This is the follow-up blog to my previous post on Per Capita versus Per Household Personal Income to illustrate how per household income is a better metric than per capita income when comparing states. As shown in the first chart, there has been a significant decline in average household size of 35 percent to 2.7 people from 4.13 people in the United States.
Maine and New Hampshire have mirrored the overall trend, but to varying degrees. Maine’s household size was generally larger than the national average and New Hampshire from 1929 to the 1970s. By the late 1970s/early 1980s, Maine’s household size had fallen below the national average and below New Hampshire. After the 1980s, New Hampshire’s household size stabilized just below the national average while Maine’s continued to slide lower.
The next two charts then compares the growth in real, per capita income with that of real, per household income for both Maine and New Hampshire. With the two charts next to each other, it is easy to see that a larger gap begins to develop in per household income, relative to per capita income, between Maine in New Hampshire beginning in the 1980s–the same time that New Hampshire’s household size overtook Maine’s.
Overall, Maine’s relative economic performance relative to New Hampshire has been boosted, in the short-term, by a falling household size. However, when adjusting for this effect Maine’s economic performance worsens substantially. On per capita terms, New Hampshire’s income is 17 percent higher than Maine’s in 2009; while on per household terms it is 26 percent higher.
I will post another blog in the next few days showing how this relationship changes among all 50 states.
Note: The historical data from the Census Bureau on households was only available every ten years during the decennial census. Intervening years were interpolated. Since 2000, household size has been reported on a more frequent basis under the new American Community Survey (ACS). However, for technical reasons, this analysis only uses the 2009 ACS data with the intervening years being interpolated. This will be replaced by data from the 2010 decennial census when it becomes available.