This　paper　focuses　on　the　recursive　parameter　estimation　problem　of　the　exponential　autoregressive　(ExpAR)　model.　Applying　the　Newton　search　and　multi-innovation　theory,　a　multi-innovation　Newton　recursive　algorithm　is　presented　for　identifying　the　ExpAR　model.　In　order　to　improve　the　computational　efficiency,　the　hierarchical　identification　principle　is　employed　to　decompose　an　ExpAR　model　into　two　sub-models,　and　to　derive　a　hierarchical　multi-innovation　Newton　recursive　algorithm.　A　simulation　example　is　provided　to　demonstrate　the　effectiveness　of　the　proposed　algorithms.