The graph above shows two series related to household debt that have received a great deal of attention lately: consumer credit (mostly lines of credit and credit cards) and student loans. These series show stark increases especially in recent years. But one has to be careful before jumping to conclusions, as the eye may be deceived here. First, the student loans shown here are only those that come directly from the federal government, and that specific program was introduced in 1994. So part of the increase is simply this program ramping up. But more importantly, one has to consider the important factors for the time period shown here: overall prices increased, population grew, and real incomes increased as well. Thus, it could be that these graphs simply show the increases in these three factors and nothing else.
To make things clearer, we need to divide by a measure that also increases along with these three factors and thus represents the size of the economy over the years. One popular candidate for this is nominal (that is, not real) GDP. It accounts for price, population, and productivity growth. The graph below is the same as the above, except that both series are divided by nominal GDP. The new graph still shows an increase for both series, but it’s not as dramatic. It also has the advantage of providing a frame of reference for the numbers: Total outstanding consumer credit currently amounts to about 20% of national income, and student debt is 6%. Whether this is excessive is open to debate. But one should focus on the data in percentages, not in billions of dollars.
How these graphs were created: Search for “consumer credit” and click on the desired series. Once you have the graph, go to the “Edit Graph” section and open the “Add Line” panel. Search for “student loans” and take the series with a longer time range. Apply formula a/1000 so that the units match. You have now the first graph. For the second, add a series to each line by searching for “GDP” (do not take real GDP) and apply formulas a/b and a/b/1000, respectively.
Suggested by Christian Zimmermann.