Kumar -

You might look at a confidence interval about the difference in their means, rather than look at their ratios.  The latter might be more easily misinterpreted.  Don't worry about "significance."  That word is a misnomer.  

 https://www.researchgate.net/publication/262971440_Practical_Interpretation_of_Hypothesis_Tests_-_letter_to_the_editor_-_TAS

You might plot one against the other in a scatterplot, if we are talking about continuous data. 

Regressing one against the other could be instructive. If you have two continuous variables, and say suppose when x is zero then y must be, and the relationship is linear (see scatterplots), then under those circumstances,  the regression coefficient would be the ratio you may have had in mind.  That would likely be heteroscedastic.  The classical ratio estimator would generally be robust for that.  Here are a couple of references that look at finite population sampling, but the b and V(b) estimates would apply regardless:

See page section 5, page 10 here (ignore total, T, as it is for a finite population,  unless you want that):

 https://www.researchgate.net/publication/261474159_Some_Proposed_Optional_Estimators_for_Totals_and_their_Relative_Standard_Errors_for_a_set_of_Weekly_Quasi-Cutoff_Sample_Establishment_Surveys

And see page 2 (for b), and bottom 2/3s of page 7 (for estimated variance of the prediction error, which, by the way, is the square of STDI in SAS PROC REG, for example) in here:

 https://www.researchgate.net/publication/263036348_Properties_of_Weighted_Least_Squares_Regression_for_Cutoff_Sampling_in_Establishment_Surveys

(Note: I mistakenly used the word "confidence" rather than "prediction" there for the intervals, but that does not really change anything of practical consequence.)  

Cheers - Jim

Article Some Proposed Optional Estimators for Totals and their Relat...

Article Practical Interpretation of Hypothesis Tests - letter to the...

Article Properties of Weighted Least Squares Regression for Cutoff S...

Thanks a lot James. Does effect size (Cohen's D) will be of any help in this case ?

Yes, it would appear so. 

You might want to take a look at the following:

 http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3444174/