Sometimes this is the kind of output that ETS and ARIMA produce. When the historical data doesn't have strong seasonality the forecasting model may find it difficult to predict the future. Therefore it simply takes average of your previous values.
