One of the challenges we've taken on this year is developing more real-time indicators of the health of the U.S. economy. One of the other challenges we've taken on is the determination of a consumption-based National Dividend - a concept for measuring the well-being of the people of a nation that was first proposed by Irving Fisher in 1906, but which couldn't be developed because it was too difficult to amass the consumption data needed to calculate it. At least, until we finally got our hands on the problem earlier this year!
The problem with making our national dividend calculation into a near-real time indicator of the economic health of the American people is once again the data. The Consumer Expenditure Survey that provides the data we need to calculate the national dividend is presented semi-annually. If we want to make it a near-real time indicator, we need that kind of data a lot more frequently.
That's where we realized that our observation that there is an almost perfect 1-to-1 correlation between a consumer unit's average annual consumption expenditures and the median household income would be an important discovery - we can use the monthly median household income data collected and published by Sentier Research as a reasonable substitute for monthly consumer expenditure data!
With that in mind, here's the nominal and inflation-adjusted median household income that has been reported by Sentier Research in each month since January 2000 through the most recent data available for March 2015:
To reasonably estimate both the nominal and real national dividend then, we would just need to multiply the monthly values for these figures by the number of consumer units in the U.S. Only here, we took the data for July of each year that overlaps the annual Consumer Expenditure Survey reports so we could determine the correlation between the two. Given the once-a-decade data revisions in the Census Bureau's monthly household estimates however, it was more complicated that it sounds. The chart below reveals the correlation we found for each year, which we will apply to the appropriate monthly data to get a monthly estimate of the number of consumer units in the U.S.
Here's how that math translates the U.S. Census Bureau's monthly estimates of the number of U.S. households translates into the number of U.S. "consumer units" with the correlations we identified in the previous chart after we graft them together, which fits together a lot better than you might have expected from the correlations we identified:
That was more work than it really needed to be, but now that we've done it, let's get to our results. The chart below shows our estimate of the nominal and real national dividend on a monthly basis since January 2000:
Here's what we observe in the data:
- The negative impact of the 2001 recession would appear to have been concentrated outside of U.S. households, as the national dividend rose during the months spanning the recession.
- The recovery from that recession would appear to have been especially slow, as the national dividend remained largely flat for about 18 months afterward. This observation is consistent with the general perception that this period represented a jobless recovery.
- The national dividend then began to grow steadily up until the start of the so-called "Great Recession".
- In real, inflation-adjusted terms, the Great Recession was really more of a double-dip affair, with a false recovery that reversed into a deeper decline in the last quarter of 2008, first with the financial system crisis triggered by the collapse of Lehman Brothers, then by the implosion of the U.S. automotive industry.
- In real terms again, the Great Recession went on far longer than the official NBER recession dates indicate. It didn't really bottom until the second quarter of 2011 - two years after the second quarter of 2009 that the National Bureau of Economic Research marked as the end of the Great Recession.
- We see that the phenomenon of a first quarter dip has been a recurring feature in the U.S. economy during the past decade. The suggestion that seasonal adjustments in the U.S. construction industry are off is not supported by our consumption-based national dividend calculation.
Looking at the most recent months, since December 2014, we see that our nominal and inflation-adjusted consumption-based measure of the economic well being of the ultimate end consumers in the U.S. has turned downward, which would so far appear to coincide with a period of greatly decelerated GDP growth, if not outright contraction. We'll get the official word on that for the first quarter of 2015 near the end of June, but our estimate of the national dividend certainly supports that outcome.
One thing that will be interesting to consider is if the impact of the Federal Reserve's various quantitative easing programs has had any effect upon the national dividend, which will be something that we'll explore in an upcoming post.
We'll conclude by repeating our previous observation that the national dividend has primarily risen since January 2000 because of the population of the U.S. increasing over time, or more accurately, because of the increase in the number of "consumer units" or households in the U.S., rather than real increases in the incomes earned by U.S. households.
Previously on Political Calculations
Once upon a time, less than two months ago, we solved a problem that had stymied economists since 1906. And we made it look easy!
Also, a rough draft of this post appeared briefly on the morning of 12 May 2015 giving a sneak peak at what we've been working on developing, under the working title "The National Dividend Monthly", which sounded way too much like the name of some obscure magazine!
Sentier Research. Household Income Trends: February 2015. [PDF Document]. 23 April 2015.
U.S. Census Bureau. Housing Vacancies and Homeownership (CPS/HVS). Historical Tables. Table 13. Monthly Household Estimates: 1955 to Present. [Excel Spreadsheet]. Accessed 3 May 2015.
U.S. Department of Labor Bureau of Labor Statistics. Consumer Price Index, All Urban Consumers - (CPI-U), U.S. City Average, All Items, 1982-84=100. Accessed 3 May 2015.