i am using panel data from 1990 to 2017 for 43 countries. I have two alternate dependent variables "y" and six alternates for variable "x" , when i estimate model with country fixed effects the coefficients are consistently significant with negative sign for variable "x" in all 12 models (2 alternate "y" * 6 alternate "x"), but when i also add year-fixed-effects "i.year" with country fixed-effects "i.ccode" the results become insignificant for variable x in all 12 models. Is there any way out of this situation? or what if i ignore year-fixed effects and proceed with only country-fixed effects results? please help
Note: indepyear is the last two digits of the country year of independence, and i.colonyg is the colonial regime
Code:
xtset ccode year
ivreg2 y (x=msphund) lnpn pse religion l.polity2 l.gdpg popd indepyear i.colonyg i.ccode, endog(x)
ivreg2 y (x=msphund) lnpn pse religion l.polity2 l.gdpg popd indepyear i.colonyg i.ccode i.year, endog(x)
I face similar problem when i also use cluster errors.

Code:
ivreg2 y (x=msphund) lnpn pse religion l.polity2 l.gdpg popd indepyear i.colonyg i.ccode, cluster(ccode) endog(x)
ivreg2 y (x=msphund) lnpn pse religion l.polity2 l.gdpg popd indepyear i.colonyg i.ccode i.year, cluster(ccode)  endog(x)