Carmen Fernández, Eduardo Ley, and Mark J. F. Steel, "Model Uncertainty
in Cross-Country Growth Regressions", Journal of Applied Econometrics,
Vol. 16, No. 5, 2001, pp. 563-576.

(revised August 28, 2006)

The file fls.zip contains several files, which are described below. Text
files are ASCII files in DOS format.

Text Files:
===========

(1) growth.f     Contains the f77 code used in the paper; it can only handle
                 #reg =<52.

(2) fls060828.f  Updated f77 code for #reg =< 104, and to compute jointness
                 measures.

(3) growth.par   See growth.f header for explanation regarding the parameters.

(4) growth.dat   Dataset used in paper; 42 vars and 72 observations.
		 The first 41 lines contain the names of the variables, then
		 the dataset (y,Z) follows in (7f14.6) format (6 lines per
		 observation). See 'setup' routine in growth.f for more
		 details. Descriptive stats are:

                              mean        std
y ..... growth             0.2073E-01  0.1813E-01 
z( 1).. PrSc Enroll          0.7953      0.2440    
z( 2).. Life Exp              56.58       11.37    
z( 3).. GDPsh560              7.492      0.8785    
z( 4).. Mining               0.4487E-01  0.7651E-01
z( 5).. Eco Org               3.542       1.258    
z( 6).. Yrs Open             0.4393      0.3528    
z( 7).. English %            0.7571E-01  0.2377    
z( 8).. Foreign %            0.3744      0.4191    
z( 9).. R FEX Dist            121.7       40.72    
z(10).. Equip Inv            0.4429E-01  0.3426E-01
z(11).. NEquip Inv           0.1495      0.5429E-01
z(12).. std(BMP)              45.60       95.13    
z(13).. Outwar Or            0.3889      0.4875    
z(14).. Bl Mkt Pm            0.1567      0.2893    
z(15).. Area                  972.9       2038.    
z(16).. LatAmerica           0.2778      0.4479    
z(17).. SubSahara            0.2083      0.4061    
z(18).. High Enroll          0.4335E-01  0.5138E-01
z(19).. %Publ Edu            0.2465E-01  0.9381E-02
z(20).. Rev & Coup           0.1824      0.2362    
z(21).. War Dummy            0.4028      0.4905    
z(22).. Pol Rights            3.451       1.883    
z(23).. Civl Lib              3.466       1.700    
z(24).. Abs Lat               25.73       17.13    
z(25).. Age                   23.71       37.05    
z(26).. Brit Col             0.3194      0.4663    
z(27).. Buddha               0.5611E-01  0.1828    
z(28).. Catholic             0.4218      0.3938    
z(29).. Confuncious          0.1903E-01  0.8681E-01
z(30).. EthnoL Frac          0.3706      0.2941    
z(31).. French Col           0.1250      0.3307    
z(32).. Hindu                0.1806E-01  0.1006    
z(33).. Jewish               0.1271E-01  0.9591E-01
z(34).. Muslim               0.1475      0.2925    
z(35).. Pr Exports           0.6732      0.2967    
z(36).. Protestants          0.1730      0.2498    
z(37).. Rule of Law          0.5509      0.3329    
z(38).. Spanish Col          0.2222      0.4157    
z(39).. Pop g                0.2039E-01  0.1014E-01
z(40).. Work/Pop            -0.9537      0.1880    
z(41).. Lab Force             9305.      0.2473E+05

(5) fls.par      parameter file for new code (first line contains namenamename)

(6) grk41t72.dat is the data file used in the paper except that growth (y)
                 is now expressed in percentage points.

(7) grk67t88.dat is the data file used in the BACE paper in AER by SDM.

The structure of the *.dat file is 1st line with kreg, 2nd line
with nobs, then kreg lines w/ varnames, and then nobs lines each
with (kreg+1) data values: (y,Z):
................................................................
kreg       Note: kreg must be =< 104
nobs       No limit but if nobs > 88 change maxn=88 throughout
<# kreg lines with the names of the kreg vars>
<# nobs lines with kreg+1 format-free variables: (y,Z) >

(8) ab_test_00.out is a sample 100K/500K run that takes about one hour on
                   an old 800Mhz PPCG4 iMac 


Stata Files:
===========

(9) millions.dta  Larger Growth dataset originally from Sala-i-Martin;
                  140 countries, 64 variables. Not all vars available for
                  each country, see below.  This version dates from 1998.

(10) millions.desc The output from the Stata describe command.