This　study　proposes　a　combination　model　to　assess　upper　variance　spillover　effects　in　14　stock　markets.　By　integrating　G-normal　distribution　model　and　the　connectedness　approach,　we　measure　spillover　effects　and　compute　upper　variance.　Empirical　findings　reveal　that　developed　countries　(e.g.,　Britain,　France,　Germany)　contribute　more　to　upper　variance　risk,　while　developing　countries　(e.g.,　Philippines,　Brazil,　Indonesia)　receive　it.　During　the　crisis,　the　total　spillover　index　increases　from　24.81%　to　66.01%.　French　and　German　stock　markets＇　spillover　rises　by　144.21%　and　44.94%　respectively,　while　China＇s　spillover　is　relatively　smaller.　Dynamic　upper　variance　exhibits　an　upward　trend,　sensitive　to　major　economic　shocks.