reg y s_ds ds s_dsj dsj if code==12

sum quar_sale if code==12

local mm = r(mean)

test s_dsj*`mm' + dsj = 0

非常感谢上面两位朋友的耐心细致的解答，但为什么结果还是和原来一样呢，是因为观测值太少吗？

. reg y s_ds ds s_dsj dsj if code==12

Source | SS df MS Number of obs = 15
-------------+------------------------------ F( 4, 10) = 0.35
Model | 7.24851891 4 1.81212973 Prob > F = 0.8388
Residual | 51.8826712 10 5.18826712 R-squared = 0.1226
-------------+------------------------------ Adj R-squared = -0.2284
Total | 59.1311901 14 4.22365644 Root MSE = 2.2778

------------------------------------------------------------------------------
y | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
s_ds | 3.80e-18 4.74e-18 0.80 0.441 -6.77e-18 1.44e-17
ds | -1.61e-09 2.33e-09 -0.69 0.505 -6.81e-09 3.58e-09
s_dsj | -1.21e-19 1.65e-19 -0.73 0.480 -4.88e-19 2.46e-19
dsj | 5.62e-11 8.56e-11 0.66 0.527 -1.35e-10 2.47e-10
_cons | -.3151068 1.087856 -0.29 0.778 -2.739001 2.108787
------------------------------------------------------------------------------

.
.
.
.
. sum quar_sale if code==12

Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
quar_sale | 18 5.88e+07 7.55e+08 -1.74e+09 7.15e+08

. local mm = r(mean)

.
.
.
.
. test s_dsj*`mm' + dsj = 0

( 1) 5.88e+07 s_dsj + dsj = 0
Constraint 1 dropped

F( 0, 10) = .
Prob > F = .

.
.
. test s_dsj * 5.88e+07 +dsj=0

( 1) 5.88e+07 s_dsj + dsj = 0
Constraint 1 dropped

F( 0, 10) = .
Prob > F = .

建议你调整各个变量的统计口径，使他们尽量一致，这样出来的结果也便于解释。