Analyzing the Last Five Years of the US Economy for an Intermediate Macro Course

The objective of this assignment is for students to build a correspondence in their minds between the macroeconomic variables that comprise standard theories of consumption and investment, and the real world values and trends in those variables.

• Students should be able to find the requested data series from a reputable website.
• In locating the data, students should be able to differentiate between nominal and real measures.
• Students should discover that the components of consumption and investment expenditure embody a strong time trend.
• Students determine appropriate ways of characterizing the data, e.g. average trend growth rates over specific time frames, peaks and troughs in the cyclical data.
• Students should be able to display data in both tabular and graphical formats, appropriately formatted and labeled.
• Students should be able to correctly and completely cite the sources of their data.
• Students should be able to come up with appropriate measures of explanatory variables, e.g. real personal disposable income for income, stock market prices, housing prices, or household net worth for wealth.
• Students should be able to produce reasonable conclusions based on correlations between dependent and independent variables about which theoretical models seem to fit which dependent variables best.

In my experience, there is a disconnect in the intermediate theory course between student understanding of the economy as characterized by standard macroeconomic models and faculty understanding of the same models. Students find it difficult to really grasp what abstract models are explaining. The problem isn't in the models, but rather in how students perceive them. One way to bridge this disconnect is to ask students to dig into the data for the variables the models are using, which is what this assignment asks them to do.

The assignment asks students to work in groups of three or four. Organizing this as a group project makes it feasible for class sizes as large as 50 students (or more with TA support). This assignment should be given after the underlying theoretical material has been covered in the course, that is, after the chapter on consumption expenditure or investment expenditure has been completed. The assignment can be done either by itself (either or both consumption or investment expenditures) or as part of a larger project which explores more sectors of the economy. Each assignment takes perhaps 10 minutes to introduce in class (less the second time), and 50 minutes to review afterwards.

Students are told that the purpose of this example is to give them initial experience in the type of applied data analysis done by professional economists. The analysis is not very sophisticated, and that's as it should be at this level.

This example is actually two assignments: the first on consumption expenditures and the second on investment expenditures. In the first assignment, students are asked to collect quarterly data over the last five years on the components of consumption expenditure, including consumer durable, consumer nondurable and service expenditures. In the second, they do the same for the components of investment expenditure: durable equipment and software, non-residential investment, residential investment, and inventory investment expenditures,

Students must first create tables and time series charts for each variable, as well as provide complete citation information for where the data was obtained. Next they must think about how to characterized each data sample. Should they use levels or growth rates? Should they identify changes in the trend, or turning points in cyclical data? Noodling this out is where working as a group can be particularly useful.

Next, students need to think about the theoretical models they have studied regarding consumption or investment expenditure. What variables should be examined for the simple Keynesian consumption function, the Life-cycle model, or the Permanent Income Hypothesis? What variables should be examined for the simple Keynesian model of investment, the Accelerator model, or the Neoclassical cost of capital model? What are appropriate measures of interest rates, or income, or wealth? Clearly, there are a variety of choices for students to make.

Finally, how are the explanatory variables correlated with the dependent variables? Are these judgments based on formal statistical correlations or visual approximations? How are lags taken account of in the analysis? What conclusions can be drawn about the ability of the different models to explain each of the dependent variables?

The assignment handout is [here].

Teaching materials needed include:

• Access to print or online data sources
• Spreadsheet software
• Display hardware and software for presenting the results in class

• Basing their analysis on incorrect data.
• Failing to differentiate between real and nominal measures.
• Many students have no idea about how to describe a data sample. They tend to think in terms of minimum and maximum values, or mean levels. With time series data, this isn't very helpful.
• Not thinking of growth rates as a good way to characterize data with time trends.
• Not understanding that data are noisy and that correlations between variables probably won't be exact.
• Not using a holistic approach to assessing models.
• Rejecting models because of an overly mechanistic approach to assessing them, e.g. the (contemporaneous) correlation coefficient wasn't strong, when there's an obvious visual correlation when lags are incorporated into the story.
• Occasionally, students have tried this assignment by downloading data directly into statistical software from which they estimate the models using regression analysis. They may think they are doing advanced work, but such an approach fails to give them a real sense of the data or its messiness, and often leads to poor regression results due to lagged effects, which they haven't accounted for. Regression is not a perfect substitute for examining the data with your eyes.
• Not understanding that data sets don't always use the same names for variables as theoretical models, for example, gross fixed capital formation is investment, change in business inventories is inventory investment, etc. Not thinking about data as having an organizational structure, which parallels the structure outlined in textbooks or class lectures. Thus, while the national income and product accounts (http://bea.gov/national/nipaweb/Index.asp), may give variables different names that the textbook author does, the structure is the same. This problem comes from the google mentality that students increasingly have.
• Groups that don't work together and therefore don't learn the pieces of the assignment that they didn't complete themselves.

• Are the data accurate? Do the tables/charts show the correct patterns in the data.
• Did the students choose reasonable measures of explanatory variables?
• Did the student characterize the data in a reasonable way?
• Did the student draw conclusions in a reasonable way? Did they correctly discriminate between alternative models?