Garth Holloway, Simeon Ehui, and Amare Teklu, "Bayes Estimates of 
Distance to Market: Transactions Costs, Cooperatives and Milk Market 
Development in the Ethiopian Highlands", Journal of Applied 
Econometrics, Vol. 23, No. 5, 2008, pp. 683-696.

The data used in this paper are kindly made available by the 
International Livestock Research Institute (ILRI), where this research 
was conducted while the corresponding author was on sabbatical, 
1999-2001.  Authors wishing to use the data are kindly requested to 
acknowledge ILRI in their work.

The data set contains a panel of observations constructed for 68 
households during 3 visits in the 1997 production year, a total of 204 
observations. The data set contains 90 measures.  However, only 9 
quantities are used in the article.  These 9 measures include the sales 
and production data (dependent in the regressions) and observations on 
seven covariates (respectively, cross-breed cows, local-breed cows, 
extension visitation, minutes-time to walk bucketed fluid milk to the 
milk cooperative, years of formal schooling, and two site-specific 
binary variables).

The file het-files.zip contains two ASCII files, bayesdistancetomarket.m 
and distancetomarketdata.ascii. UNIX users should use 'unzip -a'.

The data are in an ASCII file in DOS format and are loaded into the 
MATLAB code (in the file entitled bayesdistancetomarket.m and copied 
into the text appendix entitled het-appendix-12-09-06) as lines 68 to 
79, namely

  %Fetch
  data=load('distancetomarketdata.ascii');
  production=data(:,2);
  sales=data(:,12);
  crossbreed=data(:,86);
  localbreed=data(:,87);
  extension=data(:,90);
  distance=data(:,20);
  education=data(:,24);
  ilukura=data(:,32);
  mirti=data(:,33);
  %pause

Thus the production data reside in column 2 of the data matrix, the 
sales data reside in column 12, and so on.  The remainder of the code 
appears in the same format, with a one-word descriptor explaining the 
block, and a pause commented at the end of the block.

Throughout, an attempt has been made to ensure that anyone with a very 
basic familiarity with programming constructs and basic linear algebra 
can follow and interpret the MATLAB code.  However, I strongly encourage 
users who encounter difficulty to email me at garth.holloway [AT] 
reading.ac.uk.

Finally, I repeat my request that colleagues wishing to use the data in 
their own research acknowledge ILRI, who -- in addition to providing 
this very informative data set which is amenable to composed-error 
random-censored Tobit modelling in a multivariate framework -- provided 
a research environment that was unsurpassed in my experience.

Thank you.

Garth Holloway