Peter L. Ormosi, "A Tip of the Iceberg? The Probability of Catching 
Cartels", Journal of Applied Econometrics, Vol. 29, No. 4, 2014,
pp. 549-566.

The data records firms that were involved in EU-wide cartels and were
discovered by the European Commission between 1985-2009. Software MARK
was used for the estimation. MARK is an open source statistical
application designed for estimating animal population parameters. The
software and instruction manuals are available at:

http://www.phidot.org/software/mark/download/

Data file - 1yr_rev.inp

As MARK cannot read .txt extensions the data is saved as input file
(.inp). It can be also read by any ASCII editor.  The 1; at the end of
each line denotes the number of observations for each firm (1 in this
case), which is required by MARK.

The data are organised in an n x K matrix, where n is the number of
individual colluding firms captured, K is the number of years in the
analysis, n=588 and K=26. This capture history matrix records whether
an investigation in year K discovered firm n. 1 records captures, 0
records no capture. Row m of the capture history matrix is the capture
history of firm m (where m is an element of n). The corresponding
names of firms are stored in the file firm_codes.txt. The columns each
represent a year from 1984 to 2009

To replicate the results, it is necessary to acquire some familiarity
with MARK. To replicate the estimates in Section 3.4, open MARK and
start a New File. Name the new project and then click on 'Select file'
to open the data (1yr_rev.inp). On the LHS, select 'Recaptures only'
for CJS models. For 'Ecounter occasions' give the number of years (26)
and click OK. You can now now set up the same models run for this
paper following instructions in the manual for MARK. 

File with stored models - 1YR_REV.dbf

The file 1YR_REV.dbf contains the CJS models estimated and discussed
in Section 3.4. Section 3.3 of the paper explains how these model
names have to be interpreted and what the corresponding statistics
mean. For example the model called "{Phi(./.) p(treatment5year) DM}"
was used for the estimates reported in Table 3 under the "5-year
windows" column. The file can be opened in MARK.

All files are zipped in the file plo-files.zip.