EE625 : ECONOMETRICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction ◮ Economics concerns with relations among economic variables. Econometrics concerns the analysis of data describing economic relationship. ◮ We may also ask by curiosity whether a change in one variable ( x ) causes a change in another variable ( y )? ◮ Examples of economic question: ◮ What is the effect of school spending on student performance? ◮ Does having another year of education cause an increase in salary? ◮ What is the effect of registering debt collectors on low-income debtors? ◮ Does reducing class size cause an improvement in student performance? ◮ What is the effect of prohibiting political campaign on voting outcomes? ◮ What is the effect of minimum wage on unemployment? ◮ and so on... ◮ These questions have something in common. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Causality and the Notion of Ceteris Paribus in Econometric Analysis ◮ What is a causal effect of x on y ? ◮ For example, x could be ”institutions”, y could be ”economic development”, or x could be ”schooling”, and y could be ”wage” ◮ Suppose x is correlated with y , can we interpret this relationship as causation? ◮ Consider the following story: in 1988, someone conducted a series of interviews with freshmen and found that those who had taken SAT preparation courses scored on average 63 points lower than those who hadn’t. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ The person then concluded that SAT preparation courses were not helpful. Is his/her conclusion valid? ◮ How can we isolate the effect of x on y , and quantitatively establish that x matters for y ? ◮ If we can run a controlled experiment, this may allow a simple correlation analysis to uncover causality. But this is rarely the case in economics. ◮ We generally must accept the conditions under which people act and the responses occur. Typically we cannot choose the level of a treatment and then record the outcome, but we can often observe people behavior as recorded in nonexperimental or observational data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Experimentation ◮ For example, how can we isolate the effect of institutions on economic performance, and quantitatively establish that institutions matter for economic development? ◮ Suppose we can conduct the following experiment: Pick 2 identical economies, holding all other factors fixed, change institutions of only one country, and then watch what happen to economic development of these 2 countries. ◮ Then we can convincingly attribute the difference in development paths to institutional change. ◮ Fortunately, we cannot do that. Instead, we can use econometric methods to effectively hold other factors fixed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ Experiments are conducted less often in the social sciences than in the natural sciences. ◮ Thus experimental data that are often collected in laboratory environments in the natural sciences, are more difficult to obtain in the social sciences. ◮ Although some social experiments can be devised, it is often expensive, or morally repugnant to conduct the kinds of controlled experiments that would be needed to address economic issues. ◮ What we usually have are nonexperimental or observational data. And econometrics has evolved as a separate discipline from statistics because it focuses on the problems inherent in collecting and analyzing nonexperimental economic data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ Causal effect refers to the answer to the following counterfactual thought experiment: if, all else being equal, but only x changes exogenously, what would be the effect on y ? ◮ Thus the notion of ceteris paribus which means other (relevant) factors being equal plays an important role in causal analysis. ◮ Answering such causal questions is quite challenging, because it is hard to hold all other relevant factor fixed. ◮ The key question in most empirical studies is: Have enough other factors been held fixed to make a case for causality? If some relevant variables are omitted, it is then difficult to isolate changes in endogenous variables that are not driven by omitted factors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ Econometrics is useful because if it is carefully applied, it can simulate a ceteris paribus experiment. ◮ For example, we might be interested in the effect of another week of job training on wages, with all other components being equal (in particular, education and experience). ◮ If we succeed in holding all other relevant factors fixed and then find a link between job training and wages, we can conclude that job training has a causal effect on worker productivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ Example: Measuring the Return to Education ◮ If a person is chosen from the population and given another year of education, by how much will his or her wage increase? ◮ This is a ceteris paribus question where all other factors are held fixed while another year of education is given to the person. ◮ If we can conduct an experiment to measure the return to education? How would we set up this experiment? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ One way is to randomly pick and assign each person an amount of education; some are given no education at all, some are given a high school education, some are given two years of college, and so on. ◮ Subsequently, we measures wages for each group of people. ◮ However, the experiment described above is infeasible. We cannot give someone only a high school education if he or she already has a college degree and so on. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ Even though experimental data cannot be obtained for measuring the return to education, we can collect nonexperimental data on education levels and wages for a large group by sampling randomly from the population of working people. ◮ Example of such data: the Current Population Survey (CPS), the Labor Force Survey (LFS). ◮ A common feature of many observational data is self-selection. Usually people choose their own levels of education. Therefore education levels are probably not determined independently of all other factors affecting wage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ If we settle on a list of controls, and if all factors in the list can be observed, then estimating the causal effect of x on y is quite straightforward. But some factors in the list may not be observable. ◮ For example, to estimate the causal effect of education on wage, we might decide that the relevant list to control for is years of workforce experience, and innate ability. ◮ Since pursuing more education generally requires postponing entering the workforce, those with more education usually have less experience. ◮ Thus, in a nonexperimental data set on wages and education, education is likely to be negatively associated with experience. ◮ People with more innate ability often choose higher levels of education. Since higher ability leads to higher wages, there should be a positive relationship between education and ability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
◮ As it is not difficult to measure experience, it is likely to have this variable in nonexperimental data set. ◮ So accounting for observed factors, such as experience, when estimating the ceteris paribus effect of another variable, such as education, is relatively straightforward. ◮ Ability, on the other hand, is difficult to measure. ◮ Accounting for inherently unobservable factors, such as ability, is much more problematic. Many of the advances in econometric methods have tried to deal with unobserved factors in econometric models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Types of Data ◮ Some econometric methods can be applied across different kinds of data sets. ◮ But some data sets might have special features that must be accounted for or should be exploited in particular. ◮ The most important data structures encountered in applied work are ◮ 1) Cross-Section Data ◮ 2) Time Series Data ◮ 3) Pooled Cross Sections ◮ 4) Panel Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Recommend
More recommend
