I am hoping to express ACS 2013-2017 dollar-denominated data as 2010 dollars. Following this resource, my understanding is that I would take the average CPI value for 2010 and divide it by the average CPI value of 2017 (the most recent/last year in the series). This would give me the inflation adjustment ratio which I would then multiply by an ACS estimate for 2013-2017. Why is the CPI value for the last year in the series used (i.e. 2017) rather than an average CPI value for all 5-years (i.e. 2013-2017)?
Census provides specific advise for comparing ACS data. For the 2017 ACS release look here:
https://www.census.gov/programs-surveys/acs/guidance/comparing-acs-data.2017.html
In sum, the recommendation is that one use the CPI Research Series to conduct comparisons where you inflate or deflate reported values. You can find information about the values for the CPI-U-RS (or R-CPI-U-RS as it now seems to be called) here:
https://www.bls.gov/cpi/research-series/r-cpi-u-rs-home.htm
My recollection, and someone please correct me if wrong, is that ACS dollar values are adjusted to the last month in the period, so to adjust 2013-17 values down to the average value for 2010 you would use the ratio of the 2010 average to the December 2017 index figure.
Thanks, Cliff. Based on your note about ACS dollars being adjusted to the last month in the period, I looked into this a bit more to see how multi-year estimates are created. I found this Census resource (page 22) which explains how multiyear estimates are created and provides an example which is helpful. Here is the excerpt (emphasis added):