Clustering st. errors removes significance

Dear all, I have been running a regression
reg y x1 x2 x3 x4, vce (cluster region)

where x1 - log GDP per capita, x2 - generalized trust in country and region is region code from 1 - latin america to 8 south east asia.
When I run simple OLS regression the variables of interest (x1 and x2) are both significant at the 5% level. However when I run the code above, this variables lose their significance. How can I explain this in my paper? Any ideas what can be driving this in clustered regression?

Furthremore, when i run

cgmreg y x1 x2 x3 x4, cluster (region legal_origins)
where legal origins range from 1 (socialist) to 5 (scandinavian)

x1 and x2 further become insignificant


Thanks