July 1975 -- 8.0% July 1976 -- 6.4% July 1977 -- 5.9% July 1978 -- 6.5% July 1979 -- 9.9% July 1980 -- 14.3% July 1981 -- 11.2% July 1982 -- 7.4% January 1984 -- 3.5% January 1985 -- 3.5% January 1986 -- 3.1% January 1987 -- 1.3% January 1988 -- 4.2% January 1989 -- 4.0% January 1990 -- 4.7% January 1991 -- 5.4% January 1992 -- 3.7% January 1993 -- 3.0% January 1994 -- 2.6% January 1995 -- 2.8% January 1996 -- 2.6% January 1997 -- 2.9%

January 1998 -- 2.1% January 1999 -- 1.3% January 2000 -- 2.5% January 2001 -- 3.5% January 2002 -- 2.6% January 2003 -- 1.4% January 2004 -- 2.1% January 2005 -- 2.7% January 2006 -- 4.1% January 2007 -- 3.3% January 2008 -- 2.3% January 2009 -- 5.8% January 2010 -- 0.0% January 2011 -- 0.0% January 2012 -- 3.6% January 2013 -- 1.7% January 2014 -- 1.5% January 2015 -- 1.7% January 2016 -- 0.0% January 2017 -- 0.3% January 2018 -- 2.0% January 2019 -- 2.8%

Before 1975, Social Security benefit increases were set by legislation. 

• More at the 2016 COLA Press Release (in pdf)
  
  

CPI-U and CPI-W:  As a result of budget cuts and because little difference was seen between the CPI-U and CPI-W measures during 1978–1980, the BLS discontinued the separate but overlapping samples of individual items and outlets maintained from 1978–1980 for the CPI-U and CPI-W. Because maintaining two independent samples was not sustainable, BLS economists track spending and prices by using the CPI-U sample of geographic areas, outlets, items, and prices. The CPI-W is then derived by adjusting the weights for various spending categories, reflecting that the spending habits of the wage earner population differ somewhat from the all urban consumer population.  For more, see Why does BLS provide both the CPI-W and CPI-U?