Interaction term in a mlogit

[Frederik Thorbjoern]

Hello

I have estimated the following mlogit:

mlogit EntryMode AS InternalUncertainty ExternalUncertainty AS_ExtUnc StrategicFlexibility DemandVolatility ManufacturingIndustry LnExchangeVolatility LnFirmSize, baseoutcome(1)
fitstat

After that I have made an analysis of predicted probabilities and marginal effects for example AS with the following codes:

//Predicted Prob
margins, at(AS=(0(0.01)0.8717034)) predict(outcome(1)) saving(ASPP1, replace)
margins, at(AS=(0(0.01)0.8717034)) predict(outcome(2)) saving(ASPP2, replace)
margins, at(AS=(0(0.01)0.8717034)) predict(outcome(3)) saving(ASPP3, replace)
combomarginsplot ASPP1 ASPP2 ASPP3, xlabel(0 "Min" 0.2335184 "Mean" 0.8717034 "Max") xtitle("Asset specificity") ytitle("Predicted Probability") title("") recastci(rline) ciopts(lpattern(shortdash_dot))

//Predicted prob at specified levels
prvalue, x(AS=0.0147176) rest(mean) delta //5th percentile
prvalue, x(AS=0.2335184) rest(mean) delta //mean
prvalue, x(AS=0.5752195) rest(mean) delta //95th percentile

//Marginal Effects
margins, dydx(AS) at(AS=(0(0.01)0.8717034)) predict(outcome(1)) saving(ASME1, replace)
margins, dydx(AS) at(AS=(0(0.01)0.8717034)) predict(outcome(2)) saving(ASME2, replace)
margins, dydx(AS) at(AS=(0(0.01)0.8717034)) predict(outcome(3)) saving(ASME3, replace)
combomarginsplot ASME1 ASME2 ASME3, xlabel(0 "Min" 0.2335184 "Mean" 0.8717034 "Max") xtitle("Asset specificity") ytitle("Marginal Effects") title("") recastci(rline) ciopts(lpattern(shortdash_dot))
//AME & MEM
margins, dydx(AS)
margins, dydx(AS) atmeans

However, now I want to test if there is a moderation effect of AS_ExtUnc which is AS*ExtUnc. Of what I have read, you cannot use the estimates. I have tried to use the inteff command, but it seems to not work with mlogit. Can I do the same procedure as with AS or is that wrongful in dealing with an interaction term?

Best, Frederik