foreach & levelsof, using two variables

[Dennis Uhrskov]

Hi Statalist,

I have constructed 27 portfolios across 1962 - 2018. Now i want to find the market cap average percentiles across the years and the portfolios for June.

I have tried this:

// We only keep the June data to identify each portfolio
keep if month(dofm(YM)) == 6

drop if portfolio27 == .

// Identify percentiles in each portfolio
levelsof june_year, local(years)
levelsof portfolio27, local(level)
gen int size_percentiles = .
foreach y of local years | x of local level {
    pctile cutoffs = MktCap if june_year  == `y' | portfolio27 == `x', nq(3)
    xtile pf = MktCap, cutpoints(cutoffs)
    replace size_percentiles = pf if june_year == `y' | portfolio27 == `x'
    drop pf cutoffs
}

but i get this error message ''{ required''. So i tried to add another set of brackets and got this error ''program error: code follows on the same line as open brace'', even without the error, is this really doing what i want it to?

PS. I know that the code works, because i used it to construct the portfolios, but that was only on june_year, so in this example i have tried to add another levelsof, namely the portfolio27.

Data sample:

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(june_year MktCap portfolio27)
1974    6.0145 19
1975  15.68175 12
1976   14.4495 12
1977         . 11
1971  242.5963 17
1972 254.29945 17
1973 250.79996 16
1974  246.0596  9
1975         .  9
1977 293.78162 16
1978 308.89575 16
1979  292.1955 17
1980         . 17
1982  287.0055 18
1987    1.6495 19
1988  6.654375 19
1989  6.329674 12
1990  5.013796  2
1991     7.152  3
1992    34.344  2
1993    126.45 10
1994 190.25626  6
1982   12.7875  3
1983    12.705  3
1984     10.83  3
1985    19.944  1
1986    16.874 10
1976  7.890738 21
1977         . 21
1979  21.77247  3
1980  62.57572 10
1981  52.41597 13
1982 202.86795  6
1983  157.0747  9
1984 210.75395  9
1985   537.444  7
1986   686.475  7
1986 228.45325  4
1987   356.094  8
1988 188.35873 16
1989 257.36996 17
1990    253.74  7
1991  318.9904  9
1992 583.59436  9
1993    296.67  7
1994  405.6938 16
1995  431.8043  8
1996 367.74625  9
1997    559.13 17
1998  979.1423 18
1999 2172.8984  9
2000 1097.8752 18
2001  870.4118 18
2002 1115.4672 16
2003  1074.688 25
2004   771.005 26
2005 1317.0795 22
2006 1134.2701 27
2007  969.8489 27
1988  541.9273  9
1989    379.33 18
1990 295.57333 18
1991 317.95575 26
1992  470.8856 26
1993  634.4625 26
1994  889.4475 18
1995  1040.405 18
1996 1102.0231 26
1997         . 25
1974  6.316873 19
1975  8.252998 19
1976     7.422 19
1977    10.392 19
1978 13.152375 20
1971  1160.392  9
1972  406.7928  9
1973   283.815  9
1974   382.561  7
1975         . 16
1979  324.8184 17
1980  532.4825 25
1981  357.6225 18
1982   385.009 18
1983  428.0753 17
1984   489.843 25
2015 15206.462  7
1994         .  3
1997  22.17456 10
1998   29.3615  1
1999  18.31578  3
2000    15.796 10
2001   22.9968 10
2002   17.9622  2
2003 17.925648 11
2004  30.27879 10
2005    90.784  3
2006  167.8835  3
2007  88.04961  3
2008  56.83838  3
2009   67.2382 19
end

Let me know if something is unclear

Dennis

[FernandoRios]

I think you just need two loops, one for each variable

 levelsof june_year, local(years)
levelsof portfolio27, local(level)
gen int size_percentiles = .
foreach y of local years {  
      foreach x of local level {    
           pctile cutoffs = MktCap if june_year  == `y' | portfolio27 == `x', nq(3)    
           xtile pf = MktCap, cutpoints(cutoffs)    
            replace size_percentiles = pf if june_year == `y' | portfolio27 == `x'    
           drop pf cutoffs    
       }
}

HTH
Fernando

[Dennis Uhrskov]

Thanks, it works

Just not as intended but thats another problem..