Nalan Basturk, Cem Cakmakli, Pinar Ceyhan, and Herman van Dijk,
"Posterior-Predictive Evidence on US Inflation using Phillips Curve
Models with Non-filtered Data", Journal of Applied Econometrics,
Vol. 29, No. 7, 2014, pp. 1164-1182.

The data are stored in the file bccv-data.xls file. It includes three
sheets:

  Defl (GDP Deflator)
  RMC (Real marginal cost series where we use labor share in non-farm
     business sector as a proxy)
  SInf (Survey Based Inflation Expectations from University of Michigan
     Surveys of Consumers)

The file bccv-data.xls is zipped in bccv-excel.zip.

In addition, the data are also contained in three CSV files (comma
separated ASCII files, in DOS format). These files (Defl.csv, 
RMC.csv, and SInf.csv) are zipped in the file bccv-csv.zip. Unix/Linux
users should use "unzip -a".

For Defl, the source is the Federal Reserve Bank of St. Louis (FRED
Dataset): https://research.stlouisfed.org/fred2/series/GDPDEF

For RMC, the source is the Federal Reserve Bank of St. Louis (FRED
Dataset). Logarithm of the series is used:
https://research.stlouisfed.org/fred2/series/PRS85006173

For SInf, the source is University of Michigan Surveys of Consumers.
The mean of the answers of the consumers to the question: "By about
what percent do you expect to go up, on the average, during the next
12 months?" is used. Historical differences between CPI inflation and
GDP price deflator are computed in real time and survey expectations
are modified by adding these historical differences at each period. 
 
http://www.sca.isr.umich.edu/tables.php
http://research.stlouisfed.org/fred2/series/CPIAUCSL