$R^2$ is too high- reasons? What are the reasons of too high values of $R^2$? I only know that its value increases when the number of independent variables increases. What are the other factors?
 A: Based on your comment describing your problem in response to @PsychometStats, it appears you are regressing a variable that tends to increase over time - population - on a variable that tends to increase over time - GDP, and possibly on others.   This will cause high $R^2$ values regardless of whether there is any real relationship between the target variable and the regressors.  
The general term relating to this phenomenon is cointegration; see the answers to the Stackexchange question Linear regression of nonstationary variables? for some explanation or the Wikipedia page https://en.m.wikipedia.org/wiki/Cointegration (probably best to start the latter at the "History" section; the fourth paragraph has a particularly appropriate example.)  It is very common with basic economic series, and this link https://www.fmf.uni-lj.si/finmath09/Cointegration.pdf contains an explanation oriented towards people who work with these series.
Note that two variables don't have to be cointegrated in order for there to be a spurious regression effect; I highlight cointegration because that appears often in economic time series.  
For a straightforward example, consider regressing annual GDP from 2002 through last year against a single regressor comprised of my daughter's age plus the number of trips she took to Germany in that year, with a trip counted double if she travelled alone:

Here's the regression (in R):
> summary(lm(gdp~age))

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) 11.05660    0.22344   49.48  < 2e-16 ***
age          0.46161    0.01982   23.29 9.06e-14 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.4797 on 16 degrees of freedom
Multiple R-squared:  0.9713,    Adjusted R-squared:  0.9695 
F-statistic: 542.3 on 1 and 16 DF,  p-value: 9.056e-14

That's a pretty impressive $R^2$ by most standards, but... the regression is still meaningless.
If we instead try to see if we can explain changes in annual GDP by changes in the modified age variable, we get:
> summary(lm(diff(gdp)~diff(age)))

Coefficients:
            Estimate Std. Error t value Pr(>|t|)  
(Intercept)  0.61327    0.33390   1.837   0.0862 .
diff(age)   -0.09082    0.26495  -0.343   0.7365  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.6361 on 15 degrees of freedom
Multiple R-squared:  0.007772,  Adjusted R-squared:  -0.05838 
F-statistic: 0.1175 on 1 and 15 DF,  p-value: 0.7365

which, fortunately for economic forecasters everywhere, indicates that the modified age variable does not help us predict year-over-year changes in GDP.
For how to address this problem, I recommend doing some research on cointegration; the Wikipedia page linked to above is not a bad place to start, nor is the linked PDF. 
A: I would check for very highly correlated variables. This is known as multicollinearity, which may substantially overinflate your beta coefficients. Consequently, your $R^2$ may be unreasonably high. If that is the case, remove one of such variables and check if $R^2$ decreases.    

EDIT: 
  For clarification purposes: The above implied to check if the DV itself was very highly correlated with one of the predictors.

