Charts showing data over time, called time series, are an excellent way to show trends and to make forecasts for the future. However, time series data may need to be adjusted before plotting to avoid errors in analysis.

First, if the data being displayed is in monetary units (for instance, dollars), then one has to be careful to take into account the effect of inflation. For short periods, omitting inflation isn’t a critical error. However, as the time scale increases, the error can become significant, especially if the time scale includes the 1970s and 1980s when inflation was quite high.

When looking at historical price or monetary data, the unadjusted raw dollar amounts are called nominal values.  The adjusted dollar amounts are called real. (One has to be aware that there are many different inflation adjustments available, but that is beyond the scope of this blog post.)

Second, when comparing country economic data such as gross domestic product (GDP) across multiple years, it should be corrected for population growth. Both the U.S. and the world population are much larger today than 50 years ago while the population of many European countries and Japan are stable or shrinking. Thus, when comparing the well-being of these countries, one should use changes in per capita real GDP. Notice that the U.S. nominal GDP must grow about 4 percent a year just to keep per capita real GDP constant.

Finally, when comparing spending on a particular good or service, the data should be adjusted for changes in household income, consumption patterns, and the quality of the good or service. This is often difficult to do and there are disagreement on the best way to do this. The biggest example is change in house prices. For example, the median house today is much larger than a house 20 years ago. Also, the quality of appliances and materials is much higher than 20 years ago. This is true for the median home and can be very expensive in high-end homes, which will cause the average to be skewed.

Consumption patterns are changing as income increases. The median U.S. and world household income is higher today that it was 50 years ago, even adjusting for inflation). However, much of this gain has been directed toward college graduates who work in jobs that are located in big cities on the east and west coast. This has driven the price of houses upward in cities like New York, Boston, San Francisco, and Los Angeles relative to houses in smaller cities away from the coasts. The interaction between income and house prices is hard to analyze separately.

Rising income also affects the composition of goods and services a household consumes. As a proportion of income, people spend less on food and durable goods and more on entertainment, education, and health care.

Example of a badly drawn chart and a corrected version

In January 2011, Bain & Company released a report on electronic publishing that included a warning that publishers should not repeat the mistakes of the music industry. The warning was illustrated the following chart.


The rise and fall of the music industry. Image from Bain

Michael DeGusta picks apart this chart in a Feb 2011 Business Insider article. He notes that the chart makes it look like the music industry grew steadily from 1973 a peak in 1999 and has since fallen about 40%.

However, he corrects the chart for changes in purchasing power and population growth as shown in the revised chart below.


The rise, fall, rise, and collapse of the music industry. Image from Business Insider

This revised chart is much more interesting. It shows that real per capita music sales peaked in 1978 and began to fall until CDs arrived. Sales reached a new peak in 1999, but then fell. Digital music sales have not been able to stem the fall and per capital sales in 2009 have fallen 63% (not 40%) from the level of ten years earlier.

Mr. DeGusta goes on to provide an insightful analysis of why sales of digital music have not been able to replace CDs. Read his blog for more details.