The results show that both t-statistics and R square are not good, meaning the relationship is not strong that suggests either that important variables are absent or the data is not good for your purpose, or the approach is problematic.
I see your sample runs from 1997 to 2013, that is 16 years, but your sample size is only 195, so it seems the data is monthly.
If that is true, then such an approach is problematic. Exchange rate is changing almost instantly and using monthly data would not be necessarily appropriate. How have other people approached this problem?

The results show that both t-statistics and R square are not good, meaning the relationship is not strong that suggests either that important variables are absent or the data is not good for
your purpose, or the approach is problematic.
I see your sample runs from 1997 to 2013, that is 16 years, but your sample size is only 195, so it seems the data is monthly.
If that is true, then such an approach is problematic. Exchange rate is changing almost instantly and using monthly data would not be necessarily appropriate. How have other people approached this problem?