Note:  To do this in STATA one must first create and save the program nlpiece.do 
     containing the function to be estimated, as shown below.  Then one gives then command
     nl piece apply where apply is teh dependent variable.  (Created 20 Mar 2006.)
     
     1 . use "D:\s209\ler01.dta", clear
      
     2 . generate x1=sumnews
      
     3 . do nlpiece
      
     4 . program nlpiece
        1.         version 8.0
        2.         if "`1'" == "?" {
        3.                 global S_1 "B0 B1 B2 XP"
        4.                 global B0=200000
        5.                 global B1=2100
        6.                 global B2=-2300
        7.                 global XP=60
        8.                 exit
        9.         }
       10.         replace `1' = $B0 + $B1*x1 + $B2*(x1-$XP)*(x1>=$XP)
       11. end
      
     5 . 
      end of do-file
      
     6 . nl piece apply
      (obs = 40)
      
      Iteration 0:   residual SS =  5.03e+10
      Iteration 1:   residual SS =  4.70e+10
      Iteration 2:   residual SS =  4.65e+10
      Iteration 3:   residual SS =  4.64e+10
      Iteration 4:   residual SS =  4.64e+10
      Iteration 5:   residual SS =  4.64e+10
      Iteration 6:   residual SS =  4.64e+10
      Iteration 7:   residual SS =  4.64e+10
      Iteration 8:   residual SS =  4.63e+10
      Iteration 9:   residual SS =  4.63e+10
      
            Source |       SS       df       MS            Number of obs =        40
      -------------+------------------------------         F(  3,    36) =     19.95
             Model |  7.7040e+10     3  2.5680e+10         Prob > F      =    0.0000
          Residual |  4.6350e+10    36  1.2875e+09         R-squared     =    0.6244
      -------------+------------------------------         Adj R-squared =    0.5931
             Total |  1.2339e+11    39  3.1638e+09         Root MSE      =  35881.64
                                                           Res. dev.     =  948.3392
      (piece)
      ------------------------------------------------------------------------------
             apply |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
      -------------+----------------------------------------------------------------
                B0 |   183781.5   12407.43    14.81   0.000     158618.1    208944.9
                B1 |    2519.42   454.8478     5.54   0.000     1596.946    3441.894
                B2 |  -2393.364   614.2728    -3.90   0.000    -3639.167   -1147.561
                XP |         49   9.908172     4.95   0.000      28.9053    69.09471
      ------------------------------------------------------------------------------
      * Parameter B0 taken as constant term in model & ANOVA table
       (SEs, P values, CIs, and correlations are asymptotic approximations)
      
     7 . translate @Results piece.txt