Census 2010 contains the latest available information on the block level of census geography. In order to estimate population variables on this level of granularity for the year 2016, simplyfying assumptions have to be made.
In this case, a model which assumes a relatively unchanged demographic structure is taken. Using the 2010 data, the distribution of population, households and housing units within each block group is calculated. These distributions are then used in conjunction with the most relevant socioeconomic information from the 5-year American Community Survey 2016 dataset to produce estimates on block level.