Erik Kole, Dick van Dijk, "How to Identify and Forecast Bull and Bear 
Markets?", Journal of Applied Econometrics, Vol. 32, No. 1, 2017,
pp. 120-139.

There are five data files. They are all ASCII files in DOS format.
These files are zipped in the file kvd-data.zip. Unix/linux users
should use "unzip -a".

The file data_raw_week.txt contains the raw weekly data, as it was
downloaded. First date is Friday December 30, 1960. Last date is
Friday December 27, 2013. The columns give the date (yyyy-mm-dd),
the value of the S&P500 index (SP500), the 3-month T-bill rate (in
% per year; WTB3MS), the yield on AAA rated bonds (in % per year;
WAAA), the yield on BAA rated bonds (in % per year; WBAA), the
yield on 10-year government bonds (in % per year; WGS10YR), and the
dividends over the past year.

The first five series were downloaded from the FRED database of the
Federal Reserve Bank in St. Louis (mnemonics are given in
parentheses). Dividends wee taken from the website of Robert Shiller.
Shiller gives the sum of past dividends for each month. We assume that
the dividend observation for a given a month pertains to all weeks in
that particular month.

The files ur_rt.txt, ip_rt.txt, and cpi_rt.txt contain vintage data
for the annual change in the unemployment rate, the annual relative
change in industrial production, and the annual relative change in the
cpi. The columns give the vintage and the observation dates
(yyyy-mm-dd), the real-time observation, and the revised observation.
The data were downloaded from the ALFRED database of the Federal
Reserve Bank in St. Louis.

The file data_joined_week.txt contains the transformed weekly data,
as explained in Section 3 of the paper and Appendix B. First date
is Friday January 5, 1962. Last date is Friday December 27, 2013.
The columns give the date (yyyy-mm-dd), the discrete return on the
S&P500 index, the 1-week risk-free rate (the T-bill rate divided by
5200), the inflation rate (transformed to yearly change of the log
CPI), the growth rate of industrial production (transformed to
yearly change of the log IP), the yearly change in the unemployment
rate, the t-bill rate (difference from the average over the past
year), the term spread (10-year yield minus 3-month yield), the
credit spread (AAA-yield minus BAA-yield), the dp-ratio (difference
from the average over the past year), and the volatility
(RiskMetrics method applied to log returns, lambda = 0.95).