Someone pointed me to data showing a recent very steep decline in incoming migrants. I had a look at the data and plotted it. Unfortunately, the dataset has some problematic missing data that make the decline appear more steep than it is. We’re at 2014 levels, not 2011’s.
Main limitation of the below is that I used univariate imputation, i.e. did not exploit correlations in time trends between countries to impute missing values. It probably did not matter much because all the methods agreed about the crucial last few values from 2018. See this neat paper for how to do time series imputation in R.