E. M. Iglesias and G. D. A. Phillips, "Improved Instrumental Variables 
Estimation of Simultaneous Equations under Conditionally Heteroskedastic 
Disturbances", Journal of Applied Econometrics, Vol. 27, No. 3, 2012, 
pp. 474-499.

The dataset and the description were found at:

http://fmwww.bc.edu/ec-p/data/wooldridge/FISH.des
http://fmwww.bc.edu/ec-p/data/wooldridge/FISH.dta

The data are in fish.dta (a Stata dataset) and fish.txt (an ASCII file in
DOSD format). The former is zipped in fish-dta.zip and the latter in
fish-txt.zip. Unix/Linux users should use "unzip -a" for fish-txt.zip but
not for fish-dta.zip.

Dataset as described by Stata:

obs:             97                          
vars:            20                          19 Nov 2001 23:15
size:         5,529 (99.9% of memory free)
-------------------------------------------------------------------------------
              storage  display     value
variable name   type   format      label      variable label
-------------------------------------------------------------------------------
prca            float  %9.0g                  price for Asian buyers
prcw            float  %9.0g                  price for white buyers
qtya            int    %9.0g                  quantity sold to Asians
qtyw            int    %9.0g                  quantity sold to whites
mon             byte   %9.0g                  =1 if Monday
tues            byte   %9.0g                  =1 if Tuesday
wed             byte   %9.0g                  =1 if Wednesday
thurs           byte   %9.0g                  =1 if Thursday
speed2          byte   %9.0g                  min past 2 days wind speeds
wave2           float  %9.0g                  avg max last 2 days wave height
speed3          byte   %9.0g                  3 day lagged max windspeed
wave3           float  %9.0g                  avg max wave hghts of 3 & 4 day
                                                lagged hghts
avgprc          float  %9.0g                  ((prca*qtya) + (prcw*qtyw))/(qtya
                                                + qtyw)
totqty          int    %9.0g                  qtya + qtyw
lavgprc         float  %9.0g                  log(avgprc)
ltotqty         float  %9.0g                  log(totqty)
t               byte   %9.0g                  time trend
lavgp_1         float  %9.0g                  lavgprc[_n-1]
gavgprc         float  %9.0g                  lavgprc - lavgp_1
gavgp_1         float  %9.0g                  gavgprc[_n-1]