Carlo Altavilla, Raffaella Giacomini, and Giuseppe Ragusa, "Anchoring
the Yield Curve Using Survey Expectations", Journal of Applied
Econometrics, Vol. 32, No. 6, 2017, pp. 1055-1068.

## Data

The yields data are pooled from two sources.

The three-month yield is from the Fed's H-15 release. The one used is
from the FRED data repository (code [DGS3MO]
(https://fred.stlouisfed.org/series/DGS3MO)).

For longer maturities, we use zero-coupon yields constructed in
Gurkaynak and Wright (2007). The excel spreadsheet containing the data
can be obtained [here]
(https://www.federalreserve.gov/econresdata/researchdata/feds200628.xls).
We focus on average-of-the-month data from January 1985 to December
2011. We consider yields of the following 17 maturities (in months):
3, 6, 9, 12, 15, 18, 21, 24, 30, 36, 48, 60, 72, 84, 96, 108, 120.
This choice provides us with a panel of 324 monthly observations on 17
different yields.

For convenience, we provide csv files with the data for both the
Gurkaynak and Wright yields and the 3-month Treasury yields obtained
from FRED in the files DGS3MO.csv and feds200628.csv, respectively.
These files are ASCII files in DOS format. They are zipped in the file
agr-data.zip. Unix/Linux users should use "unzip -a".

The macroeconomic variables used in the macro-augmented DNS model
consist of monthly observations on 23 U.S. macroeconomic time series
from 1985:1 through 2011:12. Table 2 of the online appendix describes
the data and the transformations applied to each individual series.
These data are in the file dataset_factors_data.csv, which is also in
agr-data.zip.

Data on yield expectations are from the Blue Chip Financial Indicators.

Data on the Blue Chip monthly expectation are proprietary; hence the
data cannot be lodged here. The data can be acquired by [Wolters
Kluwer]
(https://lrus.wolterskluwer.com/store/products/blue-chip-financial-forecasts-prod-000000000010041429/journal-item-1-000000000010041429).
Data on expectation were not available electronically and thus manual
scraping was necessary.

## Replication code

The zip file `agr-anchoring.zip` contains all the data (except the
Blue Chip data) and the code needed to replicate figures and tables
reported in the paper and the online appendix. The code is written in
`R`. Several R packages are used, but the code will automatically
install them provided the package `pacman` is installed (see
instructions below).

Note that the files in agr-anchoring.zip are *not* in DOS format, and
some of them are binary. The data directory contains Unix versions of
the three csv files.

Once unzipped, agr-anchoring.zip creates a tree with several 
directories:

R
  00-kalman.R
  01-helper.R
  02-load_data.R
  03-setup_model.R
Rmd
  figures
  functions.R
  header.tex
  kalman-tikzDictionary
  kalman.Rmd
  kalman.md
  kalman.pdf
  latex.R
  tables
data
  DGS3MO.RData
  DGS3MO.csv
  dataset_factors_data.csv
  feds200628.csv
results
  dns_kalman.Rda

6 directories, 17 files

The code is in the `replication/R` folder. The main program file is
`00-kalman.R`, which once executed calls `01-helper.R`,
`02-load_data.R`, and `03-setup_model.R`. `00-kalman.R` generates a R
data file, `dns_model.RData`, in `replication/results/`. This file
contains all the models estimated. The tables and figures in the paper
in the online appendix are obtained by "knitting" the rmarkdown file
`kalman.Rmd`, which uses the information in `dns_model.RData` and
yield and survey data files descibed above.

The following instructions can be used to produce `dns_model.RData`:

unzip agr-anchoring.zip
cd replication/R
## Install pacman package
R -e 'options(repos = c(CRAN = "http://cran.rstudio.com")); install.packages("pacman")'
## Run the code
R -e 'source("00-kalman.R")'

The following commands produce `replication/Rmd/kalman.pdf`

cd replication/Rmd
## Knitr kalman.rmd
R -e 'library(knitr); knit2pdf("kalman.Rmd")'

This last step produces a pdf file, `results/kalman.pdf` containing
the tables and figures reported in the paper and in the online
appendix.

A pdf file generated on March 23, 2017 is included in the replication
material.