Abstract: PMI composite output index, one of the reliable early indicators of national economic growth, is composed of manufacturing PMI output and services PMI business activity in weighted average method. This paper establishes dynamic models for the first time of manufacturing PMI output, services PMI business activity and PMI composite output using ARDL, based on lag effects of demand-supply. It shows that firstly, new orders, purchases, gap of new orders and inventory of finished goods as well as the profit trend (new orders-input prices) have significant hysteresis effects on PMI output and business activity, especially in new orders; secondly, new orders, gap and services profit trend have significant hysteresis effects on PMI composite output, especially in service new orders and gap; and thirdly, manufacturing new orders have negative hysteresis effect on composite output during the short period. This paper concludes at the sectoral level, expanding domestic demand respectively is still the most essential way to boost production. However, destocking and services demand determine to a large extent the growth pattern for sector-wide production growth.

Key words: composite PMI output, lagged effect, ARDL model