Lengthy Economic Time Series - Their Use and Misuses.

Samuel Williamson, MeasuringWorth & Miami University

For the United States there exist three types of well used continuous time series that measure economic performance from the late 18 century to today. They are the CPI (consumer price index), wage indexes for skilled and unskilled workers, and GDP (the output index). Economists and others use these series for measuring economic growth and improvements in standard of living. Each of these series consist of over 220 observations and have been downloaded from the MeasuringWorth website between 2,000 to 10,000 times a year. My presentation would first discuss the “theory” that guided the definition of these three types of variables in the past and how understanding this is important when using them today. Examples of issues they might address are: How does the bundle underlying the CPI measures change over time? What does an hour of a type of work mean over 220 years? How are we counting total output when its underlying composition has changed so much and the share of output “going through the market” has increased? In contemporary periodicals and scholarly journals, when looking at historic amounts, there is often an attempt to differentiate between the nominal and the real value, or the historic and the current day value. Often with spectacular misconceptions, particular when the CPI is used to deflate more than a generation ago. The second part of my presentation would be to proposed set of “rules” as to how historical amounts should be presented in scholarly works with the recommendation that in most cases they would be more than one adjusted number. It would also be the suggestion that the profession should require that authors explain their choice of the relative values used the same we ask for statistical significance of the regressions used.

See paper

 Presented in Session 39. Problems with Data and Measurement