Log-linear estimation

[RenzoGuzman]

Hi there,
I'm trying to estimate this model for manufactured exports in my country for 1991-2011 : LOG(MX) = b0 + b1*LOG(ITCR) + b2*LOG(GDPUSA) + b3*LOG(PROMA)
where:
XM : Manufactured Exports
ITCR : Real Exchange Rate
GDPUSA : United States GDP
PROMA : Workforce Productivity

The problem is that b1 and b3 are supposed to be positive, but after I estimate this model, those parametres are negative.

In addition, I estimated this model for a shorter sample (2001-2011) and those parametres were positive in this last estimation, so I don't know what is the reason of it. I don't think there's a breakpoint, because I added a dummy variable and it was non-significant. So could you help me to solve this problem? Thanks in advance. =)