player, especially a player in high school. High school leagues vary from region to region and year to year, so the numbers a player posts in high school are nothing more than a rough indication of a player’s skill against an ever changing competition level. The pressure of high school ball is incomparable to the major leagues, and mental blocks are unpredictable if the player has never been in a comparable situation. The Red Sox may have used aspects of the team designed proposed in the movie, but one can be certain the management of the championship team understood the non-numerical side of the game as well.
In chapter 4 of Poor Economics, Banerjee and Duflo focus on the education problems in most developing and underdeveloped countries. The arguments of experts who believe more schools are the answer against the argument of a need for the market to demand more schooling are explored. While children do need schools in order to learn and are unlikely to travel to school if it is a significant distance away, schools are a waste of government resources if the students and the teachers do not want to learn or teach respectively. The authors also examine parents’ views of education and problems that stem from their beliefs, such as a belief created education trap or the choice of which child, if any, gets to attend school.
Jim Belshaw’s blog post The education trap, takes a view of the education trap mentioned in Poor Economics from the perspective of a developed nation, more specifically Australia. Instead of the education trap being in the minds of parents deciding on whether or not to send their children to school, in Australia the trap has to do with lack of real world learning. One of the reasons parents gave to researchers in developing countries about why they did not think school was worth their children’s time was that no ‘practical’ skills were taught. They did not think it was important to learn scholarly subjects if their child was destined to become a farmer or other trade laborer. Belshaw appears to agree with them, for he views going to medical school as a waste because he learned nothing of how the world operates. He argues that in a developed country understanding a cash economy is more useful than the majority of subjects taught in schools, similar to learning trades being more useful in the eyes of parents in less developed places.
In addition, Belshaw quotes an article mentioning the failure trap in the educational system of Papau New Guinea. The system in that country decides if students are permitted to continue their schooling past grade 8 and then loses thousands of students in grades 10 to 12. As previously stated, these students are not being taught how to survive in their country’s economy, only the ideas of scholarly subjects. For the students not allowed or unable to continue their education, this is especially problematic, because their time spent in school up to that point could be used to learn more useful skills that would help them in the workforce. Banerjee and Duflo also examine school systems that make decision about students, usually based on their performance on standardize test. They point to the flaws such as poor teachers causing students to perform poorly, or students not learning as much as they should because all lessons are focused on passing the test instead of learning. More interestingly, statistics indicating student psychological state may affect these tests instead of entirely how intelligent a student is, like the case of students in a lower caste or a gender expected to perform poorer on the test.
I find the arguments in Poor Economics more convincing, but that may be because I trust statistics more than a person generalizing for a nation from personal experience. Although Banergee and Duflo warn against a paternal view of developing countries, I am more likely to excuse their misunderstandings of the educational system or trust their belief that learning practical skills is more important in their community. People unhappy with their degree in developed countries made the choice to earn that degree and had the resources to make an informed decision about what degree to pursue. Most degrees are marketable in a developed country, and it is up to the person to make it known to employers that they are a desirable employee.
Data Set: UCDP/PRIO Armed Conflict Dataset and CSCW Geographical and Resource Database
Thesis: When one of the spoils of a war is land with an abundance of valuable natural resources such as diamonds or petroleum, the fighting is bloodier than in wars for land with less abundant natural resources.
Many people have seen the movies Blood Diamond or Lord of War, and have had a Hollywood introduction to the horrors associated with wars in countries where valuable resources are plentiful. I am interested in how these resources relate to the bloodiness of the wars. First, the term bloodiness must be defined and there are many options at my disposal. Bloodiness could mean the total casualties in a war, the number of civilian casualties, or even the average number of casualties per battle. All have inherent problems, but provide a possibility to provide insight into the motivation of the army leaders.
If a strong correlation is found between valuable resources, my data will focus on the resources diamonds and petroleum, and bloodiness in wars, there are numerous policy implications. These implications include allocation of relief funds and use of foreign armies in local wars. A land with substantial natural resources can already request military help from more powerful countries in need of their resources, but may not receive equivalent relief aid after the war. Furthermore, a correlation could provide insight into the mind of the leaders of countries in developing countries who are normally portrayed as crazy and bloodthirsty, because they may simply be cold, calculating leaders motivated by economic incentives.
As previously stated, the definition of bloodiness could influence the strength of the correlation. Looking at total casualties would favor longer wars, even if each battle within the war was not particularly violent. On the other hand, deaths per battle would weigh current battles heavier because advances in weaponry have made warfare more deadly. Civilian casualties could destroy any correlation as genocides would have the highest civilian death rates, but may have no resources serving as motivation for the war. Luckily, the data for all these definitions of bloodiness is available and running regressions could tell which definition is closest related to the amount of resources in a country.
There are some difficulties immediately apparent in finding such a correlation. One being the location of the data is clustered in Africa, Southeast Asia, and the Middle East, so any implications might be limited in their scope to that section of the world; however, this would still be helpful to policy makers. With the dataset dealing with armed conflicts being so large, narrowing the scope to any one of these reason could still result in enough observations to yield a significant correlation. Many countries have connections to countries all around the world, and a country’s allies could affect how bloody a war is if the allies are supplying the country with modern weaponry. Hopefully, the world is interconnected to the point where all countries have powerful allies, especially the countries with excess natural resources, that the effect of those allies will be negligible.
nateecon350 