{smcl}
{* 30aug2013}{...}
{hline}
help for {hi:cpigen}
{hline}

{title:Add US price index CPI-U to current data}

{title:Syntax}

{p 6 16 2}
{cmd:cpigen} [{cmd:,} {cmd:replace}]

{synoptset 20 tabbed}{...}
{synopthdr}
{synoptline}
{synopt:{opt replace}}replace existing variables named {cmd:cpi} and {cmd:cpiu}{p_end}
{synoptline}
{p2colreset}{...}

{title:Description}

{p}{cmd:cpigen} uses a numeric variable {cmd:year} on the current data, and
a numeric variable {cmd:quarter} or a numeric variable {cmd:month} if available, 
defining calendar time in the usual way, to generate
two variables {cmd:cpi} and {cmd:cpiu} containing versions of the Consumer Price
Index commonly used to convert US dollar amounts to "real" dollars.  The first, {cmd:cpi}
(supported range 1913-2012),
is the research series (applying more consistent methods over time) CPI-U-RS or CPI-U-X1
where available, and
an adjusted version of the CPI-U where no research series is available, as is
{browse "http://www.census.gov/hhes/www/income/income03/cpiurstxt.html":standard Census practice}.  The
second, {cmd:cpiu} (supported range 1913-2013),
is the seasonally adjusted CPI-U, for all goods, for urban consumers, also available
from {stata "findit freduse":freduse CPIAUCSL}. {p_end}

{p}The research series data for 1978 to 2006 comes from the BLS publication
{browse "http://www.bls.gov/cpi/cpiurs1978_2006.pdf":http://www.bls.gov/cpi/cpiurs1978_2006.pdf}
updated at 
{browse "http://www.bls.gov/cpi/cpiurs1978_2010.pdf":http://www.bls.gov/cpi/cpiursai1978_2010.pdf}. For more discussion of these series, and other indices in the CPI family,
see {browse "http://www.bls.gov/cpi":http://www.bls.gov/cpi}
and the article by Kenneth J. Stewart and Stephen B. Reed, "CPI Research Series Using Current Methods, 1978-98," 
in the Monthly Labor Review, June 1999, downloadable at {browse "http://stats.bls.gov/opub/mlr/1999/06/cpimlr.pdf":http://stats.bls.gov/opub/mlr/1999/06/cpimlr.pdf}.
A useful basic description is {browse "http://www.irp.wisc.edu/faqs/faq5.htm":http://www.irp.wisc.edu/faqs/faq5.htm}.{p_end}

{p}Earlier data (monthly CPI-U data for 1913 
through 1920) comes from
{browse "ftp://ftp.bls.gov/pub/special.requests/cpi/cpiai.txt":ftp://ftp.bls.gov/pub/special.requests/cpi/cpiai.txt}.
Though undocumented, this source matches the unadjusted CPI-U {stata "findit freduse":freduse} data,
which shows 174 for December 2000, whereas the seasonally adjusted CPI-U is 174.6; 
the ratio of December 2006 to December 1977 is 3.250 for unadjusted and 3.255 for seasonally adjusted CPI-U, whereas
that ratio is 2.96 for the research series (seasonal adjustment matters more for ratios formed from months within a year, i.e.
short term inflation, but such differences are likely swamped by measurement error in the CPI).
{p_end}

{p}Note that the indices {cmd:cpi} and {cmd:cpiu} 
created by {cmd:cpigen} apply to national trends
in prices of all goods and services purchased
by the 9/10 of consumers who are 
classified as urban. Many indices are available 
for particular categories of goods, e.g. food or clothing or
medical care (1998 clothing prices were about 1.5 times 1978 prices, while
1998 medical care prices were about 4 times 1978 prices), or geographic areas, e.g. New York 
or large northeastern cities or small southern 
cities (large northeastern cities tend to have 
prices nearly twice as high as small southern 
cities). See Stewart and Reed (link above) or 
{browse "http://www.bls.gov/cpi/cpifaq.htm#Question_16":http://www.bls.gov/cpi/cpifaq.htm#Question_16}
{browse "http://www.bls.gov/cpi/cpifaq.htm#Question_19":http://www.bls.gov/cpi/cpifaq.htm#Question_19} for more.{p_end}

{p}For indices not in the CPI family, see e.g. {browse "http://www.bea.gov/papers/pdf/cpi_pce.pdf":http://www.bea.gov/papers/pdf/cpi_pce.pdf} and
{browse "http://www.ssa.gov/OACT/COLA/AWI.html":http://www.ssa.gov/OACT/COLA/AWI.html}.{p_end}

{p}The variables {cmd:cpi} and {cmd:cpiu} created by {cmd:cpigen} are normalized to 2000 dollars in annual data, or December 2000 dollars in monthly data.
To turn nominal income, say, into real income measured in 2000 dollars, simply divide {cmd:income} by {cmd:cpi}.
To turn nominal income into real income measured in 2005 dollars, first divide {cmd:income} by {cmd:cpi}, then multiply by the value of {cmd:cpi} in 2005.
The idea is, what cost a dollar in 2000 cost about 1.13 in 2005, so a dollar of income in 2005 made you about as well off as 88.5 cents of income in 2000.{p_end}

{title:Example}

webuse nlswork, clear
replace year=year+1900
cpigen
g nomwage=exp(ln_w)*cpi
su cpi if year==1978
g wage78=nomwage/cpi*r(mean)
la var wage78 "Real Wage, 1978 dollars"
g x=exp(_n/20-1) in 1/80
kdensity wage78 if year==1968, at(x) g(d68) nogr
kdensity wage78 if year==1978, at(x) g(d78) nogr
kdensity wage78 if year==1988, at(x) g(d88) nogr
la var d68 "1968 Wages"
la var d78 "1978 Wages"
la var d88 "1988 Wages"
line d68 d78 d88 x, xsc(log)

{title:Author}

    Austin Nichols
    Urban Institute
    Washington, DC, USA
    austinnichols@gmail.com

{title:Also see}

{p 1 10}On-line: {stata "findit freduse": freduse (on SSC)}{p_end}